Changeset 9cdfb4d0

doc/proposals/concurrency/.gitignore

r326cd2b	r9cdfb4d0
24	24	build/*.toc
25	25	*.pdf
	26	*.png
	27	figures/*.tex
26	28
27	29	examples

doc/proposals/concurrency/Makefile

-                      r326cd2b
+                      r9cdfb4d0
 ## Define the appropriate configuration variables.
 TeXLIB = .:./style:./text:./annex:./build:../../LaTeXmacros:../../LaTeXmacros/listings:../../LaTeXmacros/enumitem:~/bibliographies:
+TeXLIB = .:./style:./text:./annex:./build:../../LaTeXmacros:../../LaTeXmacros/listings:../../LaTeXmacros/enumitem:~/bibliographies:/usr/local/bibliographies:
 LaTeX  = TEXINPUTS=${TeXLIB} && export TEXINPUTS && latex -halt-on-error -output-directory=build -interaction=nonstopmode
 BibTeX = BIBINPUTS=${TeXLIB} && export BIBINPUTS && bibtex -terse
 …
         build/*.tex     \
         build/*.toc     \
+        build/*.lof     \
+        build/*.lol     \
+        build/*.lot     \
+        figures/*.tex   \
+        *.png           \
 …
         fig2dev -L pstex_t -p $@ $< > $@_t
+figures/%.tex: build/%.pstex
+        echo -n         "\documentclass[preview]{standalone}\n"         \
+                        "\usepackage[T1]{fontenc}\n"                    \
+                        "\usepackage[usenames]{color}\n"                \
+                        "\usepackage{graphicx}\n"                       \
+                        "\usepackage{listings}\n"                       \
+                        "\usepackage{xspace}\n"                         \
+                        "\input{style}\n"                               \
+                        "\\\\begin{document}\n"                         \
+                        "{\\\\resizebox{3\\\\textwidth}{!}{\input{${basename ${notdir $@}}.pstex_t}}}\n" \
+                        "\end{document}" > $@
+%.png : build/%.pstex figures/%.tex
+        echo ${basename $@}
+        ${LaTeX} figures/${basename $@}.tex
+        dvips build/${basename $@}.dvi -o build/${basename $@}.ps
+        ps2pdf build/${basename $@}.ps
+        convert -negate ${basename $@}.pdf $@
 # Local Variables: #

doc/proposals/concurrency/annex/glossary.tex

r326cd2b	r9cdfb4d0
98	98	\newacronym{tls}{TLS}{Thread Local Storage}
99	99	\newacronym{api}{API}{Application Program Interface}
100		\newacronym{raii}{RAII}{Ressource Acquisition Is Initialization}
	100	\newacronym{raii}{RAII}{Resource Acquisition Is Initialization}
101	101	\newacronym{numa}{NUMA}{Non-Uniform Memory Access}

doc/proposals/concurrency/annex/local.bib

-                      r326cd2b
+                      r9cdfb4d0
 @article{TBB,
+        key     = {TBB},
         keywords        = {Intel, TBB},
         title   = {Intel Thread Building Blocks},
 …
 @manual{www-cfa,
+        key     = {CFA},
         keywords        = {Cforall},
+        title   = {Cforall Programmming Language},
+        address = {https://plg.uwaterloo.ca/~cforall/}
+        author  = {C$\forall$},
+        title   = {C$\forall$ Programmming Language},
+        note    = {\url{https://plg.uwaterloo.ca/~cforall}},
+}
 @article{rob-thesis,
+@mastersthesis{rob-thesis,
         keywords        = {Constructors, Destructors, Tuples},
         author  = {Rob Schluntz},
         title   = {Resource Management and Tuples in Cforall},
+        year            = 2017
+        year            = 2017,
+        school  = {University of Waterloo},
+        note    = {\url{https://uwspace.uwaterloo.ca/handle/10012/11830}},
+}
 …
 @article{BankTransfer,
+        key     = {Bank Transfer},
         keywords        = {Bank Transfer},
         title   = {Bank Account Transfer Problem},
 …
+}
+@book{Herlihy93,
+        title={Transactional memory: Architectural support for lock-free data structures},
+        author={Herlihy, Maurice and Moss, J Eliot B},
+        volume={21},
+        number={2},
+        year={1993},
+        publisher={ACM}
+@article{Herlihy93,
+        author  = {Herlihy, Maurice and Moss, J. Eliot B.},
+        title   = {Transactional memory: architectural support for lock-free data structures},
+        journal = {SIGARCH Comput. Archit. News},
+        issue_date      = {May 1993},
+        volume  = {21},
+        number  = {2},
+        month   = may,
+        year    = {1993},
+        pages   = {289--300},
+        numpages        = {12},
+        publisher       = {ACM},
+        address = {New York, NY, USA},
+}
 @manual{affinityLinux,
+        key     = {TBB},
         title           = "{Linux man page - sched\_setaffinity(2)}"
+}
 …
+}
 @misc{NodeJs,
         title           = "{Node.js}",

doc/proposals/concurrency/figures/system.fig

-                      r326cd2b
+                      r9cdfb4d0
 0 0 50 -1 0 11 0.0000 2 120 570 4125 7126 thread 4\001
 -6
+6975 4050 9525 7875
+2 0 1 0 7 50 -1 -1 0.000 0 1 -1 0 0 5
+5400 7575 5400 7575 5850 7125 5850 7125 5400
+2 0 1 0 7 50 -1 18 0.000 0 1 -1 0 0 5
+4200 7575 4200 7575 4650 7125 4650 7125 4200
+2 0 1 0 7 50 -1 45 0.000 0 1 -1 0 0 5
+4800 7575 4800 7575 5250 7125 5250 7125 4800
+2 0 1 0 7 50 -1 -1 0.000 0 1 -1 0 0 5
+4050 9525 4050 9525 7875 6975 7875 6975 4050
+2 0 2 0 7 49 -1 -1 0.000 1 0 0 10
+6900 7500 6975 7200 7050 7500 7125 7200 7200 7500 7275
+7350 7500 7425 7200 7500 7350 7575
+.000 -1.000 -1.000 -1.000 -1.000 -1.000 -1.000 -1.000
+         -1.000 0.000
+2 0 4 0 7 49 -1 -1 0.000 1 0 0 10
+6000 7500 6075 7200 6150 7500 6225 7200 6300 7500 6375
+6450 7500 6525 7200 6600 7350 6675
+.000 -1.000 -1.000 -1.000 -1.000 -1.000 -1.000 -1.000
+         -1.000 0.000
+0 0 50 -1 0 11 0.0000 2 120 945 7725 4500 Pthread stack\001
+0 0 50 -1 0 11 0.0000 2 150 1530 7725 5100 Pthread stack (stolen)\001
+0 0 50 -1 0 11 0.0000 2 120 540 7725 6375 Pthread\001
+0 0 50 -1 0 11 0.0000 2 150 1065 7725 7275 $\\CFA$ thread\001
+0 0 50 -1 0 11 0.0000 2 150 990 7725 5700 $\\CFA$ stack\001
+-6
 2 0 1 0 7 50 -1 -1 0.000 1 3.1416 3150 5250 750 450 2400 4800 3900 5700
 1 0 1 0 7 50 -1 -1 0.000 0 1 -1 0 0 2
 …
 1 0 1 0 7 50 -1 -1 0.000 0 1 -1 0 0 2
 3900 3825 5025
-2 0 1 0 7 50 -1 -1 0.000 0 1 -1 0 0 5
-3525 7575 3525 7575 3975 7125 3975 7125 3525
-2 0 1 0 7 50 -1 -1 0.000 0 1 -1 0 0 5
-2175 9525 2175 9525 6000 6975 6000 6975 2175
 1 0 1 0 7 50 -1 -1 0.000 0 1 -1 0 0 2
 6225 2400 5400
 …
 2 0 1 0 7 50 -1 45 0.000 0 1 -1 0 0 5
 975 5850 975 5850 2625 5100 2625 5100 975
-2 0 1 0 7 50 -1 18 0.000 0 1 -1 0 0 5
-2325 7575 2325 7575 2775 7125 2775 7125 2325
-2 0 1 0 7 50 -1 45 0.000 0 1 -1 0 0 5
-2925 7575 2925 7575 3375 7125 3375 7125 2925
 2 0 1 0 7 50 -1 45 0.000 0 1 -1 0 0 5
 7425 1275 7425 1275 9075 525 9075 525 7425
 …
 2 0 1 0 7 50 -1 18 0.000 0 1 -1 0 0 5
 2625 2775 2625 2775 975 2025 975 2025 2625
-2 0 2 0 7 49 -1 -1 0.000 1 0 0 10
-5025 7500 5100 7200 5175 7500 5250 7200 5325 7500 5400
-5475 7500 5550 7200 5625 7350 5700
-.000 -1.000 -1.000 -1.000 -1.000 -1.000 -1.000 -1.000
-         -1.000 0.000
-2 0 4 0 7 49 -1 -1 0.000 1 0 0 10
-4125 7500 4200 7200 4275 7500 4350 7200 4425 7500 4500
-4575 7500 4650 7200 4725 7350 4800
-.000 -1.000 -1.000 -1.000 -1.000 -1.000 -1.000 -1.000
-         -1.000 0.000
 0 0 50 -1 0 18 0.0000 2 30 225 4500 3150 ...\001
 0 0 50 -1 0 18 0.0000 2 30 225 3750 4500 ...\001
 0 0 50 -1 0 11 0.0000 2 120 705 2775 5325 Scheduler\001
-0 0 50 -1 0 11 0.0000 2 120 945 7725 2625 Pthread stack\001
-0 0 50 -1 0 11 0.0000 2 150 1530 7725 3225 Pthread stack (stolen)\001
-0 0 50 -1 0 11 0.0000 2 120 540 7725 4500 Pthread\001
-0 0 50 -1 0 11 0.0000 2 150 1065 7725 5400 $\\CFA$ thread\001
 0 0 50 -1 0 18 0.0000 2 30 225 4950 6600 ...\001
 0 0 50 -1 0 18 0.0000 2 30 225 4200 5850 ...\001
-0 0 50 -1 0 11 0.0000 2 150 990 7725 3825 $\\CFA$ stack\001

doc/proposals/concurrency/text/basics.tex

-                      r326cd2b
+                      r9cdfb4d0
 At its core, concurrency is based on having multiple call-stacks and scheduling among threads of execution executing on these stacks. Concurrency without parallelism only requires having multiple call stacks (or contexts) for a single thread of execution.
 Execution with a single thread and multiple stacks where the thread is self-scheduling deterministically across the stacks is called coroutining. Execution with a single and multiple stacks but where the thread is scheduled by an oracle (non-deterministic from the thread perspective) across the stacks is called concurrency.
 Therefore, a minimal concurrency system can be achieved by creating coroutines, which instead of context-switching among each other, always ask an oracle where to context-switch next. While coroutines can execute on the caller?s stack-frame, stack-full coroutines allow full generality and are sufficient as the basis for concurrency. The aforementioned oracle is a scheduler and the whole system now follows a cooperative threading-model (aka non-preemptive scheduling). The oracle/scheduler can either be a stack-less or stack-full entity and correspondingly require one or two context-switches to run a different coroutine. In any case, a subset of concurrency related challenges start to appear. For the complete set of concurrency challenges to occur, the only feature missing is preemption.
+Execution with a single thread and multiple stacks where the thread is self-scheduling deterministically across the stacks is called coroutining. Execution with a single and multiple stacks but where the thread is scheduled by an oracle (non-deterministic from the thread's perspective) across the stacks is called concurrency.
+Therefore, a minimal concurrency system can be achieved by creating coroutines (see Section \ref{coroutine}), which instead of context-switching among each other, always ask an oracle where to context-switch next. While coroutines can execute on the caller's stack-frame, stack-full coroutines allow full generality and are sufficient as the basis for concurrency. The aforementioned oracle is a scheduler and the whole system now follows a cooperative threading-model (a.k.a., non-preemptive scheduling). The oracle/scheduler can either be a stack-less or stack-full entity and correspondingly require one or two context-switches to run a different coroutine. In any case, a subset of concurrency related challenges start to appear. For the complete set of concurrency challenges to occur, the only feature missing is preemption.
 A scheduler introduces order of execution uncertainty, while preemption introduces uncertainty about where context switches occur. Mutual exclusion and synchronization are ways of limiting non-determinism in a concurrent system. Now it is important to understand that uncertainty is desirable; uncertainty can be used by runtime systems to significantly increase performance and is often the basis of giving a user the illusion that tasks are running in parallel. Optimal performance in concurrent applications is often obtained by having as much non-determinism as correctness allows.
 \section{\protect\CFA's Thread Building Blocks}
 One of the important features that are missing in C is threading. On modern architectures, a lack of threading is unacceptable~\cite{Sutter05, Sutter05b}, and therefore modern programming languages must have the proper tools to allow users to write efficient concurrent programs to take advantage of parallelism. As an extension of C, \CFA needs to express these concepts in a way that is as natural as possible to programmers familiar with imperative languages. And being a system-level language means programmers expect to choose precisely which features they need and which cost they are willing to pay.
+One of the important features that are missing in C is threading\footnote{While the C11 standard defines a ``threads.h'' header, it is minimal and defined as optional. As such, library support for threading is far from widespread. At the time of writing the thesis, neither \texttt{gcc} nor \texttt{clang} support ``threads.h'' in their respective standard libraries.}. On modern architectures, a lack of threading is unacceptable~\cite{Sutter05, Sutter05b}, and therefore modern programming languages must have the proper tools to allow users to write efficient concurrent programs to take advantage of parallelism. As an extension of C, \CFA needs to express these concepts in a way that is as natural as possible to programmers familiar with imperative languages. And being a system-level language means programmers expect to choose precisely which features they need and which cost they are willing to pay.
 \section{Coroutines: A Stepping Stone}\label{coroutine}
 While the main focus of this proposal is concurrency and parallelism, it is important to address coroutines, which are actually a significant building block of a concurrency system. Coroutines need to deal with context switches and other context-management operations. Therefore, this proposal includes coroutines both as an intermediate step for the implementation of threads, and a first-class feature of \CFA. Furthermore, many design challenges of threads are at least partially present in designing coroutines, which makes the design effort that much more relevant. The core \acrshort{api} of coroutines revolves around two features: independent call-stacks and \code{suspend}/\code{resume}.
+While the main focus of this proposal is concurrency and parallelism, it is important to address coroutines, which are actually a significant building block of a concurrency system. \textbf{Coroutine}s are generalized routines which have predefined points where execution is suspended and can be resumed at a later time. Therefore, they need to deal with context switches and other context-management operations. This proposal includes coroutines both as an intermediate step for the implementation of threads, and a first-class feature of \CFA. Furthermore, many design challenges of threads are at least partially present in designing coroutines, which makes the design effort that much more relevant. The core \acrshort{api} of coroutines revolves around two features: independent call-stacks and \code{suspend}/\code{resume}.
 \begin{table}
 …
 \end{tabular}
 \end{center}
 \caption{Different implementations of a Fibonacci sequence generator in C.},
+\caption{Different implementations of a Fibonacci sequence generator in C.}
 \label{lst:fibonacci-c}
 \end{table}
 A good example of a problem made easier with coroutines is generators, like the Fibonacci sequence. This problem comes with the challenge of decoupling how a sequence is generated and how it is used. Table \ref{lst:fibonacci-c} shows conventional approaches to writing generators in C. All three of these approach suffer from strong coupling. The left and centre approaches require that the generator have knowledge of how the sequence is used, while the rightmost approach requires holding internal state between calls on behalf of the generator and makes it much harder to handle corner cases like the Fibonacci seed.
+A good example of a problem made easier with coroutines is generators, e.g., generating the Fibonacci sequence. This problem comes with the challenge of decoupling how a sequence is generated and how it is used. Listing \ref{lst:fibonacci-c} shows conventional approaches to writing generators in C. All three of these approach suffer from strong coupling. The left and centre approaches require that the generator have knowledge of how the sequence is used, while the rightmost approach requires holding internal state between calls on behalf of the generator and makes it much harder to handle corner cases like the Fibonacci seed.
 Listing \ref{lst:fibonacci-cfa} is an example of a solution to the Fibonacci problem using \CFA coroutines, where the coroutine stack holds sufficient state for the next generation. This solution has the advantage of having very strong decoupling between how the sequence is generated and how it is used. Indeed, this version is as easy to use as the \code{fibonacci_state} solution, while the implementation is very similar to the \code{fibonacci_func} example.
 …
 One important design challenge for implementing coroutines and threads (shown in section \ref{threads}) is that the runtime system needs to run code after the user-constructor runs to connect the fully constructed object into the system. In the case of coroutines, this challenge is simpler since there is no non-determinism from preemption or scheduling. However, the underlying challenge remains the same for coroutines and threads.
 The runtime system needs to create the coroutine?s stack and more importantly prepare it for the first resumption. The timing of the creation is non-trivial since users both expect to have fully constructed objects once execution enters the coroutine main and to be able to resume the coroutine from the constructor. There are several solutions to this problem but the chosen option effectively forces the design of the coroutine.
+The runtime system needs to create the coroutine's stack and, more importantly, prepare it for the first resumption. The timing of the creation is non-trivial since users expect both to have fully constructed objects once execution enters the coroutine main and to be able to resume the coroutine from the constructor. There are several solutions to this problem but the chosen option effectively forces the design of the coroutine.
 Furthermore, \CFA faces an extra challenge as polymorphic routines create invisible thunks when cast to non-polymorphic routines and these thunks have function scope. For example, the following code, while looking benign, can run into undefined behaviour because of thunks:
 …
+}
 \end{ccode}
 The problem in this example is a storage management issue, the function pointer \code{_thunk0} is only valid until the end of the block, which limits the viable solutions because storing the function pointer for too long causes Undefined Behaviour; i.e., the stack-based thunk being destroyed before it can be used. This challenge is an extension of challenges that come with second-class routines. Indeed, GCC nested routines also have the limitation that nested routine cannot be passed outside of the declaration scope. The case of coroutines and threads is simply an extension of this problem to multiple call stacks.
+The problem in this example is a storage management issue, the function pointer \code{_thunk0} is only valid until the end of the block, which limits the viable solutions because storing the function pointer for too long causes undefined behaviour; i.e., the stack-based thunk being destroyed before it can be used. This challenge is an extension of challenges that come with second-class routines. Indeed, GCC nested routines also have the limitation that nested routine cannot be passed outside of the declaration scope. The case of coroutines and threads is simply an extension of this problem to multiple call stacks.
 \subsection{Alternative: Composition}
 …
 symmetric_coroutine<>::yield_type
 \end{cfacode}
 Often, the canonical threading paradigm in languages is based on function pointers, pthread being one of the most well-known examples. The main problem of this approach is that the thread usage is limited to a generic handle that must otherwise be wrapped in a custom type. Since the custom type is simple to write in \CFA and solves several issues, added support for routine/lambda based coroutines adds very little.
 A variation of this would be to use a simple function pointer in the same way pthread does for threads :
+Often, the canonical threading paradigm in languages is based on function pointers, \texttt{pthread} being one of the most well-known examples. The main problem of this approach is that the thread usage is limited to a generic handle that must otherwise be wrapped in a custom type. Since the custom type is simple to write in \CFA and solves several issues, added support for routine/lambda based coroutines adds very little.
+A variation of this would be to use a simple function pointer in the same way \texttt{pthread} does for threads:
 \begin{cfacode}
 void foo( coroutine_t cid, void* arg ) {
 …
 \subsection{Alternative: Trait-Based Coroutines}
 Finally, the underlying approach, which is the one closest to \CFA idioms, is to use trait-based lazy coroutines. This approach defines a coroutine as anything that satisfies the trait \code{is_coroutine} and is used as a coroutine.
+Finally, the underlying approach, which is the one closest to \CFA idioms, is to use trait-based lazy coroutines. This approach defines a coroutine as anything that satisfies the trait \code{is_coroutine} (as defined below) and is used as a coroutine.
 \begin{cfacode}
 …
 forall( dtype T | is_coroutine(T) ) void resume (T&);
 \end{cfacode}
 This ensures an object is not a coroutine until \code{resume} is called on the object. Correspondingly, any object that is passed to \code{resume} is a coroutine since it must satisfy the \code{is_coroutine} trait to compile. The advantage of this approach is that users can easily create different types of coroutines, for example, changing the memory layout of a coroutine is trivial when implementing the \code{get_coroutine} routine. The \CFA keyword \code{coroutine} only has the effect of implementing the getter and forward declarations required for users to implement the main routine.
+This ensures that an object is not a coroutine until \code{resume} is called on the object. Correspondingly, any object that is passed to \code{resume} is a coroutine since it must satisfy the \code{is_coroutine} trait to compile. The advantage of this approach is that users can easily create different types of coroutines, for example, changing the memory layout of a coroutine is trivial when implementing the \code{get_coroutine} routine. The \CFA keyword \code{coroutine} simply has the effect of implementing the getter and forward declarations required for users to implement the main routine.
 \begin{center}
 …
 \end{cfacode}
 Obviously, for this thread implementation to be useful it must run some user code. Several other threading interfaces use a function-pointer representation as the interface of threads (for example \Csharp~\cite{Csharp} and Scala~\cite{Scala}). However, this proposal considers that statically tying a \code{main} routine to a thread supersedes this approach. Since the \code{main} routine is already a special routine in \CFA (where the program begins), it is a natural extension of the semantics using overloading to declare mains for different threads (the normal main being the main of the initial thread). As such the \code{main} routine of a thread can be defined as
+Obviously, for this thread implementation to be useful it must run some user code. Several other threading interfaces use a function-pointer representation as the interface of threads (for example \Csharp~\cite{Csharp} and Scala~\cite{Scala}). However, this proposal considers that statically tying a \code{main} routine to a thread supersedes this approach. Since the \code{main} routine is already a special routine in \CFA (where the program begins), it is a natural extension of the semantics to use overloading to declare mains for different threads (the normal main being the main of the initial thread). As such the \code{main} routine of a thread can be defined as
 \begin{cfacode}
 thread foo {};
 …
 A consequence of the strongly typed approach to main is that memory layout of parameters and return values to/from a thread are now explicitly specified in the \acrshort{api}.
 Of course for threads to be useful, it must be possible to start and stop threads and wait for them to complete execution. While using an \acrshort{api} such as \code{fork} and \code{join} is relatively common in the literature, such an interface is unnecessary. Indeed, the simplest approach is to use \acrshort{raii} principles and have threads \code{fork} after the constructor has completed and \code{join} before the destructor runs.
+Of course, for threads to be useful, it must be possible to start and stop threads and wait for them to complete execution. While using an \acrshort{api} such as \code{fork} and \code{join} is relatively common in the literature, such an interface is unnecessary. Indeed, the simplest approach is to use \acrshort{raii} principles and have threads \code{fork} after the constructor has completed and \code{join} before the destructor runs.
 \begin{cfacode}
 thread World;
 …
 \end{cfacode}
 However, one of the drawbacks of this approach is that threads always form a lattice, i.e., they are always destroyed in the opposite order of construction because of block structure. This restriction is relaxed by using dynamic allocation, so threads can outlive the scope in which they are created, much like dynamically allocating memory lets objects outlive the scope in which they are created.
+However, one of the drawbacks of this approach is that threads always form a tree where nodes must always outlive their children, i.e., they are always destroyed in the opposite order of construction because of C scoping rules. This restriction is relaxed by using dynamic allocation, so threads can outlive the scope in which they are created, much like dynamically allocating memory lets objects outlive the scope in which they are created.
 \begin{cfacode}

doc/proposals/concurrency/text/cforall.tex

-                      r326cd2b
+                      r9cdfb4d0
 % ======================================================================
 \section{Operators}
 Overloading also extends to operators. The syntax for denoting operator-overloading is to name a routine with the symbol of the operator and question marks where the arguments of the operation occur, e.g.:
+Overloading also extends to operators. The syntax for denoting operator-overloading is to name a routine with the symbol of the operator and question marks where the arguments of the operation appear, e.g.:
 \begin{cfacode}
 int ++? (int op);                       //unary prefix increment
 …
 % ======================================================================
 \section{Constructors/Destructors}
 Object lifetime is often a challenge in concurrency. \CFA uses the approach of giving concurrent meaning to object lifetime as a means of synchronization and/or mutual exclusion. Since \CFA relies heavily on the lifetime of objects, constructors and destructors is a core feature required for concurrency and parallelism. \CFA uses the following syntax for constructors and destructors :
+Object lifetime is often a challenge in concurrency. \CFA uses the approach of giving concurrent meaning to object lifetime as a means of synchronization and/or mutual exclusion. Since \CFA relies heavily on the lifetime of objects, constructors and destructors is a core feature required for concurrency and parallelism. \CFA uses the following syntax for constructors and destructors:
 \begin{cfacode}
 struct S {
 …
 % ======================================================================
 \section{Parametric Polymorphism}
+Routines in \CFA can also be reused for multiple types. This capability is done using the \code{forall} clause, which gives \CFA its name. \code{forall} clauses allow separately compiled routines to support generic usage over multiple types. For example, the following sum function works for any type that supports construction from 0 and addition :
+\label{s:ParametricPolymorphism}
+Routines in \CFA can also be reused for multiple types. This capability is done using the \code{forall} clauses, which allow separately compiled routines to support generic usage over multiple types. For example, the following sum function works for any type that supports construction from 0 and addition:
 \begin{cfacode}
 //constraint type, 0 and +
 …
 Since writing constraints on types can become cumbersome for more constrained functions, \CFA also has the concept of traits. Traits are named collection of constraints that can be used both instead and in addition to regular constraints:
 \begin{cfacode}
 trait sumable( otype T ) {
+trait summable( otype T ) {
         void ?{}(T *, zero_t);          //constructor from 0 literal
         T ?+?(T, T);                            //assortment of additions
 …
         T ?++(T *);
 };
 forall( otype T | sumable(T) )  //use trait
+forall( otype T | summable(T) ) //use trait
 T sum(T a[], size_t size);
 \end{cfacode}
 …
 % ======================================================================
 \section{with Clause/Statement}
 Since \CFA lacks the concept of a receiver, certain functions end-up needing to repeat variable names often. To remove this inconvenience, \CFA provides the \code{with} statement, which opens an aggregate scope making its fields directly accessible (like Pascal).
+Since \CFA lacks the concept of a receiver, certain functions end up needing to repeat variable names often. To remove this inconvenience, \CFA provides the \code{with} statement, which opens an aggregate scope making its fields directly accessible (like Pascal).
 \begin{cfacode}
 struct S { int i, j; };

doc/proposals/concurrency/text/concurrency.tex

-                      r326cd2b
+                      r9cdfb4d0
 % ======================================================================
 % ======================================================================
 Several tools can be used to solve concurrency challenges. Since many of these challenges appear with the use of mutable shared-state, some languages and libraries simply disallow mutable shared-state (Erlang~\cite{Erlang}, Haskell~\cite{Haskell}, Akka (Scala)~\cite{Akka}). In these paradigms, interaction among concurrent objects relies on message passing~\cite{Thoth,Harmony,V-Kernel} or other paradigms closely relate to networking concepts (channels~\cite{CSP,Go} for example). However, in languages that use routine calls as their core abstraction mechanism, these approaches force a clear distinction between concurrent and non-concurrent paradigms (i.e., message passing versus routine calls). This distinction in turn means that, in order to be effective, programmers need to learn two sets of design patterns. While this distinction can be hidden away in library code, effective use of the library still has to take both paradigms into account.
+Several tools can be used to solve concurrency challenges. Since many of these challenges appear with the use of mutable shared state, some languages and libraries simply disallow mutable shared state (Erlang~\cite{Erlang}, Haskell~\cite{Haskell}, Akka (Scala)~\cite{Akka}). In these paradigms, interaction among concurrent objects relies on message passing~\cite{Thoth,Harmony,V-Kernel} or other paradigms closely relate to networking concepts (channels~\cite{CSP,Go} for example). However, in languages that use routine calls as their core abstraction mechanism, these approaches force a clear distinction between concurrent and non-concurrent paradigms (i.e., message passing versus routine calls). This distinction in turn means that, in order to be effective, programmers need to learn two sets of design patterns. While this distinction can be hidden away in library code, effective use of the library still has to take both paradigms into account.
 Approaches based on shared memory are more closely related to non-concurrent paradigms since they often rely on basic constructs like routine calls and shared objects. At the lowest level, concurrent paradigms are implemented as atomic operations and locks. Many such mechanisms have been proposed, including semaphores~\cite{Dijkstra68b} and path expressions~\cite{Campbell74}. However, for productivity reasons it is desirable to have a higher-level construct be the core concurrency paradigm~\cite{HPP:Study}.
 …
 \subsection{Mutual-Exclusion}
 As mentioned above, mutual-exclusion is the guarantee that only a fix number of threads can enter a critical section at once. However, many solutions exist for mutual exclusion, which vary in terms of performance, flexibility and ease of use. Methods range from low-level locks, which are fast and flexible but require significant attention to be correct, to  higher-level mutual-exclusion methods, which sacrifice some performance in order to improve ease of use. Ease of use comes by either guaranteeing some problems cannot occur (e.g., being deadlock free) or by offering a more explicit coupling between data and corresponding critical section. For example, the \CC \code{std::atomic<T>} offers an easy way to express mutual-exclusion on a restricted set of operations (e.g.: reading/writing large types atomically). Another challenge with low-level locks is composability. Locks have restricted composability because it takes careful organizing for multiple locks to be used while preventing deadlocks. Easing composability is another feature higher-level mutual-exclusion mechanisms often offer.
+As mentioned above, mutual-exclusion is the guarantee that only a fix number of threads can enter a critical section at once. However, many solutions exist for mutual exclusion, which vary in terms of performance, flexibility and ease of use. Methods range from low-level locks, which are fast and flexible but require significant attention to be correct, to  higher-level concurrency techniques, which sacrifice some performance in order to improve ease of use. Ease of use comes by either guaranteeing some problems cannot occur (e.g., being deadlock free) or by offering a more explicit coupling between data and corresponding critical section. For example, the \CC \code{std::atomic<T>} offers an easy way to express mutual-exclusion on a restricted set of operations (e.g., reading/writing large types atomically). Another challenge with low-level locks is composability. Locks have restricted composability because it takes careful organizing for multiple locks to be used while preventing deadlocks. Easing composability is another feature higher-level mutual-exclusion mechanisms often offer.
 \subsection{Synchronization}
 As for mutual-exclusion, low-level synchronization primitives often offer good performance and good flexibility at the cost of ease of use. Again, higher-level mechanisms often simplify usage by adding better coupling between synchronization and data, e.g.: message passing, or offering a simpler solution to otherwise involved challenges. As mentioned above, synchronization can be expressed as guaranteeing that event \textit{X} always happens before \textit{Y}. Most of the time, synchronization happens within a critical section, where threads must acquire mutual-exclusion in a certain order. However, it may also be desirable to guarantee that event \textit{Z} does not occur between \textit{X} and \textit{Y}. Not satisfying this property is called barging. For example, where event \textit{X} tries to effect event \textit{Y} but another thread acquires the critical section and emits \textit{Z} before \textit{Y}. The classic example is the thread that finishes using a resource and unblocks a thread waiting to use the resource, but the unblocked thread must compete again to acquire the resource. Preventing or detecting barging is an involved challenge with low-level locks, which can be made much easier by higher-level constructs. This challenge is often split into two different methods, barging avoidance and barging prevention. Algorithms that use flag variables to detect barging threads are said to be using barging avoidance, while algorithms that baton-pass locks~\cite{Andrews89} between threads instead of releasing the locks are said to be using barging prevention.
+As with mutual-exclusion, low-level synchronization primitives often offer good performance and good flexibility at the cost of ease of use. Again, higher-level mechanisms often simplify usage by adding either better coupling between synchronization and data (e.g., message passing) or offering a simpler solution to otherwise involved challenges. As mentioned above, synchronization can be expressed as guaranteeing that event \textit{X} always happens before \textit{Y}. Most of the time, synchronization happens within a critical section, where threads must acquire mutual-exclusion in a certain order. However, it may also be desirable to guarantee that event \textit{Z} does not occur between \textit{X} and \textit{Y}. Not satisfying this property is called \textbf{barging}. For example, where event \textit{X} tries to effect event \textit{Y} but another thread acquires the critical section and emits \textit{Z} before \textit{Y}. The classic example is the thread that finishes using a resource and unblocks a thread waiting to use the resource, but the unblocked thread must compete to acquire the resource. Preventing or detecting barging is an involved challenge with low-level locks, which can be made much easier by higher-level constructs. This challenge is often split into two different methods, barging avoidance and barging prevention. Algorithms that use flag variables to detect barging threads are said to be using barging avoidance, while algorithms that baton-pass locks~\cite{Andrews89} between threads instead of releasing the locks are said to be using barging prevention.
 % ======================================================================
 …
 % ======================================================================
 % ======================================================================
 A monitor is a set of routines that ensure mutual exclusion when accessing shared state. This concept is generally associated with Object-Oriented Languages like Java~\cite{Java} or \uC~\cite{uC++book} but does not strictly require OO semantics. The only requirement is the ability to declare a handle to a shared object and a set of routines that act on it :
+A \textbf{monitor} is a set of routines that ensure mutual-exclusion when accessing shared state. More precisely, a monitor is a programming technique that associates mutual-exclusion to routine scopes, as opposed to mutex locks, where mutual-exclusion is defined by lock/release calls independently of any scoping of the calling routine. This strong association eases readability and maintainability, at the cost of flexibility. Note that both monitors and mutex locks, require an abstract handle to identify them. This concept is generally associated with object-oriented languages like Java~\cite{Java} or \uC~\cite{uC++book} but does not strictly require OO semantics. The only requirement is the ability to declare a handle to a shared object and a set of routines that act on it:
 \begin{cfacode}
 typedef /*some monitor type*/ monitor;
 int f(monitor& m);
+int f(monitor & m);
 int main() {
 …
 % ======================================================================
 % ======================================================================
 The above monitor example displays some of the intrinsic characteristics. First, it is necessary to use pass-by-reference over pass-by-value for monitor routines. This semantics is important, because at their core, monitors are implicit mutual-exclusion objects (locks), and these objects cannot be copied. Therefore, monitors are implicitly non-copy-able objects (\code{dtype}).
 Another aspect to consider is when a monitor acquires its mutual exclusion. For example, a monitor may need to be passed through multiple helper routines that do not acquire the monitor mutual-exclusion on entry. Pass through can occur for generic helper routines (\code{swap}, \code{sort}, etc.) or specific helper routines like the following to implement an atomic counter :
+The above monitor example displays some of the intrinsic characteristics. First, it is necessary to use pass-by-reference over pass-by-value for monitor routines. This semantics is important, because at their core, monitors are implicit mutual-exclusion objects (locks), and these objects cannot be copied. Therefore, monitors are non-copy-able objects (\code{dtype}).
+Another aspect to consider is when a monitor acquires its mutual exclusion. For example, a monitor may need to be passed through multiple helper routines that do not acquire the monitor mutual-exclusion on entry. Passthrough can occur for generic helper routines (\code{swap}, \code{sort}, etc.) or specific helper routines like the following to implement an atomic counter:
 \begin{cfacode}
 …
 \end{tabular}
 \end{center}
 Notice how the counter is used without any explicit synchronization and yet supports thread-safe semantics for both reading and writing, which is similar in usage to \CC \code{atomic} template.
+Notice how the counter is used without any explicit synchronization and yet supports thread-safe semantics for both reading and writing, which is similar in usage to the \CC template \code{std::atomic}.
 Here, the constructor (\code{?\{\}}) uses the \code{nomutex} keyword to signify that it does not acquire the monitor mutual-exclusion when constructing. This semantics is because an object not yet con\-structed should never be shared and therefore does not require mutual exclusion. Furthermore, it allows the implementation greater freedom when it initializes the monitor locking. The prefix increment operator uses \code{mutex} to protect the incrementing process from race conditions. Finally, there is a conversion operator from \code{counter_t} to \code{size_t}. This conversion may or may not require the \code{mutex} keyword depending on whether or not reading a \code{size_t} is an atomic operation.
 …
 \end{figure}
 Having both \code{mutex} and \code{nomutex} keywords is redundant based on the meaning of a routine having neither of these keywords. For example, it is reasonable that it should default to the safest option (\code{mutex}) when given a routine without qualifiers \code{void foo(counter_t & this)}, whereas assuming \code{nomutex} is unsafe and may cause subtle errors. On the other hand, \code{nomutex} is the ``normal'' parameter behaviour, it effectively states explicitly that ``this routine is not special''. Another alternative is making exactly one of these keywords mandatory, which provides the same semantics but without the ambiguity of supporting routines with neither keyword. Mandatory keywords would also have the added benefit of being self-documented but at the cost of extra typing. While there are several benefits to mandatory keywords, they do bring a few challenges. Mandatory keywords in \CFA would imply that the compiler must know without doubt whether or not a parameter is a monitor or not. Since \CFA relies heavily on traits as an abstraction mechanism, the distinction between a type that is a monitor and a type that looks like a monitor can become blurred. For this reason, \CFA only has the \code{mutex} keyword and uses no keyword to mean \code{nomutex}.
+Having both \code{mutex} and \code{nomutex} keywords can be redundant, depending on the meaning of a routine having neither of these keywords. For example, it is reasonable that it should default to the safest option (\code{mutex}) when given a routine without qualifiers \code{void foo(counter_t & this)}, whereas assuming \code{nomutex} is unsafe and may cause subtle errors. On the other hand, \code{nomutex} is the ``normal'' parameter behaviour, it effectively states explicitly that ``this routine is not special''. Another alternative is making exactly one of these keywords mandatory, which provides the same semantics but without the ambiguity of supporting routines with neither keyword. Mandatory keywords would also have the added benefit of being self-documented but at the cost of extra typing. While there are several benefits to mandatory keywords, they do bring a few challenges. Mandatory keywords in \CFA would imply that the compiler must know without doubt whether or not a parameter is a monitor or not. Since \CFA relies heavily on traits as an abstraction mechanism, the distinction between a type that is a monitor and a type that looks like a monitor can become blurred. For this reason, \CFA only has the \code{mutex} keyword and uses no keyword to mean \code{nomutex}.
 The next semantic decision is to establish when \code{mutex} may be used as a type qualifier. Consider the following declarations:
 \begin{cfacode}
 int f1(monitor& mutex m);
+int f1(monitor & mutex m);
 int f2(const monitor & mutex m);
 int f3(monitor** mutex m);
 int f4(monitor* mutex m []);
 int f5(graph(monitor*) & mutex m);
+int f3(monitor ** mutex m);
+int f4(monitor * mutex m []);
+int f5(graph(monitor *) & mutex m);
 \end{cfacode}
 The problem is to identify which object(s) should be acquired. Furthermore, each object needs to be acquired only once. In the case of simple routines like \code{f1} and \code{f2} it is easy to identify an exhaustive list of objects to acquire on entry. Adding indirections (\code{f3}) still allows the compiler and programmer to identify which object is acquired. However, adding in arrays (\code{f4}) makes it much harder. Array lengths are not necessarily known in C, and even then, making sure objects are only acquired once becomes none-trivial. This problem can be extended to absurd limits like \code{f5}, which uses a graph of monitors. To make the issue tractable, this project imposes the requirement that a routine may only acquire one monitor per parameter and it must be the type of the parameter with at most one level of indirection (ignoring potential qualifiers). Also note that while routine \code{f3} can be supported, meaning that monitor \code{**m} is acquired, passing an array to this routine would be type-safe and yet result in undefined behaviour because only the first element of the array is acquired. However, this ambiguity is part of the C type-system with respects to arrays. For this reason, \code{mutex} is disallowed in the context where arrays may be passed:
 \begin{cfacode}
 int f1(monitor& mutex m);   //Okay : recommended case
 int f2(monitor* mutex m);   //Okay : could be an array but probably not
+int f1(monitor & mutex m);    //Okay : recommended case
+int f2(monitor * mutex m);    //Not Okay : Could be an array
 int f3(monitor mutex m []);  //Not Okay : Array of unknown length
 int f4(monitor** mutex m);  //Not Okay : Could be an array
 int f5(monitor* mutex m []); //Not Okay : Array of unknown length
+int f4(monitor ** mutex m);   //Not Okay : Could be an array
+int f5(monitor * mutex m []); //Not Okay : Array of unknown length
 \end{cfacode}
 Note that not all array functions are actually distinct in the type system. However, even if the code generation could tell the difference, the extra information is still not sufficient to extend meaningfully the monitor call semantic.
 Unlike object-oriented monitors, where calling a mutex member \emph{implicitly} acquires mutual-exclusion of the receiver object, \CFA uses an explicit mechanism to acquire mutual-exclusion. A consequence of this approach is that it extends naturally to multi-monitor calls.
+Unlike object-oriented monitors, where calling a mutex member \emph{implicitly} acquires mutual-exclusion of the receiver object, \CFA uses an explicit mechanism to specify the object that acquires mutual-exclusion. A consequence of this approach is that it extends naturally to multi-monitor calls.
 \begin{cfacode}
 int f(MonitorA & mutex a, MonitorB & mutex b);
 …
 The \gls{multi-acq} monitor lock allows a monitor lock to be acquired by both \code{bar} or \code{baz} and acquired again in \code{foo}. In the calls to \code{bar} and \code{baz} the monitors are acquired in opposite order.
 However, such use leads to the lock acquiring order problems. In the example above, the user uses implicit ordering in the case of function \code{foo} but explicit ordering in the case of \code{bar} and \code{baz}. This subtle difference means that calling these routines concurrently may lead to deadlock and is therefore Undefined Behaviour. As shown~\cite{Lister77}, solving this problem requires:
+However, such use leads to lock acquiring order problems. In the example above, the user uses implicit ordering in the case of function \code{foo} but explicit ordering in the case of \code{bar} and \code{baz}. This subtle difference means that calling these routines concurrently may lead to deadlock and is therefore undefined behaviour. As shown~\cite{Lister77}, solving this problem requires:
 \begin{enumerate}
         \item Dynamically tracking of the monitor-call order.
+        \item Dynamically tracking the monitor-call order.
         \item Implement rollback semantics.
 \end{enumerate}
 …
 \subsection{\code{mutex} statement} \label{mutex-stmt}
 The call semantics discussed above have one software engineering issue, only a named routine can acquire the mutual-exclusion of a set of monitor. \CFA offers the \code{mutex} statement to work around the need for unnecessary names, avoiding a major software engineering problem~\cite{2FTwoHardThings}. Table \ref{lst:mutex-stmt} shows an example of the \code{mutex} statement, which introduces a new scope in which the mutual-exclusion of a set of monitor is acquired. Beyond naming, the \code{mutex} statement has no semantic difference from a routine call with \code{mutex} parameters.
+The call semantics discussed above have one software engineering issue: only a routine can acquire the mutual-exclusion of a set of monitor. \CFA offers the \code{mutex} statement to work around the need for unnecessary names, avoiding a major software engineering problem~\cite{2FTwoHardThings}. Table \ref{lst:mutex-stmt} shows an example of the \code{mutex} statement, which introduces a new scope in which the mutual-exclusion of a set of monitor is acquired. Beyond naming, the \code{mutex} statement has no semantic difference from a routine call with \code{mutex} parameters.
 \begin{table}
 …
 \end{cfacode}
 Like threads and coroutines, monitors are defined in terms of traits with some additional language support in the form of the \code{monitor} keyword. The monitor trait is :
+Like threads and coroutines, monitors are defined in terms of traits with some additional language support in the form of the \code{monitor} keyword. The monitor trait is:
 \begin{cfacode}
 trait is_monitor(dtype T) {
 …
 };
 \end{cfacode}
 Note that the destructor of a monitor must be a \code{mutex} routine to prevent deallocation while a thread is accessing the monitor. As with any object, calls to a monitor, using \code{mutex} or otherwise, is Undefined Behaviour after the destructor has run.
+Note that the destructor of a monitor must be a \code{mutex} routine to prevent deallocation while a thread is accessing the monitor. As with any object, calls to a monitor, using \code{mutex} or otherwise, is undefined behaviour after the destructor has run.
 % ======================================================================
 …
 % ======================================================================
 % ======================================================================
 In addition to mutual exclusion, the monitors at the core of \CFA's concurrency can also be used to achieve synchronization. With monitors, this capability is generally achieved with internal or external scheduling as in~\cite{Hoare74}. Since internal scheduling within a single monitor is mostly a solved problem, this thesis concentrates on extending internal scheduling to multiple monitors. Indeed, like the \gls{bulk-acq} semantics, internal scheduling extends to multiple monitors in a way that is natural to the user but requires additional complexity on the implementation side.
 First, here is a simple example of internal scheduling :
+In addition to mutual exclusion, the monitors at the core of \CFA's concurrency can also be used to achieve synchronization. With monitors, this capability is generally achieved with internal or external scheduling as in~\cite{Hoare74}. With \textbf{scheduling} loosely defined as deciding which thread acquires the critical section next, \textbf{internal scheduling} means making the decision from inside the critical section (i.e., with access to the shared state), while \textbf{external scheduling} means making the decision when entering the critical section (i.e., without access to the shared state). Since internal scheduling within a single monitor is mostly a solved problem, this thesis concentrates on extending internal scheduling to multiple monitors. Indeed, like the \gls{bulk-acq} semantics, internal scheduling extends to multiple monitors in a way that is natural to the user but requires additional complexity on the implementation side.
+First, here is a simple example of internal scheduling:
 \begin{cfacode}
 …
+}
 \end{cfacode}
 There are two details to note here. First, the \code{signal} is a delayed operation, it only unblocks the waiting thread when it reaches the end of the critical section. This semantic is needed to respect mutual-exclusion, i.e., the signaller and signalled thread cannot be in the monitor simultaneously. The alternative is to return immediately after the call to \code{signal}, which is significantly more restrictive. Second, in \CFA, while it is common to store a \code{condition} as a field of the monitor, a \code{condition} variable can be stored/created independently of a monitor. Here routine \code{foo} waits for the \code{signal} from \code{bar} before making further progress, ensuring a basic ordering.
 An important aspect of the implementation is that \CFA does not allow barging, which means that once function \code{bar} releases the monitor, \code{foo} is guaranteed to resume immediately after (unless some other thread waited on the same condition). This guarantee offers the benefit of not having to loop around waits to recheck that a condition is met. The main reason \CFA offers this guarantee is that users can easily introduce barging if it becomes a necessity but adding barging prevention or barging avoidance is more involved without language support. Supporting barging prevention as well as extending internal scheduling to multiple monitors is the main source of complexity in the design and implementation of \CFA concurrency.
+There are two details to note here. First, \code{signal} is a delayed operation; it only unblocks the waiting thread when it reaches the end of the critical section. This semantics is needed to respect mutual-exclusion, i.e., the signaller and signalled thread cannot be in the monitor simultaneously. The alternative is to return immediately after the call to \code{signal}, which is significantly more restrictive. Second, in \CFA, while it is common to store a \code{condition} as a field of the monitor, a \code{condition} variable can be stored/created independently of a monitor. Here routine \code{foo} waits for the \code{signal} from \code{bar} before making further progress, ensuring a basic ordering.
+An important aspect of the implementation is that \CFA does not allow barging, which means that once function \code{bar} releases the monitor, \code{foo} is guaranteed to be the next thread to acquire the monitor (unless some other thread waited on the same condition). This guarantee offers the benefit of not having to loop around waits to recheck that a condition is met. The main reason \CFA offers this guarantee is that users can easily introduce barging if it becomes a necessity but adding barging prevention or barging avoidance is more involved without language support. Supporting barging prevention as well as extending internal scheduling to multiple monitors is the main source of complexity in the design and implementation of \CFA concurrency.
 % ======================================================================
 …
 % ======================================================================
 % ======================================================================
 It is easier to understand the problem of multi-monitor scheduling using a series of pseudo-code examples. Note that for simplicity in the following snippets of pseudo-code, waiting and signalling is done using an implicit condition variable, like Java built-in monitors. Indeed, \code{wait} statements always use the implicit condition variable as parameters and explicitly names the monitors (A and B) associated with the condition. Note that in \CFA, condition variables are tied to a \emph{group} of monitors on first use (called branding), which means that using internal scheduling with distinct sets of monitors requires one condition variable per set of monitors. The example below shows the simple case of having two threads (one for each column) and a single monitor A.
+It is easy to understand the problem of multi-monitor scheduling using a series of pseudo-code examples. Note that for simplicity in the following snippets of pseudo-code, waiting and signalling is done using an implicit condition variable, like Java built-in monitors. Indeed, \code{wait} statements always use the implicit condition variable as parameters and explicitly name the monitors (A and B) associated with the condition. Note that in \CFA, condition variables are tied to a \emph{group} of monitors on first use (called branding), which means that using internal scheduling with distinct sets of monitors requires one condition variable per set of monitors. The example below shows the simple case of having two threads (one for each column) and a single monitor A.
 \begin{multicols}{2}
 …
 \noindent This version uses \gls{bulk-acq} (denoted using the {\sf\&} symbol), but the presence of multiple monitors does not add a particularly new meaning. Synchronization happens between the two threads in exactly the same way and order. The only difference is that mutual exclusion covers a group of monitors. On the implementation side, handling multiple monitors does add a degree of complexity as the next few examples demonstrate.
 While deadlock issues can occur when nesting monitors, these issues are only a symptom of the fact that locks, and by extension monitors, are not perfectly composable. For monitors, a well-known deadlock problem is the Nested Monitor Problem~\cite{Lister77}, which occurs when a \code{wait} is made by a thread that holds more than one monitor. For example, the following pseudo-code runs into the nested-monitor problem :
+While deadlock issues can occur when nesting monitors, these issues are only a symptom of the fact that locks, and by extension monitors, are not perfectly composable. For monitors, a well-known deadlock problem is the Nested Monitor Problem~\cite{Lister77}, which occurs when a \code{wait} is made by a thread that holds more than one monitor. For example, the following pseudo-code runs into the nested-monitor problem:
 \begin{multicols}{2}
 \begin{pseudo}
 …
 % ======================================================================
 A larger example is presented to show complex issues for \gls{bulk-acq} and all the implementation options are analyzed. Listing \ref{lst:int-bulk-pseudo} shows an example where \gls{bulk-acq} adds a significant layer of complexity to the internal signalling semantics, and listing \ref{lst:int-bulk-cfa} shows the corresponding \CFA code to implement the pseudo-code in listing \ref{lst:int-bulk-pseudo}. For the purpose of translating the given pseudo-code into \CFA-code, any method of introducing a monitor is acceptable, e.g., \code{mutex} parameters, global variables, pointer parameters, or using locals with the \code{mutex}-statement.
+A larger example is presented to show complex issues for \gls{bulk-acq} and its implementation options are analyzed. Listing \ref{lst:int-bulk-pseudo} shows an example where \gls{bulk-acq} adds a significant layer of complexity to the internal signalling semantics, and listing \ref{lst:int-bulk-cfa} shows the corresponding \CFA code to implement the pseudo-code in listing \ref{lst:int-bulk-pseudo}. For the purpose of translating the given pseudo-code into \CFA-code, any method of introducing a monitor is acceptable, e.g., \code{mutex} parameters, global variables, pointer parameters, or using locals with the \code{mutex} statement.
 \begin{figure}[!t]
 …
 \end{figure}
 The complexity begins at code sections 4 and 8 in listing \ref{lst:int-bulk-pseudo}, which are where the existing semantics of internal scheduling needs to be extended for multiple monitors. The root of the problem is that \gls{bulk-acq} is used in a context where one of the monitors is already acquired and is why it is important to define the behaviour of the previous pseudo-code. When the signaller thread reaches the location where it should ``release \code{A & B}'' (listing \ref{lst:int-bulk-pseudo} line \ref{line:releaseFirst}), it must actually transfer ownership of monitor \code{B} to the waiting thread. This ownership transfer is required in order to prevent barging into \code{B} by another thread, since both the signalling and signalled threads still need monitor \code{A}. There are three options.
+The complexity begins at code sections 4 and 8 in listing \ref{lst:int-bulk-pseudo}, which are where the existing semantics of internal scheduling needs to be extended for multiple monitors. The root of the problem is that \gls{bulk-acq} is used in a context where one of the monitors is already acquired, which is why it is important to define the behaviour of the previous pseudo-code. When the signaller thread reaches the location where it should ``release \code{A & B}'' (listing \ref{lst:int-bulk-pseudo} line \ref{line:releaseFirst}), it must actually transfer ownership of monitor \code{B} to the waiting thread. This ownership transfer is required in order to prevent barging into \code{B} by another thread, since both the signalling and signalled threads still need monitor \code{A}. There are three options:
 \subsubsection{Delaying Signals}
 The obvious solution to solve the problem of multi-monitor scheduling is to keep ownership of all locks until the last lock is ready to be transferred. It can be argued that that moment is when the last lock is no longer needed because this semantics fits most closely to the behaviour of single-monitor scheduling. This solution has the main benefit of transferring ownership of groups of monitors, which simplifies the semantics from multiple objects to a single group of objects, effectively making the existing single-monitor semantic viable by simply changing monitors to monitor groups. This solution releases the monitors once every monitor in a group can be released. However, since some monitors are never released (i.e., the monitor of a thread), this interpretation means a group might never be released. A more interesting interpretation is to transfer the group until it can be disbanded, which means the group is not passed further and a thread can retain its locks.
 However, listing \ref{lst:int-secret} shows this solution can become much more complicated depending on what is executed while secretly holding B at line \ref{line:secret}, while avoiding the need to transfer ownership of a subset of the condition monitors. Listing \ref{lst:dependency} shows a slightly different example where a third thread is waiting on monitor \code{A}, using a different condition variable. Because the third thread is signalled when secretly holding \code{B}, the goal  becomes unreachable. Depending on the order of signals (listing \ref{lst:dependency} line \ref{line:signal-ab} and \ref{line:signal-a}) two cases can happen :
+The obvious solution to the problem of multi-monitor scheduling is to keep ownership of all locks until the last lock is ready to be transferred. It can be argued that that moment is when the last lock is no longer needed, because this semantics fits most closely to the behaviour of single-monitor scheduling. This solution has the main benefit of transferring ownership of groups of monitors, which simplifies the semantics from multiple objects to a single group of objects, effectively making the existing single-monitor semantic viable by simply changing monitors to monitor groups. This solution releases the monitors once every monitor in a group can be released. However, since some monitors are never released (e.g., the monitor of a thread), this interpretation means a group might never be released. A more interesting interpretation is to transfer the group until all its monitors are released, which means the group is not passed further and a thread can retain its locks.
+However, listing \ref{lst:int-secret} shows this solution can become much more complicated depending on what is executed while secretly holding B at line \ref{line:secret}, while avoiding the need to transfer ownership of a subset of the condition monitors. Listing \ref{lst:dependency} shows a slightly different example where a third thread is waiting on monitor \code{A}, using a different condition variable. Because the third thread is signalled when secretly holding \code{B}, the goal  becomes unreachable. Depending on the order of signals (listing \ref{lst:dependency} line \ref{line:signal-ab} and \ref{line:signal-a}) two cases can happen:
 \paragraph{Case 1: thread $\alpha$ goes first.} In this case, the problem is that monitor \code{A} needs to be passed to thread $\beta$ when thread $\alpha$ is done with it.
 …
 Note that ordering is not determined by a race condition but by whether signalled threads are enqueued in FIFO or FILO order. However, regardless of the answer, users can move line \ref{line:signal-a} before line \ref{line:signal-ab} and get the reverse effect for listing \ref{lst:dependency}.
 In both cases, the threads need to be able to distinguish, on a per monitor basis, which ones need to be released and which ones need to be transferred, which means knowing when to dispand a group becomes complex and inefficient (see next section) and therefore effectively precludes this approach.
+In both cases, the threads need to be able to distinguish, on a per monitor basis, which ones need to be released and which ones need to be transferred, which means knowing when to release a group becomes complex and inefficient (see next section) and therefore effectively precludes this approach.
 \subsubsection{Dependency graphs}
 …
 \end{figure}
 In listing \ref{lst:int-bulk-pseudo}, there is a solution that satisfies both barging prevention and mutual exclusion. If ownership of both monitors is transferred to the waiter when the signaller releases \code{A & B} and then the waiter transfers back ownership of \code{A} back to the signaller when it releases it, then the problem is solved (\code{B} is no longer in use at this point). Dynamically finding the correct order is therefore the second possible solution. The problem is effectively resolving a dependency graph of ownership requirements. Here even the simplest of code snippets requires two transfers and it seems to increase in a manner close to polynomial. This complexity explosion can be seen in listing \ref{lst:explosion}, which is just a direct extension to three monitors, requires at least three ownership transfer and has multiple solutions. Furthermore, the presence of multiple solutions for ownership transfer can cause deadlock problems if a specific solution is not consistently picked; In the same way that multiple lock acquiring order can cause deadlocks.
+In listing \ref{lst:int-bulk-pseudo}, there is a solution that satisfies both barging prevention and mutual exclusion. If ownership of both monitors is transferred to the waiter when the signaller releases \code{A & B} and then the waiter transfers back ownership of \code{A} back to the signaller when it releases it, then the problem is solved (\code{B} is no longer in use at this point). Dynamically finding the correct order is therefore the second possible solution. The problem is effectively resolving a dependency graph of ownership requirements. Here even the simplest of code snippets requires two transfers and has a super-linear complexity. This complexity can be seen in listing \ref{lst:explosion}, which is just a direct extension to three monitors, requires at least three ownership transfer and has multiple solutions. Furthermore, the presence of multiple solutions for ownership transfer can cause deadlock problems if a specific solution is not consistently picked; In the same way that multiple lock acquiring order can cause deadlocks.
 \begin{figure}
 \begin{multicols}{2}
 …
 \end{figure}
 Given the three threads example in listing \ref{lst:dependency}, figure \ref{fig:dependency} shows the corresponding dependency graph that results, where every node is a statement of one of the three threads, and the arrows the dependency of that statement (e.g., $\alpha1$ must happen before $\alpha2$). The extra challenge is that this dependency graph is effectively post-mortem, but the runtime system needs to be able to build and solve these graphs as the dependency unfolds. Resolving dependency graphs being a complex and expensive endeavour, this solution is not the preferred one.
+Given the three threads example in listing \ref{lst:dependency}, figure \ref{fig:dependency} shows the corresponding dependency graph that results, where every node is a statement of one of the three threads, and the arrows the dependency of that statement (e.g., $\alpha1$ must happen before $\alpha2$). The extra challenge is that this dependency graph is effectively post-mortem, but the runtime system needs to be able to build and solve these graphs as the dependencies unfold. Resolving dependency graphs being a complex and expensive endeavour, this solution is not the preferred one.
 \subsubsection{Partial Signalling} \label{partial-sig}
 Finally, the solution that is chosen for \CFA is to use partial signalling. Again using listing \ref{lst:int-bulk-pseudo}, the partial signalling solution transfers ownership of monitor \code{B} at lines \ref{line:signal1} to the waiter but does not wake the waiting thread since it is still using monitor \code{A}. Only when it reaches line \ref{line:lastRelease} does it actually wake up the waiting thread. This solution has the benefit that complexity is encapsulated into only two actions, passing monitors to the next owner when they should be released and conditionally waking threads if all conditions are met. This solution has a much simpler implementation than a dependency graph solving algorithms, which is why it was chosen. Furthermore, after being fully implemented, this solution does not appear to have any significant downsides.
 Using partial signalling, listing \ref{lst:dependency} can be solved easily :
+Finally, the solution that is chosen for \CFA is to use partial signalling. Again using listing \ref{lst:int-bulk-pseudo}, the partial signalling solution transfers ownership of monitor \code{B} at lines \ref{line:signal1} to the waiter but does not wake the waiting thread since it is still using monitor \code{A}. Only when it reaches line \ref{line:lastRelease} does it actually wake up the waiting thread. This solution has the benefit that complexity is encapsulated into only two actions: passing monitors to the next owner when they should be released and conditionally waking threads if all conditions are met. This solution has a much simpler implementation than a dependency graph solving algorithms, which is why it was chosen. Furthermore, after being fully implemented, this solution does not appear to have any significant downsides.
+Using partial signalling, listing \ref{lst:dependency} can be solved easily:
 \begin{itemize}
         \item When thread $\gamma$ reaches line \ref{line:release-ab} it transfers monitor \code{B} to thread $\alpha$ and continues to hold monitor \code{A}.
 …
 An important note is that, until now, signalling a monitor was a delayed operation. The ownership of the monitor is transferred only when the monitor would have otherwise been released, not at the point of the \code{signal} statement. However, in some cases, it may be more convenient for users to immediately transfer ownership to the thread that is waiting for cooperation, which is achieved using the \code{signal_block} routine.
 The example in table \ref{tbl:datingservice} highlights the difference in behaviour. As mentioned, \code{signal} only transfers ownership once the current critical section exits, this behaviour requires additional synchronization when a two-way handshake is needed. To avoid this explicit synchronization, the \code{condition} type offers the \code{signal_block} routine, which handles the two-way handshake as shown in the example. This feature removes the need for a second condition variables and simplifies programming. Like every other monitor semantic, \code{signal_block} uses barging prevention, which means mutual-exclusion is baton-passed both on the frond end and the back end of the call to \code{signal_block}, meaning no other thread can acquire the monitor either before or after the call.
+The example in table \ref{tbl:datingservice} highlights the difference in behaviour. As mentioned, \code{signal} only transfers ownership once the current critical section exits; this behaviour requires additional synchronization when a two-way handshake is needed. To avoid this explicit synchronization, the \code{condition} type offers the \code{signal_block} routine, which handles the two-way handshake as shown in the example. This feature removes the need for a second condition variables and simplifies programming. Like every other monitor semantic, \code{signal_block} uses barging prevention, which means mutual-exclusion is baton-passed both on the front end and the back end of the call to \code{signal_block}, meaning no other thread can acquire the monitor either before or after the call.
 % ======================================================================
 …
 % ======================================================================
 % ======================================================================
 An alternative to internal scheduling is external scheduling, e.g., in \uC.
 \begin{center}
+An alternative to internal scheduling is external scheduling (see Table~\ref{tbl:sched}).
+\begin{table}
 \begin{tabular}{|c|c|c|}
 Internal Scheduling & External Scheduling & Go\\
 …
 \end{gocode}
 \end{tabular}
+\end{center}
+This method is more constrained and explicit, which helps users reduce the non-deterministic nature of concurrency. Indeed, as the following examples demonstrates, external scheduling allows users to wait for events from other threads without the concern of unrelated events occurring. External scheduling can generally be done either in terms of control flow (e.g., Ada with \code{accept}, \uC with \code{_Accept}) or in terms of data (e.g., Go with channels). Of course, both of these paradigms have their own strengths and weaknesses, but for this project control-flow semantics was chosen to stay consistent with the rest of the languages semantics. Two challenges specific to \CFA arise when trying to add external scheduling with loose object definitions and multiple-monitor routines. The previous example shows a simple use \code{_Accept} versus \code{wait}/\code{signal} and its advantages. Note that while other languages often use \code{accept}/\code{select} as the core external scheduling keyword, \CFA uses \code{waitfor} to prevent name collisions with existing socket \acrshort{api}s.
+\caption{Different forms of scheduling.}
+\label{tbl:sched}
+\end{table}
+This method is more constrained and explicit, which helps users reduce the non-deterministic nature of concurrency. Indeed, as the following examples demonstrate, external scheduling allows users to wait for events from other threads without the concern of unrelated events occurring. External scheduling can generally be done either in terms of control flow (e.g., Ada with \code{accept}, \uC with \code{_Accept}) or in terms of data (e.g., Go with channels). Of course, both of these paradigms have their own strengths and weaknesses, but for this project, control-flow semantics was chosen to stay consistent with the rest of the languages semantics. Two challenges specific to \CFA arise when trying to add external scheduling with loose object definitions and multiple-monitor routines. The previous example shows a simple use \code{_Accept} versus \code{wait}/\code{signal} and its advantages. Note that while other languages often use \code{accept}/\code{select} as the core external scheduling keyword, \CFA uses \code{waitfor} to prevent name collisions with existing socket \acrshort{api}s.
 For the \code{P} member above using internal scheduling, the call to \code{wait} only guarantees that \code{V} is the last routine to access the monitor, allowing a third routine, say \code{isInUse()}, acquire mutual exclusion several times while routine \code{P} is waiting. On the other hand, external scheduling guarantees that while routine \code{P} is waiting, no other routine than \code{V} can acquire the monitor.
 …
 \end{tabular}
 \end{center}
+For the first two conditions, it is easy to implement a check that can evaluate the condition in a few instructions. However, a fast check for \pscode{monitor accepts me} is much harder to implement depending on the constraints put on the monitors. Indeed, monitors are often expressed as an entry queue and some acceptor queue as in the following figure:
+\begin{figure}[H]
+\begin{center}
+{\resizebox{0.4\textwidth}{!}{\input{monitor}}}
+\end{center}
+\label{fig:monitor}
+For the first two conditions, it is easy to implement a check that can evaluate the condition in a few instructions. However, a fast check for \pscode{monitor accepts me} is much harder to implement depending on the constraints put on the monitors. Indeed, monitors are often expressed as an entry queue and some acceptor queue as in Figure~\ref{fig:ClassicalMonitor}.
+\begin{figure}
+\centering
+\subfloat[Classical Monitor] {
+\label{fig:ClassicalMonitor}
+{\resizebox{0.45\textwidth}{!}{\input{monitor}}}
+}% subfloat
+\qquad
+\subfloat[\Gls{bulk-acq} Monitor] {
+\label{fig:BulkMonitor}
+{\resizebox{0.45\textwidth}{!}{\input{ext_monitor}}}
+}% subfloat
+\caption{External Scheduling Monitor}
 \end{figure}
+There are other alternatives to these pictures, but in the case of this picture, implementing a fast accept check is relatively easy. Restricted to a fixed number of mutex members, N, the accept check reduces to updating a bitmask when the acceptor queue changes, a check that executes in a single instruction even with a fairly large number (e.g., 128) of mutex members. This approach requires a unique dense ordering of routines with an upper-bound and that ordering must be consistent across translation units. For OO languages these constraints are common, since objects only offer adding member routines consistently across translation units via inheritance. However, in \CFA users can extend objects with mutex routines that are only visible in certain translation unit. This means that establishing a program-wide dense-ordering among mutex routines can only be done in the program linking phase, and still could have issues when using dynamically shared objects.
+The alternative is to alter the implementation like this:
+\begin{center}
+{\resizebox{0.4\textwidth}{!}{\input{ext_monitor}}}
+\end{center}
+There are other alternatives to these pictures, but in the case of the left picture, implementing a fast accept check is relatively easy. Restricted to a fixed number of mutex members, N, the accept check reduces to updating a bitmask when the acceptor queue changes, a check that executes in a single instruction even with a fairly large number (e.g., 128) of mutex members. This approach requires a unique dense ordering of routines with an upper-bound and that ordering must be consistent across translation units. For OO languages these constraints are common, since objects only offer adding member routines consistently across translation units via inheritance. However, in \CFA users can extend objects with mutex routines that are only visible in certain translation unit. This means that establishing a program-wide dense-ordering among mutex routines can only be done in the program linking phase, and still could have issues when using dynamically shared objects.
+The alternative is to alter the implementation as in Figure~\ref{fig:BulkMonitor}.
 Here, the mutex routine called is associated with a thread on the entry queue while a list of acceptable routines is kept separate. Generating a mask dynamically means that the storage for the mask information can vary between calls to \code{waitfor}, allowing for more flexibility and extensions. Storing an array of accepted function pointers replaces the single instruction bitmask comparison with dereferencing a pointer followed by a linear search. Furthermore, supporting nested external scheduling (e.g., listing \ref{lst:nest-ext}) may now require additional searches for the \code{waitfor} statement to check if a routine is already queued.
 …
 \end{figure}
 Note that in the second picture, tasks need to always keep track of the monitors associated with mutex routines, and the routine mask needs to have both a function pointer and a set of monitors, as is discussed in the next section. These details are omitted from the picture for the sake of simplicity.
+Note that in the right picture, tasks need to always keep track of the monitors associated with mutex routines, and the routine mask needs to have both a function pointer and a set of monitors, as is discussed in the next section. These details are omitted from the picture for the sake of simplicity.
 At this point, a decision must be made between flexibility and performance. Many design decisions in \CFA achieve both flexibility and performance, for example polymorphic routines add significant flexibility but inlining them means the optimizer can easily remove any runtime cost. Here, however, the cost of flexibility cannot be trivially removed. In the end, the most flexible approach has been chosen since it allows users to write programs that would otherwise be  hard to write. This decision is based on the assumption that writing fast but inflexible locks is closer to a solved problem than writing locks that are as flexible as external scheduling in \CFA.
 …
 \end{cfacode}
 Note that the set of monitors passed to the \code{waitfor} statement must be entirely contained in the set of monitors already acquired in the routine. \code{waitfor} used in any other context is Undefined Behaviour.
 An important behaviour to note is when a set of monitors only match partially :
+Note that the set of monitors passed to the \code{waitfor} statement must be entirely contained in the set of monitors already acquired in the routine. \code{waitfor} used in any other context is undefined behaviour.
+An important behaviour to note is when a set of monitors only match partially:
 \begin{cfacode}
 …
+}
 \end{cfacode}
 While the equivalent can happen when using internal scheduling, the fact that conditions are specific to a set of monitors means that users have to use two different condition variables. In both cases, partially matching monitor sets does not wake-up the waiting thread. It is also important to note that in the case of external scheduling the order of parameters is irrelevant; \code{waitfor(f,a,b)} and \code{waitfor(f,b,a)} are indistinguishable waiting condition.
+While the equivalent can happen when using internal scheduling, the fact that conditions are specific to a set of monitors means that users have to use two different condition variables. In both cases, partially matching monitor sets does not wakeup the waiting thread. It is also important to note that in the case of external scheduling the order of parameters is irrelevant; \code{waitfor(f,a,b)} and \code{waitfor(f,b,a)} are indistinguishable waiting condition.
 % ======================================================================
 …
 % ======================================================================
 Syntactically, the \code{waitfor} statement takes a function identifier and a set of monitors. While the set of monitors can be any list of expressions, the function name is more restricted because the compiler validates at compile time the validity of the function type and the parameters used with the \code{waitfor} statement. It checks that the set of monitors passed in matches the requirements for a function call. Listing \ref{lst:waitfor} shows various usages of the waitfor statement and which are acceptable. The choice of the function type is made ignoring any non-\code{mutex} parameter. One limitation of the current implementation is that it does not handle overloading but overloading is possible.
+Syntactically, the \code{waitfor} statement takes a function identifier and a set of monitors. While the set of monitors can be any list of expressions, the function name is more restricted because the compiler validates at compile time the validity of the function type and the parameters used with the \code{waitfor} statement. It checks that the set of monitors passed in matches the requirements for a function call. Listing \ref{lst:waitfor} shows various usages of the waitfor statement and which are acceptable. The choice of the function type is made ignoring any non-\code{mutex} parameter. One limitation of the current implementation is that it does not handle overloading, but overloading is possible.
 \begin{figure}
 \begin{cfacode}[caption={Various correct and incorrect uses of the waitfor statement},label={lst:waitfor}]
 …
         waitfor(f4, a1);     //Incorrect : f4 ambiguous
         waitfor(f2, a1, b2); //Undefined Behaviour : b2 not mutex
+        waitfor(f2, a1, b2); //Undefined behaviour : b2 not mutex
+}
 \end{cfacode}
 \end{figure}
 Finally, for added flexibility, \CFA supports constructing a complex \code{waitfor} statement using the \code{or}, \code{timeout} and \code{else}. Indeed, multiple \code{waitfor} clauses can be chained together using \code{or}; this chain forms a single statement that uses baton pass to any function that fits one of the function+monitor set passed in. To enable users to tell which accepted function executed, \code{waitfor}s are followed by a statement (including the null statement \code{;}) or a compound statement, which is executed after the clause is triggered. A \code{waitfor} chain can also be followed by a \code{timeout}, to signify an upper bound on the wait, or an \code{else}, to signify that the call should be non-blocking, which checks for a matching function call already arrived and otherwise continues. Any and all of these clauses can be preceded by a \code{when} condition to dynamically toggle the accept clauses on or off based on some current state. Listing \ref{lst:waitfor2}, demonstrates several complex masks and some incorrect ones.
+Finally, for added flexibility, \CFA supports constructing a complex \code{waitfor} statement using the \code{or}, \code{timeout} and \code{else}. Indeed, multiple \code{waitfor} clauses can be chained together using \code{or}; this chain forms a single statement that uses baton pass to any function that fits one of the function+monitor set passed in. To enable users to tell which accepted function executed, \code{waitfor}s are followed by a statement (including the null statement \code{;}) or a compound statement, which is executed after the clause is triggered. A \code{waitfor} chain can also be followed by a \code{timeout}, to signify an upper bound on the wait, or an \code{else}, to signify that the call should be non-blocking, which checks for a matching function call already arrived and otherwise continues. Any and all of these clauses can be preceded by a \code{when} condition to dynamically toggle the accept clauses on or off based on some current state. Listing \ref{lst:waitfor2} demonstrates several complex masks and some incorrect ones.
 \begin{figure}
 …
 % ======================================================================
 % ======================================================================
 An interesting use for the \code{waitfor} statement is destructor semantics. Indeed, the \code{waitfor} statement can accept any \code{mutex} routine, which includes the destructor (see section \ref{data}). However, with the semantics discussed until now, waiting for the destructor does not make any sense since using an object after its destructor is called is undefined behaviour. The simplest approach is to disallow \code{waitfor} on a destructor. However, a more expressive approach is to flip ordering of execution when waiting for the destructor, meaning that waiting for the destructor allows the destructor to run after the current \code{mutex} routine, similarly to how a condition is signalled.
+An interesting use for the \code{waitfor} statement is destructor semantics. Indeed, the \code{waitfor} statement can accept any \code{mutex} routine, which includes the destructor (see section \ref{data}). However, with the semantics discussed until now, waiting for the destructor does not make any sense, since using an object after its destructor is called is undefined behaviour. The simplest approach is to disallow \code{waitfor} on a destructor. However, a more expressive approach is to flip ordering of execution when waiting for the destructor, meaning that waiting for the destructor allows the destructor to run after the current \code{mutex} routine, similarly to how a condition is signalled.
 \begin{figure}
 \begin{cfacode}[caption={Example of an executor which executes action in series until the destructor is called.},label={lst:dtor-order}]

doc/proposals/concurrency/text/frontpgs.tex

-                      r326cd2b
+                      r9cdfb4d0
         \vspace*{2.0cm}
         Waterloo, Ontario, Canada, 2017 \\
+        Waterloo, Ontario, Canada, 2018 \\
         \vspace*{1.0cm}
         \copyright\ Thierry Delisle 2017 \\
+        \copyright\ Thierry Delisle 2018 \\
         \end{center}
 \end{titlepage}
 …
 % \newpage
+% L I S T   O F   T A B L E S
+% -----------------------------
+\addcontentsline{toc}{chapter}{List of Acronyms}
+\printglossary[type=\acronymtype,title={List of Acronyms}]
+\cleardoublepage
+\phantomsection         % allows hyperref to link to the correct page
 % Change page numbering back to Arabic numerals
 \pagenumbering{arabic}

doc/proposals/concurrency/text/future.tex

-                      r326cd2b
+                      r9cdfb4d0
 \subsection{Performance} \label{futur:perf}
 This thesis presents a first implementation of the \CFA runtime. Therefore, there is still significant work to improve performance. Many of the data structures and algorithms may change in the future to more efficient versions. For example, the number of monitors in a single \gls{bulk-acq} is only bound by the stack size, this is probably unnecessarily generous. It may be possible that limiting the number helps increase performance. However, it is not obvious that the benefit would be significant.
+This thesis presents a first implementation of the \CFA concurrency runtime. Therefore, there is still significant work to improve performance. Many of the data structures and algorithms may change in the future to more efficient versions. For example, the number of monitors in a single \gls{bulk-acq} is only bound by the stack size, this is probably unnecessarily generous. It may be possible that limiting the number helps increase performance. However, it is not obvious that the benefit would be significant.
 \subsection{Flexible Scheduling} \label{futur:sched}
 An important part of concurrency is scheduling. Different scheduling algorithms can affect performance (both in terms of average and variation). However, no single scheduler is optimal for all workloads and therefore there is value in being able to change the scheduler for given programs. One solution is to offer various tweaking options to users, allowing the scheduler to be adjusted to the requirements of the workload. However, in order to be truly flexible, it would be interesting to allow users to add arbitrary data and arbitrary scheduling algorithms. For example, a web server could attach Type-of-Service information to threads and have a ``ToS aware'' scheduling algorithm tailored to this specific web server. This path of flexible schedulers will be explored for \CFA.
 \subsection{Non-Blocking IO} \label{futur:nbio}
 While most of the parallelism tools are aimed at data parallelism and control-flow parallelism, many modern workloads are not bound on computation but on IO operations, a common case being web servers and XaaS (anything as a service). These types of workloads often require significant engineering around amortizing costs of blocking IO operations. At its core, Non-Blocking IO is an operating system level feature that allows queuing IO operations (e.g., network operations) and registering for notifications instead of waiting for requests to complete. In this context, the role of the language makes Non-Blocking IO easily available and with low overhead. The current trend is to use asynchronous programming using tools like callbacks and/or futures and promises, which can be seen in frameworks like Node.js~\cite{NodeJs} for JavaScript, Spring MVC~\cite{SpringMVC} for Java and Django~\cite{Django} for Python. However, while these are valid solutions, they lead to code that is harder to read and maintain because it is much less linear.
+\subsection{Non-Blocking I/O} \label{futur:nbio}
+While most of the parallelism tools are aimed at data parallelism and control-flow parallelism, many modern workloads are not bound on computation but on IO operations, a common case being web servers and XaaS (anything as a service). These types of workloads often require significant engineering around amortizing costs of blocking IO operations. At its core, non-blocking I/O is an operating system level feature that allows queuing IO operations (e.g., network operations) and registering for notifications instead of waiting for requests to complete. In this context, the role of the language makes Non-Blocking IO easily available and with low overhead. The current trend is to use asynchronous programming using tools like callbacks and/or futures and promises, which can be seen in frameworks like Node.js~\cite{NodeJs} for JavaScript, Spring MVC~\cite{SpringMVC} for Java and Django~\cite{Django} for Python. However, while these are valid solutions, they lead to code that is harder to read and maintain because it is much less linear.
 \subsection{Other Concurrency Tools} \label{futur:tools}

doc/proposals/concurrency/text/internals.tex

-                      r326cd2b
+                      r9cdfb4d0
 There are several challenges specific to \CFA when implementing concurrency. These challenges are a direct result of \gls{bulk-acq} and loose object definitions. These two constraints are the root cause of most design decisions in the implementation. Furthermore, to avoid contention from dynamically allocating memory in a concurrent environment, the internal-scheduling design is (almost) entirely free of mallocs. This approach avoids the chicken and egg problem~\cite{Chicken} of having a memory allocator that relies on the threading system and a threading system that relies on the runtime. This extra goal means that memory management is a constant concern in the design of the system.
 The main memory concern for concurrency is queues. All blocking operations are made by parking threads onto queues and all queues are designed with intrusive nodes, where each node has pre-allocated link fields for chaining, to avoid the need for memory allocation. Since several concurrency operations can use an unbound amount of memory (depending on \gls{bulk-acq}), statically defining information in the intrusive fields of threads is insufficient.The only way to use a variable amount of memory without requiring memory allocation is to pre-allocate large buffers of memory eagerly and store the information in these buffers. Conveniently, the call stack fits that description and is easy to use, which is why it is used heavily in the implementation of internal scheduling, particularly variable-length arrays. Since stack allocation is based on scopes, the first step of the implementation is to identify the scopes that are available to store the information, and which of these can have a variable-length array. The threads and the condition both have a fixed amount of memory, while mute routines and blocking calls allow for an unbound amount, within the stack size.
+The main memory concern for concurrency is queues. All blocking operations are made by parking threads onto queues and all queues are designed with intrusive nodes, where each node has pre-allocated link fields for chaining, to avoid the need for memory allocation. Since several concurrency operations can use an unbound amount of memory (depending on \gls{bulk-acq}), statically defining information in the intrusive fields of threads is insufficient.The only way to use a variable amount of memory without requiring memory allocation is to pre-allocate large buffers of memory eagerly and store the information in these buffers. Conveniently, the call stack fits that description and is easy to use, which is why it is used heavily in the implementation of internal scheduling, particularly variable-length arrays. Since stack allocation is based on scopes, the first step of the implementation is to identify the scopes that are available to store the information, and which of these can have a variable-length array. The threads and the condition both have a fixed amount of memory, while \code{mutex} routines and blocking calls allow for an unbound amount, within the stack size.
 Note that since the major contributions of this thesis are extending monitor semantics to \gls{bulk-acq} and loose object definitions, any challenges that are not resulting of these characteristics of \CFA are considered as solved problems and therefore not discussed.
 …
 % ======================================================================
 The first step towards the monitor implementation is simple mute routines. In the single monitor case, mutual-exclusion is done using the entry/exit procedure in listing \ref{lst:entry1}. The entry/exit procedures do not have to be extended to support multiple monitors. Indeed it is sufficient to enter/leave monitors one-by-one as long as the order is correct to prevent deadlock~\cite{Havender68}. In \CFA, ordering of monitor acquisition relies on memory ordering. This approach is sufficient because all objects are guaranteed to have distinct non-overlapping memory layouts and mutual-exclusion for a monitor is only defined for its lifetime, meaning that destroying a monitor while it is acquired is Undefined Behaviour. When a mutex call is made, the concerned monitors are aggregated into a variable-length pointer array and sorted based on pointer values. This array persists for the entire duration of the mutual-exclusion and its ordering reused extensively.
+The first step towards the monitor implementation is simple \code{mutex} routines. In the single monitor case, mutual-exclusion is done using the entry/exit procedure in listing \ref{lst:entry1}. The entry/exit procedures do not have to be extended to support multiple monitors. Indeed it is sufficient to enter/leave monitors one-by-one as long as the order is correct to prevent deadlock~\cite{Havender68}. In \CFA, ordering of monitor acquisition relies on memory ordering. This approach is sufficient because all objects are guaranteed to have distinct non-overlapping memory layouts and mutual-exclusion for a monitor is only defined for its lifetime, meaning that destroying a monitor while it is acquired is undefined behaviour. When a mutex call is made, the concerned monitors are aggregated into a variable-length pointer array and sorted based on pointer values. This array persists for the entire duration of the mutual-exclusion and its ordering reused extensively.
 \begin{figure}
 \begin{multicols}{2}
 …
 \end{figure}
 \subsection{ Details: Interaction with polymorphism}
+\subsection{Details: Interaction with polymorphism}
 Depending on the choice of semantics for when monitor locks are acquired, interaction between monitors and \CFA's concept of polymorphism can be more complex to support. However, it is shown that entry-point locking solves most of the issues.
 First of all, interaction between \code{otype} polymorphism and monitors is impossible since monitors do not support copying. Therefore, the main question is how to support \code{dtype} polymorphism. It is important to present the difference between the two acquiring options : \glspl{callsite-locking} and entry-point locking, i.e., acquiring the monitors before making a mutex routine-call or as the first operation of the mutex routine-call. For example:
+First of all, interaction between \code{otype} polymorphism (see Section~\ref{s:ParametricPolymorphism}) and monitors is impossible since monitors do not support copying. Therefore, the main question is how to support \code{dtype} polymorphism. It is important to present the difference between the two acquiring options: \glspl{callsite-locking} and entry-point locking, i.e., acquiring the monitors before making a mutex routine-call or as the first operation of the mutex routine-call. For example:
 \begin{table}[H]
 \begin{center}
 …
 \end{cfacode}
 Both entry point and \gls{callsite-locking} are feasible implementations. The current \CFA implementation uses entry-point locking because it requires less work when using \gls{raii}, effectively transferring the burden of implementation to object construction/destruction. It is harder to use \gls{raii} for call-site locking, as it does not necessarily have an existing scope that matches exactly the scope of the mutual exclusion, i.e.: the function body. For example, the monitor call can appear in the middle of an expression. Furthermore, entry-point locking requires less code generation since any useful routine multiple times, but there is only one entry point for many call sites.
+Both entry point and \gls{callsite-locking} are feasible implementations. The current \CFA implementation uses entry-point locking because it requires less work when using \gls{raii}, effectively transferring the burden of implementation to object construction/destruction. It is harder to use \gls{raii} for call-site locking, as it does not necessarily have an existing scope that matches exactly the scope of the mutual exclusion, i.e., the function body. For example, the monitor call can appear in the middle of an expression. Furthermore, entry-point locking requires less code generation since any useful routine is called multiple times but there is only one entry point for many call sites.
 % ======================================================================
 …
 \end{figure}
+\subsection{Processors}
+Parallelism in \CFA is built around using processors to specify how much parallelism is desired. \CFA processors are object wrappers around kernel threads, specifically \texttt{pthread}s in the current implementation of \CFA. Indeed, any parallelism must go through operating-system libraries. However, \glspl{uthread} are still the main source of concurrency, processors are simply the underlying source of parallelism. Indeed, processor \glspl{kthread} simply fetch a \gls{uthread} from the scheduler and run it; they are effectively executers for user-threads. The main benefit of this approach is that it offers a well-defined boundary between kernel code and user code, for example, kernel thread quiescing, scheduling and interrupt handling. Processors internally use coroutines to take advantage of the existing context-switching semantics.
+\subsection{Stack Management}
+One of the challenges of this system is to reduce the footprint as much as possible. Specifically, all \texttt{pthread}s created also have a stack created with them, which should be used as much as possible. Normally, coroutines also create their own stack to run on, however, in the case of the coroutines used for processors, these coroutines run directly on the \gls{kthread} stack, effectively stealing the processor stack. The exception to this rule is the Main Processor, i.e., the initial \gls{kthread} that is given to any program. In order to respect C user expectations, the stack of the initial kernel thread, the main stack of the program, is used by the main user thread rather than the main processor, which can grow very large.
 \subsection{Context Switching}
+As mentioned in section \ref{coroutine}, coroutines are a stepping stone for implementing threading, because they share the same mechanism for context-switching between different stacks. To improve performance and simplicity, context-switching is implemented using the following assumption: all context-switches happen inside a specific function call. This assumption means that the context-switch only has to copy the callee-saved registers onto the stack and then switch the stack registers with the ones of the target coroutine/thread. Note that the instruction pointer can be left untouched since the context-switch is always inside the same function. Threads, however, do not context-switch between each other directly. They context-switch to the scheduler. This method is called a 2-step context-switch and has the advantage of having a clear distinction between user code and the kernel where scheduling and other system operations happen. Obviously, this doubles the context-switch cost because threads must context-switch to an intermediate stack. The alternative 1-step context-switch uses the stack of the ``from'' thread to schedule and then context-switches directly to the ``to'' thread. However, the performance of the 2-step context-switch is still superior to a \code{pthread_yield} (see section \ref{results}). Additionally, for users in need for optimal performance, it is important to note that having a 2-step context-switch as the default does not prevent \CFA from offering a 1-step context-switch (akin to the Microsoft \code{SwitchToFiber}~\cite{switchToWindows} routine). This option is not currently present in \CFA but the changes required to add it are strictly additive.
+\subsection{Processors}
+Parallelism in \CFA is built around using processors to specify how much parallelism is desired. \CFA processors are object wrappers around kernel threads, specifically pthreads in the current implementation of \CFA. Indeed, any parallelism must go through operating-system libraries. However, \glspl{uthread} are still the main source of concurrency, processors are simply the underlying source of parallelism. Indeed, processor \glspl{kthread} simply fetch a \gls{uthread} from the scheduler and run it; they are effectively executers for user-threads. The main benefit of this approach is that it offers a well-defined boundary between kernel code and user code, for example, kernel thread quiescing, scheduling and interrupt handling. Processors internally use coroutines to take advantage of the existing context-switching semantics.
+\subsection{Stack Management}
+One of the challenges of this system is to reduce the footprint as much as possible. Specifically, all pthreads created also have a stack created with them, which should be used as much as possible. Normally, coroutines also create there own stack to run on, however, in the case of the coroutines used for processors, these coroutines run directly on the \gls{kthread} stack, effectively stealing the processor stack. The exception to this rule is the Main Processor, i.e. the initial \gls{kthread} that is given to any program. In order to respect C user expectations, the stack of the initial kernel thread, the main stack of the program, is used by the main user thread rather than the main processor, which can grow very large.
+As mentioned in section \ref{coroutine}, coroutines are a stepping stone for implementing threading, because they share the same mechanism for context-switching between different stacks. To improve performance and simplicity, context-switching is implemented using the following assumption: all context-switches happen inside a specific function call. This assumption means that the context-switch only has to copy the callee-saved registers onto the stack and then switch the stack registers with the ones of the target coroutine/thread. Note that the instruction pointer can be left untouched since the context-switch is always inside the same function. Threads, however, do not context-switch between each other directly. They context-switch to the scheduler. This method is called a 2-step context-switch and has the advantage of having a clear distinction between user code and the kernel where scheduling and other system operations happen. Obviously, this doubles the context-switch cost because threads must context-switch to an intermediate stack. The alternative 1-step context-switch uses the stack of the ``from'' thread to schedule and then context-switches directly to the ``to'' thread. However, the performance of the 2-step context-switch is still superior to a \code{pthread_yield} (see section \ref{results}). Additionally, for users in need for optimal performance, it is important to note that having a 2-step context-switch as the default does not prevent \CFA from offering a 1-step context-switch (akin to the Microsoft \code{SwitchToFiber}~\cite{switchToWindows} routine). This option is not currently present in \CFA, but the changes required to add it are strictly additive.
 \subsection{Preemption} \label{preemption}
 Finally, an important aspect for any complete threading system is preemption. As mentioned in chapter \ref{basics}, preemption introduces an extra degree of uncertainty, which enables users to have multiple threads interleave transparently, rather than having to cooperate among threads for proper scheduling and CPU distribution. Indeed, preemption is desirable because it adds a degree of isolation among threads. In a fully cooperative system, any thread that runs a long loop can starve other threads, while in a preemptive system, starvation can still occur but it does not rely on every thread having to yield or block on a regular basis, which reduces significantly a programmer burden. Obviously, preemption is not optimal for every workload. However any preemptive system can become a cooperative system by making the time slices extremely large. Therefore, \CFA uses a preemptive threading system.
 Preemption in \CFA is based on kernel timers, which are used to run a discrete-event simulation. Every processor keeps track of the current time and registers an expiration time with the preemption system. When the preemption system receives a change in preemption, it inserts the time in a sorted order and sets a kernel timer for the closest one, effectively stepping through preemption events on each signal sent by the timer. These timers use the Linux signal {\tt SIGALRM}, which is delivered to the process rather than the kernel-thread. This results in an implementation problem, because when delivering signals to a process, the kernel can deliver the signal to any kernel thread for which the signal is not blocked, i.e. :
+Preemption in \CFA\footnote{Note that the implementation of preemption is strongly tied with the underlying threading system. For this reason, only the Linux implementation is cover, \CFA does not run on Windows at the time of writting} is based on kernel timers, which are used to run a discrete-event simulation. Every processor keeps track of the current time and registers an expiration time with the preemption system. When the preemption system receives a change in preemption, it inserts the time in a sorted order and sets a kernel timer for the closest one, effectively stepping through preemption events on each signal sent by the timer. These timers use the Linux signal {\tt SIGALRM}, which is delivered to the process rather than the kernel-thread. This results in an implementation problem, because when delivering signals to a process, the kernel can deliver the signal to any kernel thread for which the signal is not blocked, i.e.:
 \begin{quote}
 A process-directed signal may be delivered to any one of the threads that does not currently have the signal blocked. If more than one of the threads has the signal unblocked, then the kernel chooses an arbitrary thread to which to deliver the signal.
 SIGNAL(7) - Linux Programmer's Manual
 \end{quote}
 For the sake of simplicity and in order to prevent the case of having two threads receiving alarms simultaneously, \CFA programs block the {\tt SIGALRM} signal on every kernel thread except one. Now because of how involuntary context-switches are handled, the kernel thread handling {\tt SIGALRM} cannot also be a processor thread.
 Involuntary context-switching is done by sending signal {\tt SIGUSER1} to the corresponding proces\-sor and having the thread yield from inside the signal handler. This approach effectively context-switches away from the signal handler back to the kernel and the signal-handler frame is eventually unwound when the thread is scheduled again. As a result, a signal-handler can start on one kernel thread and terminate on a second kernel thread (but the same user thread). It is important to note that signal-handlers save and restore signal masks because user-thread migration can cause a signal mask to migrate from one kernel thread to another. This behaviour is only a problem if all kernel threads, among which a user thread can migrate, differ in terms of signal masks\footnote{Sadly, official POSIX documentation is silent on what distinguishes ``async-signal-safe'' functions from other functions.}. However, since the kernel thread handling preemption requires a different signal mask, executing user threads on the kernel-alarm thread can cause deadlocks. For this reason, the alarm thread is in a tight loop around a system call to \code{sigwaitinfo}, requiring very little CPU time for preemption. One final detail about the alarm thread is how to wake it when additional communication is required (e.g., on thread termination). This unblocking is also done using {\tt SIGALRM}, but sent through the \code{pthread_sigqueue}. Indeed, \code{sigwait} can differentiate signals sent from \code{pthread_sigqueue} from signals sent from alarms or the kernel.
+For the sake of simplicity, and in order to prevent the case of having two threads receiving alarms simultaneously, \CFA programs block the {\tt SIGALRM} signal on every kernel thread except one.
+Now because of how involuntary context-switches are handled, the kernel thread handling {\tt SIGALRM} cannot also be a processor thread. Hence, involuntary context-switching is done by sending signal {\tt SIGUSR1} to the corresponding proces\-sor and having the thread yield from inside the signal handler. This approach effectively context-switches away from the signal handler back to the kernel and the signal handler frame is eventually unwound when the thread is scheduled again. As a result, a signal handler can start on one kernel thread and terminate on a second kernel thread (but the same user thread). It is important to note that signal handlers save and restore signal masks because user-thread migration can cause a signal mask to migrate from one kernel thread to another. This behaviour is only a problem if all kernel threads, among which a user thread can migrate, differ in terms of signal masks\footnote{Sadly, official POSIX documentation is silent on what distinguishes ``async-signal-safe'' functions from other functions.}. However, since the kernel thread handling preemption requires a different signal mask, executing user threads on the kernel-alarm thread can cause deadlocks. For this reason, the alarm thread is in a tight loop around a system call to \code{sigwaitinfo}, requiring very little CPU time for preemption. One final detail about the alarm thread is how to wake it when additional communication is required (e.g., on thread termination). This unblocking is also done using {\tt SIGALRM}, but sent through the \code{pthread_sigqueue}. Indeed, \code{sigwait} can differentiate signals sent from \code{pthread_sigqueue} from signals sent from alarms or the kernel.
 \subsection{Scheduler}
 …
 % ======================================================================
 % ======================================================================
 The following figure is the traditional illustration of a monitor (repeated from page~\pageref{fig:monitor} for convenience) :
+The following figure is the traditional illustration of a monitor (repeated from page~\pageref{fig:ClassicalMonitor} for convenience):
 \begin{figure}[H]
 …
 \end{figure}
 This picture and the proper entry and leave algorithms (see listing \ref{lst:entry2}) is the fundamental implementation of internal scheduling. Note that when a thread is moved from the condition to the AS-stack, it is conceptually split the thread into N pieces, where N is the number of monitors specified in the parameter list. The thread is woken up when all the pieces have popped from the AS-stacks and made active. In this picture, the threads are split into halves but this is only because there are two monitors. For a specific signalling operation every monitor needs a piece of thread on its AS-stack.
+This picture and the proper entry and leave algorithms (see listing \ref{lst:entry2}) is the fundamental implementation of internal scheduling. Note that when a thread is moved from the condition to the AS-stack, it is conceptually split into N pieces, where N is the number of monitors specified in the parameter list. The thread is woken up when all the pieces have popped from the AS-stacks and made active. In this picture, the threads are split into halves but this is only because there are two monitors. For a specific signalling operation every monitor needs a piece of thread on its AS-stack.
 \begin{figure}[b]
 …
 \end{figure}
 Some important things to notice about the exit routine. The solution discussed in \ref{intsched} can be seen in the exit routine of listing \ref{lst:entry2}. Basically, the solution boils down to having a separate data structure for the condition queue and the AS-stack, and unconditionally transferring ownership of the monitors but only unblocking the thread when the last monitor has transferred ownership. This solution is deadlock safe as well as preventing any potential barging. The data structures used for the AS-stack are reused extensively for external scheduling, but in the case of internal scheduling, the data is allocated using variable-length arrays on the call stack of the \code{wait} and \code{signal_block} routines.
+The solution discussed in \ref{intsched} can be seen in the exit routine of listing \ref{lst:entry2}. Basically, the solution boils down to having a separate data structure for the condition queue and the AS-stack, and unconditionally transferring ownership of the monitors but only unblocking the thread when the last monitor has transferred ownership. This solution is deadlock safe as well as preventing any potential barging. The data structures used for the AS-stack are reused extensively for external scheduling, but in the case of internal scheduling, the data is allocated using variable-length arrays on the call stack of the \code{wait} and \code{signal_block} routines.
 \begin{figure}[H]
 …
 \end{figure}
 Figure \ref{fig:structs} shows a high-level representation of these data structures. The main idea behind them is that, a thread cannot contain an arbitrary number of intrusive stacks for linking onto monitor. The \code{condition node} is the data structure that is queued onto a condition variable and, when signalled, the condition queue is popped and each \code{condition criterion} is moved to the AS-stack. Once all the criteria have been popped from their respective AS-stacks, the thread is woken up, which is what is shown in listing \ref{lst:entry2}.
+Figure \ref{fig:structs} shows a high-level representation of these data structures. The main idea behind them is that, a thread cannot contain an arbitrary number of intrusive ``next'' pointers for linking onto monitors. The \code{condition node} is the data structure that is queued onto a condition variable and, when signalled, the condition queue is popped and each \code{condition criterion} is moved to the AS-stack. Once all the criteria have been popped from their respective AS-stacks, the thread is woken up, which is what is shown in listing \ref{lst:entry2}.
 % ======================================================================
 …
 % ======================================================================
 % ======================================================================
 Similarly to internal scheduling, external scheduling for multiple monitors relies on the idea that waiting-thread queues are no longer specific to a single monitor, as mentioned in section \ref{extsched}. For internal scheduling, these queues are part of condition variables, which are still unique for a given scheduling operation (e.g., no signal statement uses multiple conditions). However, in the case of external scheduling, there is no equivalent object which is associated with \code{waitfor} statements. This absence means the queues holding the waiting threads must be stored inside at least one of the monitors that is acquired. These monitors being the only objects that have sufficient lifetime and are available on both sides of the \code{waitfor} statement. This requires an algorithm to choose which monitor holds the relevant queue. It is also important that said algorithm be independent of the order in which users list parameters. The proposed algorithm is to fall back on monitor lock ordering (sorting by address) and specify that the monitor that is acquired first is the one with the relevant waiting queue. This assumes that the lock acquiring order is static for the lifetime of all concerned objects but that is a reasonable constraint.
 This algorithm choice has two consequences :
+Similarly to internal scheduling, external scheduling for multiple monitors relies on the idea that waiting-thread queues are no longer specific to a single monitor, as mentioned in section \ref{extsched}. For internal scheduling, these queues are part of condition variables, which are still unique for a given scheduling operation (i.e., no signal statement uses multiple conditions). However, in the case of external scheduling, there is no equivalent object which is associated with \code{waitfor} statements. This absence means the queues holding the waiting threads must be stored inside at least one of the monitors that is acquired. These monitors being the only objects that have sufficient lifetime and are available on both sides of the \code{waitfor} statement. This requires an algorithm to choose which monitor holds the relevant queue. It is also important that said algorithm be independent of the order in which users list parameters. The proposed algorithm is to fall back on monitor lock ordering (sorting by address) and specify that the monitor that is acquired first is the one with the relevant waiting queue. This assumes that the lock acquiring order is static for the lifetime of all concerned objects but that is a reasonable constraint.
+This algorithm choice has two consequences:
 \begin{itemize}
         \item The queue of the monitor with the lowest address is no longer a true FIFO queue because threads can be moved to the front of the queue. These queues need to contain a set of monitors for each of the waiting threads. Therefore, another thread whose set contains the same lowest address monitor but different lower priority monitors may arrive first but enter the critical section after a thread with the correct pairing.
         \item The queue of the lowest priority monitor is both required and potentially unused. Indeed, since it is not known at compile time which monitor is the monitor which has the lowest address, every monitor needs to have the correct queues even though it is possible that some queues go unused for the entire duration of the program, for example if a monitor is only used in a specific pair.
 \end{itemize}
 Therefore, the following modifications need to be made to support external scheduling :
+Therefore, the following modifications need to be made to support external scheduling:
 \begin{itemize}
         \item The threads waiting on the entry-queue need to keep track of which routine it is trying to enter, and using which set of monitors. The \code{mutex} routine already has all the required information on its stack so the thread only needs to keep a pointer to that information.
+        \item The threads waiting on the entry queue need to keep track of which routine they are trying to enter, and using which set of monitors. The \code{mutex} routine already has all the required information on its stack, so the thread only needs to keep a pointer to that information.
         \item The monitors need to keep a mask of acceptable routines. This mask contains for each acceptable routine, a routine pointer and an array of monitors to go with it. It also needs storage to keep track of which routine was accepted. Since this information is not specific to any monitor, the monitors actually contain a pointer to an integer on the stack of the waiting thread. Note that if a thread has acquired two monitors but executes a \code{waitfor} with only one monitor as a parameter, setting the mask of acceptable routines to both monitors will not cause any problems since the extra monitor will not change ownership regardless. This becomes relevant when \code{when} clauses affect the number of monitors passed to a \code{waitfor} statement.
         \item The entry/exit routine need to be updated as shown in listing \ref{lst:entry3}.
+        \item The entry/exit routines need to be updated as shown in listing \ref{lst:entry3}.
 \end{itemize}
 \subsection{External Scheduling - Destructors}
 Finally, to support the ordering inversion of destructors, the code generation needs to be modified to use a special entry routine. This routine is needed because of the storage requirements of the call order inversion. Indeed, when waiting for the destructors, storage is needed for the waiting context and the lifetime of said storage needs to outlive the waiting operation it is needed for. For regular \code{waitfor} statements, the call stack of the routine itself matches this requirement but it is no longer the case when waiting for the destructor since it is pushed on to the AS-stack for later. The waitfor semantics can then be adjusted correspondingly, as seen in listing \ref{lst:entry-dtor}
+Finally, to support the ordering inversion of destructors, the code generation needs to be modified to use a special entry routine. This routine is needed because of the storage requirements of the call order inversion. Indeed, when waiting for the destructors, storage is needed for the waiting context and the lifetime of said storage needs to outlive the waiting operation it is needed for. For regular \code{waitfor} statements, the call stack of the routine itself matches this requirement but it is no longer the case when waiting for the destructor since it is pushed on to the AS-stack for later. The \code{waitfor} semantics can then be adjusted correspondingly, as seen in listing \ref{lst:entry-dtor}
 \begin{figure}

doc/proposals/concurrency/text/intro.tex

-                      r326cd2b
+                      r9cdfb4d0
 \chapter{Introduction}
 % ======================================================================
+This thesis provides a minimal concurrency \acrshort{api} that is simple, efficient and can be reused to build higher-level features. The simplest possible concurrency system is a thread and a lock but this low-level approach is hard to master. An easier approach for users is to support higher-level constructs as the basis of concurrency. Indeed, for highly productive concurrent programming, high-level approaches are much more popular~\cite{HPP:Study}. Examples are task based, message passing and implicit threading. The high-level approach and its minimal \acrshort{api} are tested in a dialect of C, call \CFA. Furthermore, the proposed \acrshort{api} doubles as an early definition of the \CFA language and library. This thesis also comes with an implementation of the concurrency library for \CFA as well as all the required language features added to the source-to-source translator.
+This thesis provides a minimal concurrency \acrshort{api} that is simple, efficient and can be reused to build higher-level features. The simplest possible concurrency system is a thread and a lock but this low-level approach is hard to master. An easier approach for users is to support higher-level constructs as the basis of concurrency. Indeed, for highly productive concurrent programming, high-level approaches are much more popular~\cite{HPP:Study}. Examples are task based, message passing and implicit threading. The high-level approach and its minimal \acrshort{api} are tested in a dialect of C, called \CFA. Furthermore, the proposed \acrshort{api} doubles as an early definition of the \CFA language and library. This thesis also provides an implementation of the concurrency library for \CFA as well as all the required language features added to the source-to-source translator.
 There are actually two problems that need to be solved in the design of concurrency for a programming language: which concurrency and which parallelism tools are available to the programmer. While these two concepts are often combined, they are in fact distinct, requiring different tools~\cite{Buhr05a}. Concurrency tools need to handle mutual exclusion and synchronization, while parallelism tools are about performance, cost and resource utilization.
+In the context of this thesis, a \textbf{thread} is a fundamental unit of execution that runs a sequence of code, generally on a program stack. Having multiple simultaneous threads gives rise to concurrency and generally requires some kind of locking mechanism to ensure proper execution. Correspondingly, \textbf{concurrency} is defined as the concepts and challenges that occur when multiple independent (sharing memory, timing dependencies, etc.) concurrent threads are introduced. Accordingly, \textbf{locking} (and by extension locks) are defined as a mechanism that prevents the progress of certain threads in order to avoid problems due to concurrency. Finally, in this thesis \textbf{parallelism} is distinct from concurrency and is defined as running multiple threads simultaneously. More precisely, parallelism implies \emph{actual} simultaneous execution as opposed to concurrency which only requires \emph{apparent} simultaneous execution. As such, parallelism is only observable in the differences in performance or, more generally, differences in timing.

doc/proposals/concurrency/text/parallelism.tex

-                      r326cd2b
+                      r9cdfb4d0
 % #       #     # #     # #     # ####### ####### ####### ####### ###  #####  #     #
 \chapter{Parallelism}
 Historically, computer performance was about processor speeds and instructions count. However, with heat dissipation being a direct consequence of speed increase, parallelism has become the new source for increased performance~\cite{Sutter05, Sutter05b}. In this decade, it is not longer reasonable to create a high-performance application without caring about parallelism. Indeed, parallelism is an important aspect of performance and more specifically throughput and hardware utilization. The lowest-level approach of parallelism is to use \glspl{kthread} in combination with semantics like \code{fork}, \code{join}, etc. However, since these have significant costs and limitations, \glspl{kthread} are now mostly used as an implementation tool rather than a user oriented one. There are several alternatives to solve these issues that all have strengths and weaknesses. While there are many variations of the presented paradigms, most of these variations do not actually change the guarantees or the semantics, they simply move costs in order to achieve better performance for certain workloads.
+Historically, computer performance was about processor speeds and instruction counts. However, with heat dissipation being a direct consequence of speed increase, parallelism has become the new source for increased performance~\cite{Sutter05, Sutter05b}. In this decade, it is no longer reasonable to create a high-performance application without caring about parallelism. Indeed, parallelism is an important aspect of performance and more specifically throughput and hardware utilization. The lowest-level approach of parallelism is to use \glspl{kthread} in combination with semantics like \code{fork}, \code{join}, etc. However, since these have significant costs and limitations, \glspl{kthread} are now mostly used as an implementation tool rather than a user oriented one. There are several alternatives to solve these issues that all have strengths and weaknesses. While there are many variations of the presented paradigms, most of these variations do not actually change the guarantees or the semantics, they simply move costs in order to achieve better performance for certain workloads.
 \section{Paradigms}
 \subsection{User-Level Threads}
 A direct improvement on the \gls{kthread} approach is to use \glspl{uthread}. These threads offer most of the same features that the operating system already provides but can be used on a much larger scale. This approach is the most powerful solution as it allows all the features of multithreading, while removing several of the more expensive costs of kernel threads. The downside is that almost none of the low-level threading problems are hidden; users still have to think about data races, deadlocks and synchronization issues. These issues can be somewhat alleviated by a concurrency toolkit with strong guarantees but the parallelism toolkit offers very little to reduce complexity in itself.
+A direct improvement on the \gls{kthread} approach is to use \glspl{uthread}. These threads offer most of the same features that the operating system already provides but can be used on a much larger scale. This approach is the most powerful solution as it allows all the features of multithreading, while removing several of the more expensive costs of kernel threads. The downside is that almost none of the low-level threading problems are hidden; users still have to think about data races, deadlocks and synchronization issues. These issues can be somewhat alleviated by a concurrency toolkit with strong guarantees, but the parallelism toolkit offers very little to reduce complexity in itself.
 Examples of languages that support \glspl{uthread} are Erlang~\cite{Erlang} and \uC~\cite{uC++book}.
 \subsection{Fibers : User-Level Threads Without Preemption} \label{fibers}
 A popular variant of \glspl{uthread} is what is often referred to as \glspl{fiber}. However, \glspl{fiber} do not present meaningful semantic differences with \glspl{uthread}. The significant difference between \glspl{uthread} and \glspl{fiber} is the lack of \gls{preemption} in the latter. Advocates of \glspl{fiber} list their high performance and ease of implementation as major strengths but the performance difference between \glspl{uthread} and \glspl{fiber} is controversial, and the ease of implementation, while true, is a weak argument in the context of language design. Therefore this proposal largely ignores fibers.
+A popular variant of \glspl{uthread} is what is often referred to as \glspl{fiber}. However, \glspl{fiber} do not present meaningful semantic differences with \glspl{uthread}. The significant difference between \glspl{uthread} and \glspl{fiber} is the lack of \gls{preemption} in the latter. Advocates of \glspl{fiber} list their high performance and ease of implementation as major strengths, but the performance difference between \glspl{uthread} and \glspl{fiber} is controversial, and the ease of implementation, while true, is a weak argument in the context of language design. Therefore this proposal largely ignores fibers.
 An example of a language that uses fibers is Go~\cite{Go}
 …
 \subsection{Paradigm Performance}
 While the choice between the three paradigms listed above may have significant performance implication, it is difficult to pin down the performance implications of choosing a model at the language level. Indeed, in many situations one of these paradigms may show better performance but it all strongly depends on the workload. Having a large amount of mostly independent units of work to execute almost guarantees that the \gls{pool}-based system has the best performance thanks to the lower memory overhead (i.e., no thread stack per job). However, interactions among jobs can easily exacerbate contention. User-level threads allow fine-grain context switching, which results in better resource utilization, but a context switch is more expensive and the extra control means users need to tweak more variables to get the desired performance. Finally, if the units of uninterrupted work are large enough the paradigm choice is largely amortized by the actual work done.
+While the choice between the three paradigms listed above may have significant performance implications, it is difficult to pin down the performance implications of choosing a model at the language level. Indeed, in many situations one of these paradigms may show better performance but it all strongly depends on the workload. Having a large amount of mostly independent units of work to execute almost guarantees equivalent performance across paradigms and that the \gls{pool}-based system has the best efficiency thanks to the lower memory overhead (i.e., no thread stack per job). However, interactions among jobs can easily exacerbate contention. User-level threads allow fine-grain context switching, which results in better resource utilization, but a context switch is more expensive and the extra control means users need to tweak more variables to get the desired performance. Finally, if the units of uninterrupted work are large, enough the paradigm choice is largely amortized by the actual work done.
 \section{The \protect\CFA\ Kernel : Processors, Clusters and Threads}\label{kernel}
 A \gls{cfacluster} is a group of \gls{kthread} executed in isolation. \Glspl{uthread} are scheduled on the \glspl{kthread} of a given \gls{cfacluster}, allowing organization between \glspl{uthread} and \glspl{kthread}. It is important that \glspl{kthread} belonging to a same \glspl{cfacluster} have homogeneous settings, otherwise migrating a \gls{uthread} from one \gls{kthread} to the other can cause issues. A \gls{cfacluster} also offers a plugable scheduler that can optimize the workload generated by the \glspl{uthread}.
+A \gls{cfacluster} is a group of \glspl{kthread} executed in isolation. \Glspl{uthread} are scheduled on the \glspl{kthread} of a given \gls{cfacluster}, allowing organization between \glspl{uthread} and \glspl{kthread}. It is important that \glspl{kthread} belonging to a same \glspl{cfacluster} have homogeneous settings, otherwise migrating a \gls{uthread} from one \gls{kthread} to the other can cause issues. A \gls{cfacluster} also offers a pluggable scheduler that can optimize the workload generated by the \glspl{uthread}.
 \Glspl{cfacluster} have not been fully implemented in the context of this thesis, currently \CFA only supports one \gls{cfacluster}, the initial one.
+\Glspl{cfacluster} have not been fully implemented in the context of this thesis. Currently \CFA only supports one \gls{cfacluster}, the initial one.
 \subsection{Future Work: Machine Setup}\label{machine}
 While this was not done in the context of this thesis, another important aspect of clusters is affinity. While many common desktop and laptop PCs have homogeneous CPUs, other devices often have more heterogeneous setups. For example, a system using \acrshort{numa} configurations may benefit from users being able to tie clusters and\/or kernel threads to certain CPU cores. OS support for CPU affinity is now common~\cite{affinityLinux, affinityWindows, affinityFreebsd, affinityNetbsd, affinityMacosx} which means it is both possible and desirable for \CFA to offer an abstraction mechanism for portable CPU affinity.
+While this was not done in the context of this thesis, another important aspect of clusters is affinity. While many common desktop and laptop PCs have homogeneous CPUs, other devices often have more heterogeneous setups. For example, a system using \acrshort{numa} configurations may benefit from users being able to tie clusters and/or kernel threads to certain CPU cores. OS support for CPU affinity is now common~\cite{affinityLinux, affinityWindows, affinityFreebsd, affinityNetbsd, affinityMacosx}, which means it is both possible and desirable for \CFA to offer an abstraction mechanism for portable CPU affinity.
 \subsection{Paradigms}\label{cfaparadigms}

doc/proposals/concurrency/text/results.tex

-                      r326cd2b
+                      r9cdfb4d0
 \section{Micro Benchmarks}
 All benchmarks are run using the same harness to produce the results, seen as the \code{BENCH()} macro in the following examples. This macro uses the following logic to benchmark the code :
+All benchmarks are run using the same harness to produce the results, seen as the \code{BENCH()} macro in the following examples. This macro uses the following logic to benchmark the code:
 \begin{pseudo}
 #define BENCH(run, result)
         before = gettime();
         run;
         after  = gettime();
+#define BENCH(run, result) \
+        before = gettime(); \
+        run; \
+        after  = gettime(); \
         result = (after - before) / N;
 \end{pseudo}
 …
 \subsection{Context-Switching}
 The first interesting benchmark is to measure how long context-switches take. The simplest approach to do this is to yield on a thread, which executes a 2-step context switch. In order to make the comparison fair, coroutines also execute a 2-step context-switch (\gls{uthread} to \gls{kthread} then \gls{kthread} to \gls{uthread}), which is a resume/suspend cycle instead of a yield. Listing \ref{lst:ctx-switch} shows the code for coroutines and threads with the results in table \ref{tab:ctx-switch}. All omitted tests are functionally identical to one of these tests.
+The first interesting benchmark is to measure how long context-switches take. The simplest approach to do this is to yield on a thread, which executes a 2-step context switch. Yielding causes the thread to context-switch to the scheduler and back, more precisely: from the \gls{uthread} to the \gls{kthread} then from the \gls{kthread} back to the same \gls{uthread} (or a different one in the general case). In order to make the comparison fair, coroutines also execute a 2-step context-switch by resuming another coroutine which does nothing but suspending in a tight loop, which is a resume/suspend cycle instead of a yield. Listing \ref{lst:ctx-switch} shows the code for coroutines and threads with the results in table \ref{tab:ctx-switch}. All omitted tests are functionally identical to one of these tests. The difference between coroutines and threads can be attributed to the cost of scheduling.
 \begin{figure}
 \begin{multicols}{2}
 …
 \subsection{Mutual-Exclusion}
 The next interesting benchmark is to measure the overhead to enter/leave a critical-section. For monitors, the simplest approach is to measure how long it takes to enter and leave a monitor routine. Listing \ref{lst:mutex} shows the code for \CFA. To put the results in context, the cost of entering a non-inline function and the cost of acquiring and releasing a pthread mutex lock is also measured. The results can be shown in table \ref{tab:mutex}.
+The next interesting benchmark is to measure the overhead to enter/leave a critical-section. For monitors, the simplest approach is to measure how long it takes to enter and leave a monitor routine. Listing \ref{lst:mutex} shows the code for \CFA. To put the results in context, the cost of entering a non-inline function and the cost of acquiring and releasing a \code{pthread_mutex} lock is also measured. The results can be shown in table \ref{tab:mutex}.
 \begin{figure}
 …
 \multicolumn{1}{c |}{} & \multicolumn{1}{c |}{ Median } &\multicolumn{1}{c |}{ Average } & \multicolumn{1}{c |}{ Standard Deviation} \\
 \hline
+Pthreads Condition Variable                     & 5902.5        & 6093.29       & 714.78 \\
 \uC \code{signal}                                       & 322           & 323   & 3.36   \\
 \CFA \code{signal}, 1 \code{monitor}    & 352.5 & 353.11        & 3.66   \\
 …
 \subsection{Object Creation}
 Finally, the last benchmark measures the cost of creation for concurrent objects. Listing \ref{lst:creation} shows the code for pthreads and \CFA threads, with results shown in table \ref{tab:creation}. As with all other benchmarks, all omitted tests are functionally identical to one of these tests. The only note here is that the call stacks of \CFA coroutines are lazily created, therefore without priming the coroutine, the creation cost is very low.
 \begin{figure}
 \begin{center}
+pthread
+Finally, the last benchmark measures the cost of creation for concurrent objects. Listing \ref{lst:creation} shows the code for \texttt{pthread}s and \CFA threads, with results shown in table \ref{tab:creation}. As with all other benchmarks, all omitted tests are functionally identical to one of these tests. The only note here is that the call stacks of \CFA coroutines are lazily created, therefore without priming the coroutine, the creation cost is very low.
+\begin{figure}
+\begin{center}
+\texttt{pthread}
 \begin{ccode}
 int main() {
 …
 \end{cfacode}
 \end{center}
 \begin{cfacode}[caption={Benchmark code for pthreads and \CFA to measure object creation},label={lst:creation}]
+\begin{cfacode}[caption={Benchmark code for \texttt{pthread}s and \CFA to measure object creation},label={lst:creation}]
 \end{cfacode}
 \end{figure}

doc/proposals/concurrency/text/together.tex

-                      r326cd2b
+                      r9cdfb4d0
 \section{Threads As Monitors}
 As it was subtly alluded in section \ref{threads}, \code{thread}s in \CFA are in fact monitors, which means that all monitor features are available when using threads. For example, here is a very simple two thread pipeline that could be used for a simulator of a game engine :
+As it was subtly alluded in section \ref{threads}, \code{thread}s in \CFA are in fact monitors, which means that all monitor features are available when using threads. For example, here is a very simple two thread pipeline that could be used for a simulator of a game engine:
 \begin{figure}[H]
 \begin{cfacode}[caption={Toy simulator using \code{thread}s and \code{monitor}s.},label={lst:engine-v1}]
 …
 \end{cfacode}
 \end{figure}
 One of the obvious complaints of the previous code snippet (other than its toy-like simplicity) is that it does not handle exit conditions and just goes on forever. Luckily, the monitor semantics can also be used to clearly enforce a shutdown order in a concise manner :
+One of the obvious complaints of the previous code snippet (other than its toy-like simplicity) is that it does not handle exit conditions and just goes on forever. Luckily, the monitor semantics can also be used to clearly enforce a shutdown order in a concise manner:
 \begin{figure}[H]
 \begin{cfacode}[caption={Same toy simulator with proper termination condition.},label={lst:engine-v2}]

doc/proposals/concurrency/thesis.tex

-                      r326cd2b
+                      r9cdfb4d0
 \usepackage{calc}
 \usepackage{xspace}
+\usepackage[labelformat=simple]{subfig}
+\renewcommand{\thesubfigure}{(\alph{subfigure})}
 \usepackage{graphicx}
 \usepackage{tabularx}
 …
 \usepackage[usenames]{color}
 \usepackage[pagewise]{lineno}
+\renewcommand{\linenumberfont}{\scriptsize\sffamily}
 \usepackage{fancyhdr}
 \usepackage{float}
-\renewcommand{\linenumberfont}{\scriptsize\sffamily}
 \usepackage{siunitx}
 \sisetup{ binary-units=true }
 \input{style}                                                   % bespoke macros used in the document
+\usepackage{url}
 \usepackage[dvips,plainpages=false,pdfpagelabels,pdfpagemode=UseNone,colorlinks=true,pagebackref=true,linkcolor=blue,citecolor=blue,urlcolor=blue,pagebackref=true,breaklinks=true]{hyperref}
 \usepackage{breakurl}
+\urlstyle{rm}
 \usepackage{tikz}
 \def\checkmark{\tikz\fill[scale=0.4](0,.35) -- (.25,0) -- (1,.7) -- (.25,.15) -- cycle;}
-\renewcommand{\UrlFont}{\small\sf}
 \setlength{\topmargin}{-0.45in}                         % move running title into header
 …
 \input{future}
-\clearpage
-\printglossary[type=\acronymtype]
-\printglossary
 \clearpage
+% B I B L I O G R A P H Y
+% -----------------------------
+\addcontentsline{toc}{chapter}{Bibliography}
 \bibliographystyle{plain}
+\bibliography{cw92,distSharedMem,lfp92,mlw92,parallel,parallelIO,partheory,pl,pldi92,ps,realtime,techreportsPAB,visual,local}
+\bibliography{pl,local}
+\cleardoublepage
+\phantomsection         % allows hyperref to link to the correct page
+% G L O S S A R Y
+% -----------------------------
+\addcontentsline{toc}{chapter}{Glossary}
+\printglossary
+\cleardoublepage
+\phantomsection         % allows hyperref to link to the correct page

doc/proposals/concurrency/version

r326cd2b	r9cdfb4d0
1		0.11.~~299~~
	1	0.11.403

src/CodeGen/FixMain.cc

-                      r326cd2b
+                      r9cdfb4d0
 #include "Common/SemanticError.h"  // for SemanticError
+#include "CodeGen/GenType.h"       // for GenType
 #include "SynTree/Declaration.h"   // for FunctionDecl, operator<<
 #include "SynTree/Type.h"          // for FunctionType
 …
         bool FixMain::replace_main = false;
         std::unique_ptr<FunctionDecl> FixMain::main_signature = nullptr;
+        template<typename container>
+        std::string genTypeAt(const container& p, size_t idx) {
+                return genType((*std::next(p.begin(), idx))->get_type(), "");
+        }
         void FixMain::registerMain(FunctionDecl* functionDecl)
 …
                         os << main_signature->get_scopedMangleName() << "(";
+                        switch(main_signature->get_functionType()->get_parameters().size()) {
+                                case 3: os << "argc, argv, envp"; break;
+                                case 2: os << "argc, argv"; break;
+                        const auto& params = main_signature->get_functionType()->get_parameters();
+                        switch(params.size()) {
+                                case 3: os << "(" << genTypeAt(params, 0) << ")argc, (" << genTypeAt(params, 1) << ")argv, (" << genTypeAt(params, 2) << ")envp"; break;
+                                case 2: os << "(" << genTypeAt(params, 0) << ")argc, (" << genTypeAt(params, 1) << ")argv"; break;
                                 case 0: break;
                                 default : assert(false);

src/CodeTools/TrackLoc.cc

r326cd2b	r9cdfb4d0
64	64	}
65	65	else {
66		assertf( false, "Top level node has no CodeLocation %s", ~~name~~.c_str() );
	66	assertf( false, "Top level node has no CodeLocation %s", toString( node ).c_str() );
67	67	}
68	68	}

src/Common/PassVisitor.h

-                      r326cd2b
+                      r9cdfb4d0
 #include "SynTree/Expression.h"
 #include "SynTree/Constant.h"
+#include "SynTree/TypeSubstitution.h"
+class TypeSubstitution;
 #include "PassVisitor.proto.h"
 …
 };
+#include "SynTree/TypeSubstitution.h"
 #include "PassVisitor.impl.h"

src/Common/PassVisitor.impl.h

-                      r326cd2b
+                      r9cdfb4d0
 template< typename pass_type >
 static inline void acceptAll( std::list< Declaration* > &decls, PassVisitor< pass_type >& visitor ) {
+inline void acceptAll( std::list< Declaration* > &decls, PassVisitor< pass_type >& visitor ) {
         DeclList_t* beforeDecls = visitor.get_beforeDecls();
         DeclList_t* afterDecls  = visitor.get_afterDecls();
 …
 template< typename pass_type >
 static inline void mutateAll( std::list< Declaration* > &decls, PassVisitor< pass_type >& mutator ) {
+inline void mutateAll( std::list< Declaration* > &decls, PassVisitor< pass_type >& mutator ) {
         DeclList_t* beforeDecls = mutator.get_beforeDecls();
         DeclList_t* afterDecls  = mutator.get_afterDecls();

src/Concurrency/Keywords.cc

-                      r326cd2b
+                      r9cdfb4d0
         // Generic keyword implementation
         //=============================================================================================
+        void fixupGenerics(FunctionType * func, StructDecl * decl) {
+                cloneAll(decl->parameters, func->forall);
+                for ( TypeDecl * td : func->forall ) {
+                        strict_dynamic_cast<StructInstType*>(
+                                func->parameters.front()->get_type()->stripReferences()
+                        )->parameters.push_back(
+                                new TypeExpr( new TypeInstType( noQualifiers, td->name, td ) )
+                        );
+                }
+        }
         void ConcurrentSueKeyword::postvisit(StructDecl * decl) {
                 if( decl->name == type_name && decl->body ) {
 …
                 );
                 get_type->get_parameters().push_back( this_decl );
+                get_type->get_parameters().push_back( this_decl->clone() );
                 get_type->get_returnVals().push_back(
                         new ObjectDecl(
 …
+                        )
                 );
+                fixupGenerics(get_type, decl);
                 FunctionDecl * get_decl = new FunctionDecl(
 …
                                 nullptr
                         );
+                }
+                        fixupGenerics(main_type, decl);
+                }
+                delete this_decl;
                 declsToAddBefore.push_back( forward );
 …
                                         new MemberExpr(
                                                 field,
+                                                UntypedExpr::createDeref( new VariableExpr( func->get_functionType()->get_parameters().front() ) )
+                                                new CastExpr(
+                                                        new VariableExpr( func->get_functionType()->get_parameters().front() ),
+                                                        func->get_functionType()->get_parameters().front()->get_type()->stripReferences()->clone()
+                                                )
+                                        )
+                                )

src/ControlStruct/LabelFixer.cc

r326cd2b	r9cdfb4d0
44	44
45	45	void LabelFixer::postvisit( FunctionDecl * functionDecl ) {
46		~~MLEMutator~~ mlemut( resolveJumps(), generator );
	46	PassVisitor<MLEMutator> mlemut( resolveJumps(), generator );
47	47	functionDecl->acceptMutator( mlemut );
48	48	}

src/ControlStruct/MLEMutator.cc

-                      r326cd2b
+                      r9cdfb4d0
         void MLEMutator::fixBlock( std::list< Statement * > &kids ) {
                 for ( std::list< Statement * >::iterator k = kids.begin(); k != kids.end(); k++ ) {
                         *k = (*k)->acceptMutator(*this);
+                        *k = (*k)->acceptMutator(*visitor);
                         if ( ! get_breakLabel().empty() ) {
 …
+        }
+        CompoundStmt* MLEMutator::mutate( CompoundStmt *cmpndStmt ) {
+                bool labeledBlock = !(cmpndStmt->get_labels().empty());
+        void MLEMutator::premutate( CompoundStmt *cmpndStmt ) {
+                visit_children = false;
+                bool labeledBlock = !(cmpndStmt->labels.empty());
                 if ( labeledBlock ) {
                         Label brkLabel = generator->newLabel("blockBreak", cmpndStmt);
 …
                 // a child statement may set the break label - if they do, attach it to the next statement
                 std::list< Statement * > &kids = cmpndStmt->get_kids();
+                std::list< Statement * > &kids = cmpndStmt->kids;
                 fixBlock( kids );
 …
                         enclosingControlStructures.pop_back();
                 } // if
+                return cmpndStmt;
+        }
+        template< typename LoopClass >
+        Statement *MLEMutator::handleLoopStmt( LoopClass *loopStmt ) {
+                // remember this as the most recent enclosing loop, then mutate the body of the loop -- this will determine
+                // whether brkLabel and contLabel are used with branch statements and will recursively do the same to nested
+                // loops
+                Label brkLabel = generator->newLabel("loopBreak", loopStmt);
+                Label contLabel = generator->newLabel("loopContinue", loopStmt);
+                enclosingControlStructures.push_back( Entry( loopStmt, brkLabel, contLabel ) );
+                loopStmt->set_body ( loopStmt->get_body()->acceptMutator( *this ) );
+                assert( ! enclosingControlStructures.empty() );
+                Entry &e = enclosingControlStructures.back();
+                // sanity check that the enclosing loops have been popped correctly
+                assert ( e == loopStmt );
+                // this will take the necessary steps to add definitions of the previous two labels, if they are used.
+                loopStmt->set_body( mutateLoop( loopStmt->get_body(), e ) );
+                enclosingControlStructures.pop_back();
+                return loopStmt;
+        }
+        Statement *MLEMutator::mutate( CaseStmt *caseStmt ) {
+                caseStmt->set_condition( maybeMutate( caseStmt->get_condition(), *this ) );
+                fixBlock( caseStmt->get_statements() );
+                return caseStmt;
+        }
+        template< typename IfClass >
+        Statement *MLEMutator::handleIfStmt( IfClass *ifStmt ) {
+                // generate a label for breaking out of a labeled if
+                bool labeledBlock = !(ifStmt->get_labels().empty());
+                if ( labeledBlock ) {
+                        Label brkLabel = generator->newLabel("blockBreak", ifStmt);
+                        enclosingControlStructures.push_back( Entry( ifStmt, brkLabel ) );
+                } // if
+                Parent::mutate( ifStmt );
+                if ( labeledBlock ) {
+                        if ( ! enclosingControlStructures.back().useBreakExit().empty() ) {
+                                set_breakLabel( enclosingControlStructures.back().useBreakExit() );
+                        } // if
+                        enclosingControlStructures.pop_back();
+                } // if
+                return ifStmt;
+        }
+        template< typename SwitchClass >
+        Statement *MLEMutator::handleSwitchStmt( SwitchClass *switchStmt ) {
+                // generate a label for breaking out of a labeled switch
+                Label brkLabel = generator->newLabel("switchBreak", switchStmt);
+                enclosingControlStructures.push_back( Entry(switchStmt, brkLabel) );
+                mutateAll( switchStmt->get_statements(), *this );
+                Entry &e = enclosingControlStructures.back();
+                assert ( e == switchStmt );
+                // only generate break label if labeled break is used
+                if ( e.isBreakUsed() ) {
+                        // for the purposes of keeping switch statements uniform (i.e. all statements that are direct children of a
+                        // switch should be CastStmts), append the exit label + break to the last case statement; create a default
+                        // case if there are no cases
+                        std::list< Statement * > &statements = switchStmt->get_statements();
+                        if ( statements.empty() ) {
+                                statements.push_back( CaseStmt::makeDefault() );
+                        } // if
+                        if ( CaseStmt * c = dynamic_cast< CaseStmt * >( statements.back() ) ) {
+                                Statement * stmt = new BranchStmt( Label("brkLabel"), BranchStmt::Break );
+                                stmt->labels.push_back( brkLabel );
+                                c->get_statements().push_back( stmt );
+                        } else assert(0); // as of this point, all statements of a switch are still CaseStmts
+                } // if
+                assert ( enclosingControlStructures.back() == switchStmt );
+                enclosingControlStructures.pop_back();
+                return switchStmt;
+        }
+        }
         void addUnused( Statement * stmt, const Label & originalTarget ) {
 …
         Statement *MLEMutator::mutate( BranchStmt *branchStmt ) throw ( SemanticError ) {
                 std::string originalTarget = branchStmt->get_originalTarget();
+        Statement *MLEMutator::postmutate( BranchStmt *branchStmt ) throw ( SemanticError ) {
+                std::string originalTarget = branchStmt->originalTarget;
                 std::list< Entry >::reverse_iterator targetEntry;
 …
                 // branch error checks, get the appropriate label name and create a goto
                 Label exitLabel;
                 switch ( branchStmt->get_type() ) {
+                switch ( branchStmt->type ) {
                   case BranchStmt::Break:
                                 assert( targetEntry->useBreakExit() != "");
 …
                 // add unused attribute to label to silence warnings
                 addUnused( targetEntry->get_controlStructure(), branchStmt->get_originalTarget() );
+                addUnused( targetEntry->get_controlStructure(), branchStmt->originalTarget );
                 // transform break/continue statements into goto to simplify later handling of branches
 …
+        }
+        Statement *MLEMutator::mutate( WhileStmt *whileStmt ) {
+                return handleLoopStmt( whileStmt );
+        }
+        Statement *MLEMutator::mutate( ForStmt *forStmt ) {
+                return handleLoopStmt( forStmt );
+        }
+        Statement *MLEMutator::mutate( IfStmt *ifStmt ) {
+                return handleIfStmt( ifStmt );
+        }
+        Statement *MLEMutator::mutate( SwitchStmt *switchStmt ) {
+                return handleSwitchStmt( switchStmt );
+        template< typename LoopClass >
+        void MLEMutator::prehandleLoopStmt( LoopClass * loopStmt ) {
+                // remember this as the most recent enclosing loop, then mutate the body of the loop -- this will determine
+                // whether brkLabel and contLabel are used with branch statements and will recursively do the same to nested
+                // loops
+                Label brkLabel = generator->newLabel("loopBreak", loopStmt);
+                Label contLabel = generator->newLabel("loopContinue", loopStmt);
+                enclosingControlStructures.push_back( Entry( loopStmt, brkLabel, contLabel ) );
+        }
+        template< typename LoopClass >
+        Statement * MLEMutator::posthandleLoopStmt( LoopClass * loopStmt ) {
+                assert( ! enclosingControlStructures.empty() );
+                Entry &e = enclosingControlStructures.back();
+                // sanity check that the enclosing loops have been popped correctly
+                assert ( e == loopStmt );
+                // this will take the necessary steps to add definitions of the previous two labels, if they are used.
+                loopStmt->set_body( mutateLoop( loopStmt->get_body(), e ) );
+                enclosingControlStructures.pop_back();
+                return loopStmt;
+        }
+        void MLEMutator::premutate( WhileStmt * whileStmt ) {
+                return prehandleLoopStmt( whileStmt );
+        }
+        void MLEMutator::premutate( ForStmt * forStmt ) {
+                return prehandleLoopStmt( forStmt );
+        }
+        Statement * MLEMutator::postmutate( WhileStmt * whileStmt ) {
+                return posthandleLoopStmt( whileStmt );
+        }
+        Statement * MLEMutator::postmutate( ForStmt * forStmt ) {
+                return posthandleLoopStmt( forStmt );
+        }
+        void MLEMutator::premutate( IfStmt * ifStmt ) {
+                // generate a label for breaking out of a labeled if
+                bool labeledBlock = !(ifStmt->get_labels().empty());
+                if ( labeledBlock ) {
+                        Label brkLabel = generator->newLabel("blockBreak", ifStmt);
+                        enclosingControlStructures.push_back( Entry( ifStmt, brkLabel ) );
+                } // if
+        }
+        Statement * MLEMutator::postmutate( IfStmt * ifStmt ) {
+                bool labeledBlock = !(ifStmt->get_labels().empty());
+                if ( labeledBlock ) {
+                        if ( ! enclosingControlStructures.back().useBreakExit().empty() ) {
+                                set_breakLabel( enclosingControlStructures.back().useBreakExit() );
+                        } // if
+                        enclosingControlStructures.pop_back();
+                } // if
+                return ifStmt;
+        }
+        void MLEMutator::premutate( CaseStmt *caseStmt ) {
+                visit_children = false;
+                caseStmt->condition = maybeMutate( caseStmt->condition, *visitor );
+                fixBlock( caseStmt->stmts );
+        }
+        void MLEMutator::premutate( SwitchStmt *switchStmt ) {
+                // generate a label for breaking out of a labeled switch
+                Label brkLabel = generator->newLabel("switchBreak", switchStmt);
+                enclosingControlStructures.push_back( Entry(switchStmt, brkLabel) );
+        }
+        Statement * MLEMutator::postmutate( SwitchStmt * switchStmt ) {
+                Entry &e = enclosingControlStructures.back();
+                assert ( e == switchStmt );
+                // only generate break label if labeled break is used
+                if ( e.isBreakUsed() ) {
+                        // for the purposes of keeping switch statements uniform (i.e. all statements that are direct children of a
+                        // switch should be CastStmts), append the exit label + break to the last case statement; create a default
+                        // case if there are no cases
+                        std::list< Statement * > &statements = switchStmt->statements;
+                        if ( statements.empty() ) {
+                                statements.push_back( CaseStmt::makeDefault() );
+                        } // if
+                        if ( CaseStmt * c = dynamic_cast< CaseStmt * >( statements.back() ) ) {
+                                Statement * stmt = new BranchStmt( Label("brkLabel"), BranchStmt::Break );
+                                stmt->labels.push_back( e.useBreakExit() );
+                                c->stmts.push_back( stmt );
+                        } else assert(0); // as of this point, all statements of a switch are still CaseStmts
+                } // if
+                assert ( enclosingControlStructures.back() == switchStmt );
+                enclosingControlStructures.pop_back();
+                return switchStmt;
+        }
 } // namespace ControlStruct

src/ControlStruct/MLEMutator.h

-                      r326cd2b
+                      r9cdfb4d0
 #include <string>                  // for string
+#include "Common/PassVisitor.h"
 #include "Common/SemanticError.h"  // for SemanticError
 #include "SynTree/Label.h"         // for Label
 …
 namespace ControlStruct {
 class LabelGenerator;
+        class LabelGenerator;
         class MLEMutator : public Mutator {
+        class MLEMutator : public WithVisitorRef<MLEMutator>, public WithShortCircuiting {
                 class Entry;
-                typedef Mutator Parent;
           public:
                 MLEMutator( std::map<Label, Statement *> *t, LabelGenerator *gen = 0 ) : targetTable( t ), breakLabel(std::string("")), generator( gen ) {}
                 ~MLEMutator();
+                virtual CompoundStmt *mutate( CompoundStmt *cmpndStmt ) override;
+                virtual Statement *mutate( WhileStmt *whileStmt ) override;
+                virtual Statement *mutate( ForStmt *forStmt ) override;
+                virtual Statement *mutate( BranchStmt *branchStmt ) throw ( SemanticError ) override;
+                virtual Statement *mutate( CaseStmt *caseStmt ) override;
+                virtual Statement *mutate( IfStmt *ifStmt ) override;
+                virtual Statement *mutate( SwitchStmt *switchStmt ) override;
+                void premutate( CompoundStmt *cmpndStmt );
+                Statement * postmutate( BranchStmt *branchStmt ) throw ( SemanticError );
+                void premutate( WhileStmt *whileStmt );
+                Statement * postmutate( WhileStmt *whileStmt );
+                void premutate( ForStmt *forStmt );
+                Statement * postmutate( ForStmt *forStmt );
+                void premutate( CaseStmt *caseStmt );
+                void premutate( IfStmt *ifStmt );
+                Statement * postmutate( IfStmt *ifStmt );
+                void premutate( SwitchStmt *switchStmt );
+                Statement * postmutate( SwitchStmt *switchStmt );
                 Statement *mutateLoop( Statement *bodyLoop, Entry &e );
 …
                                 loop( _loop ), breakExit( _breakExit ), contExit( _contExit ), breakUsed(false), contUsed(false) {}
                         bool operator==( const Statement *stmt ) { return ( loop == stmt ); }
                         bool operator!=( const Statement *stmt ) { return ( loop != stmt ); }
+                        bool operator==( const Statement *stmt ) { return loop == stmt; }
+                        bool operator!=( const Statement *stmt ) { return loop != stmt; }
                         bool operator==( const Entry &other ) { return ( loop == other.get_controlStructure() ); }
+                        bool operator==( const Entry &other ) { return loop == other.get_controlStructure(); }
                         Statement *get_controlStructure() const { return loop; }
 …
                 template< typename LoopClass >
                 Statement *handleLoopStmt( LoopClass *loopStmt );
+                void prehandleLoopStmt( LoopClass * loopStmt );
+                template< typename IfClass >
+                Statement *handleIfStmt( IfClass *switchStmt );
+                template< typename SwitchClass >
+                Statement *handleSwitchStmt( SwitchClass *switchStmt );
+                template< typename LoopClass >
+                Statement * posthandleLoopStmt( LoopClass * loopStmt );
                 void fixBlock( std::list< Statement * > &kids );

src/GenPoly/Box.cc

-                      r326cd2b
+                      r9cdfb4d0
         Expression *makeOp( const std::string &name, Expression *arg ) {
                 UntypedExpr *expr = new UntypedExpr( new NameExpr( name ) );
                 expr->get_args().push_back( arg );
+                expr->args.push_back( arg );
                 return expr;
+        }
 …
         Expression *makeOp( const std::string &name, Expression *lhs, Expression *rhs ) {
                 UntypedExpr *expr = new UntypedExpr( new NameExpr( name ) );
                 expr->get_args().push_back( lhs );
                 expr->get_args().push_back( rhs );
+                expr->args.push_back( lhs );
+                expr->args.push_back( rhs );
                 return expr;
+        }
 …
         /// Returns the dereference of a local pointer variable
         Expression *derefVar( ObjectDecl *var ) {
                 return makeOp( "*?", new VariableExpr( var ) );
+                return UntypedExpr::createDeref( new VariableExpr( var ) );
+        }
 …
                                 if ( ! isPolyType( arg->get_type() ) ) {
                                         UntypedExpr *deref = new UntypedExpr( new NameExpr( "*?" ) );
+                                        deref->get_args().push_back( new CastExpr( new VariableExpr( param ), new PointerType( Type::Qualifiers(), arg->get_type()->clone() ) ) );
+                                        deref->set_result( arg->get_type()->clone() );
+                                        deref->args.push_back( new CastExpr( new VariableExpr( param ), new PointerType( Type::Qualifiers(), arg->get_type()->clone() ) ) );
+                                        deref->result = arg->get_type()->clone();
+                                        deref->result->set_lvalue( true );
                                         return deref;
                                 } // if

src/GenPoly/Lvalue.cc

-                      r326cd2b
+                      r9cdfb4d0
                         if ( SymTab::dereferenceOperator ) {
                                 VariableExpr * deref = new VariableExpr( SymTab::dereferenceOperator );
                                 deref->set_result( new PointerType( Type::Qualifiers(), deref->get_result() ) );
                                 Type * base = InitTweak::getPointerBase( arg->get_result() );
                                 assertf( base, "expected pointer type in dereference (type was %s)", toString( arg->get_result() ).c_str() );
+                                deref->result = new PointerType( Type::Qualifiers(), deref->result );
+                                Type * base = InitTweak::getPointerBase( arg->result );
+                                assertf( base, "expected pointer type in dereference (type was %s)", toString( arg->result ).c_str() );
                                 ApplicationExpr * ret = new ApplicationExpr( deref, { arg } );
                                 delete ret->get_result();
                                 ret->set_result( base->clone() );
                                 ret->get_result()->set_lvalue( true );
+                                delete ret->result;
+                                ret->result = base->clone();
+                                ret->result->set_lvalue( true );
                                 return ret;
                         } else {
 …
                                         int diff = depth1-depth2;
                                         if ( diff == 0 ) {
+                                                // conversion between references of the same depth
                                                 assertf( depth1 == depth2, "non-intrinsic reference with cast of reference to reference not yet supported: %d %d %s", depth1, depth2, toString( castExpr ).c_str() );
                                                 PRINT( std::cerr << castExpr << std::endl; )
                                                 return castExpr;
                                         } else if ( diff < 0 ) {
+                                                Expression * ret = castExpr->get_arg();
+                                                // conversion from reference to reference with less depth (e.g. int && -> int &): add dereferences
+                                                Expression * ret = castExpr->arg;
                                                 for ( int i = 0; i < diff; ++i ) {
                                                         ret = mkDeref( ret );
+                                                }
+                                                ret->set_env( castExpr->get_env() );
+                                                delete ret->get_result();
+                                                ret->set_result( castExpr->get_result() );
+                                                castExpr->set_env( nullptr );
+                                                castExpr->set_arg( nullptr );
+                                                castExpr->set_result( nullptr );
+                                                ret->env = castExpr->env;
+                                                delete ret->result;
+                                                ret->result = castExpr->result;
+                                                ret->result->set_lvalue( true ); // ensure result is lvalue
+                                                castExpr->env = nullptr;
+                                                castExpr->arg = nullptr;
+                                                castExpr->result = nullptr;
                                                 delete castExpr;
                                                 return ret;
                                         } else if ( diff > 0 ) {
+                                                Expression * ret = castExpr->get_arg();
+                                                // conversion from reference to reference with more depth (e.g. int & -> int &&): add address-of
+                                                Expression * ret = castExpr->arg;
                                                 for ( int i = 0; i < diff; ++i ) {
                                                         ret = new AddressExpr( ret );
+                                                }
                                                 ret->set_env( castExpr->get_env() );
                                                 delete ret->get_result();
                                                 ret->set_result( castExpr->get_result() );
                                                 castExpr->set_env( nullptr );
                                                 castExpr->set_arg( nullptr );
                                                 castExpr->set_result( nullptr );
+                                                ret->env = castExpr->env;
+                                                delete ret->result;
+                                                ret->result = castExpr->result;
+                                                castExpr->env = nullptr;
+                                                castExpr->arg = nullptr;
+                                                castExpr->result = nullptr;
                                                 delete castExpr;
                                                 return ret;
 …
                                         PRINT( std::cerr << castExpr << std::endl; )
                                         return castExpr;
                                 } else if ( castExpr->get_arg()->get_result()->get_lvalue() ) {
+                                } else if ( castExpr->arg->result->get_lvalue() ) {
                                         // conversion from lvalue to reference
                                         // xxx - keep cast, but turn into pointer cast??
 …
                                         PRINT(
                                                 std::cerr << "convert lvalue to reference -- &" << std::endl;
                                                 std::cerr << castExpr->get_arg() << std::endl;
+                                                std::cerr << castExpr->arg << std::endl;
+                                        )
                                         AddressExpr * ret = new AddressExpr( castExpr->get_arg() );
                                         if ( refType->get_base()->get_qualifiers() != castExpr->get_arg()->get_result()->get_qualifiers() ) {
+                                        AddressExpr * ret = new AddressExpr( castExpr->arg );
+                                        if ( refType->base->get_qualifiers() != castExpr->arg->result->get_qualifiers() ) {
                                                 // must keep cast if cast-to type is different from the actual type
                                                 castExpr->set_arg( ret );
+                                                castExpr->arg = ret;
                                                 return castExpr;
+                                        }
                                         ret->set_env( castExpr->get_env() );
                                         delete ret->get_result();
                                         ret->set_result( castExpr->get_result() );
                                         castExpr->set_env( nullptr );
                                         castExpr->set_arg( nullptr );
                                         castExpr->set_result( nullptr );
+                                        ret->env = castExpr->env;
+                                        delete ret->result;
+                                        ret->result = castExpr->result;
+                                        castExpr->env = nullptr;
+                                        castExpr->arg = nullptr;
+                                        castExpr->result = nullptr;
                                         delete castExpr;
                                         return ret;
 …
+                                }
                                 assertf( false, "Only conversions to reference from lvalue are currently supported: %s", toString( castExpr ).c_str() );
                         } else if ( ReferenceType * refType = dynamic_cast< ReferenceType * >( castExpr->get_arg()->get_result() ) ) {
+                        } else if ( ReferenceType * refType = dynamic_cast< ReferenceType * >( castExpr->arg->result ) ) {
                                 (void)refType;
                                 // conversion from reference to rvalue
 …
                                         std::cerr << "was = " << castExpr << std::endl;
+                                )
                                 Expression * ret = castExpr->get_arg();
                                 TypeSubstitution * env = castExpr->get_env();
+                                Expression * ret = castExpr->arg;
+                                TypeSubstitution * env = castExpr->env;
                                 castExpr->set_env( nullptr );
                                 if ( ! isIntrinsicReference( ret ) ) {
 …
                                         ret = mkDeref( ret );
+                                }
                                 if ( ResolvExpr::typesCompatibleIgnoreQualifiers( castExpr->get_result(), castExpr->get_arg()->get_result()->stripReferences(), SymTab::Indexer() ) ) {
+                                if ( ResolvExpr::typesCompatibleIgnoreQualifiers( castExpr->result, castExpr->arg->result->stripReferences(), SymTab::Indexer() ) ) {
                                         // can remove cast if types are compatible, changing expression type to value type
+                                        ret->set_result( castExpr->get_result()->clone() );
+                                        castExpr->set_arg( nullptr );
+                                        ret->result = castExpr->result->clone();
+                                        ret->result->set_lvalue( true );  // ensure result is lvalue
+                                        castExpr->arg = nullptr;
                                         delete castExpr;
                                 } else {
                                         // must keep cast if types are different
                                         castExpr->set_arg( ret );
+                                        castExpr->arg = ret;
                                         ret = castExpr;
+                                }

src/GenPoly/ScrubTyVars.cc

-                      r326cd2b
+                      r9cdfb4d0
 namespace GenPoly {
         Type * ScrubTyVars::mutate( TypeInstType *typeInst ) {
+        Type * ScrubTyVars::postmutate( TypeInstType * typeInst ) {
                 if ( ! tyVars ) {
                         if ( typeInst->get_isFtype() ) {
 …
                                 return new PointerType( Type::Qualifiers(), new FunctionType( Type::Qualifiers(), true ) );
                         } else {
                                 PointerType *ret = new PointerType( Type::Qualifiers(), new VoidType( typeInst->get_qualifiers() ) );
+                                PointerType * ret = new PointerType( Type::Qualifiers(), new VoidType( typeInst->get_qualifiers() ) );
                                 delete typeInst;
                                 return ret;
 …
+                }
                 TyVarMap::const_iterator tyVar = tyVars->find( typeInst->get_name() );
+                TyVarMap::const_iterator tyVar = tyVars->find( typeInst->name );
                 if ( tyVar != tyVars->end() ) {
                         switch ( tyVar->second.kind ) {
 …
                           case TypeDecl::Ttype:
+                                {
                                         PointerType *ret = new PointerType( Type::Qualifiers(), new VoidType( typeInst->get_qualifiers() ) );
+                                        PointerType * ret = new PointerType( Type::Qualifiers(), new VoidType( typeInst->get_qualifiers() ) );
                                         delete typeInst;
                                         return ret;
 …
+        }
         Type * ScrubTyVars::mutateAggregateType( Type *ty ) {
+        Type * ScrubTyVars::mutateAggregateType( Type * ty ) {
                 if ( shouldScrub( ty ) ) {
                         PointerType *ret = new PointerType( Type::Qualifiers(), new VoidType( ty->get_qualifiers() ) );
+                        PointerType * ret = new PointerType( Type::Qualifiers(), new VoidType( ty->get_qualifiers() ) );
                         delete ty;
                         return ret;
 …
+        }
         Type * ScrubTyVars::mutate( StructInstType *structInst ) {
+        Type * ScrubTyVars::postmutate( StructInstType * structInst ) {
                 return mutateAggregateType( structInst );
+        }
         Type * ScrubTyVars::mutate( UnionInstType *unionInst ) {
+        Type * ScrubTyVars::postmutate( UnionInstType * unionInst ) {
                 return mutateAggregateType( unionInst );
+        }
+        Expression * ScrubTyVars::mutate( SizeofExpr *szeof ) {
+        void ScrubTyVars::primeBaseScrub( Type * type ) {
+                // need to determine whether type needs to be scrubbed to determine whether
+                // automatic recursion is necessary
+                if ( Type * t = shouldScrub( type ) ) {
+                        visit_children = false;
+                        GuardValue( dynType );
+                        dynType = t;
+                }
+        }
+        Expression * ScrubTyVars::postmutate( SizeofExpr * szeof ) {
                 // sizeof( T ) => _sizeof_T parameter, which is the size of T
                 if ( Type *dynType = shouldScrub( szeof->get_type() ) ) {
+                if ( dynType ) {
                         Expression *expr = new NameExpr( sizeofName( mangleType( dynType ) ) );
                         return expr;
-                } else {
-                        return Mutator::mutate( szeof );
                 } // if
+                return szeof;
+        }
         Expression * ScrubTyVars::mutate( AlignofExpr *algnof ) {
+        Expression * ScrubTyVars::postmutate( AlignofExpr * algnof ) {
                 // alignof( T ) => _alignof_T parameter, which is the alignment of T
                 if ( Type *dynType = shouldScrub( algnof->get_type() ) ) {
+                if ( dynType ) {
                         Expression *expr = new NameExpr( alignofName( mangleType( dynType ) ) );
                         return expr;
-                } else {
-                        return Mutator::mutate( algnof );
                 } // if
+                return algnof;
+        }
+        Type * ScrubTyVars::mutate( PointerType *pointer ) {
+//              // special case of shouldScrub that takes all TypeInstType pointer bases, even if they're not dynamic
+//              Type *base = pointer->get_base();
+//              Type *dynType = 0;
+//              if ( dynamicOnly ) {
+//                      if ( TypeInstType *typeInst = dynamic_cast< TypeInstType* >( base ) ) {
+//                              if ( tyVars.find( typeInst->get_name() ) != tyVars.end() ) { dynType = typeInst; }
+//                      } else {
+//                              dynType = isDynType( base, tyVars );
+//                      }
+//              } else {
+//                      dynType = isPolyType( base, tyVars );
+//              }
+//              if ( dynType ) {
+                if ( Type *dynType = shouldScrub( pointer->get_base() ) ) {
+                        Type *ret = dynType->acceptMutator( *this );
+        Type * ScrubTyVars::postmutate( PointerType * pointer ) {
+                if ( dynType ) {
+                        Type * ret = dynType->acceptMutator( *visitor );
                         ret->get_qualifiers() |= pointer->get_qualifiers();
                         pointer->set_base( 0 );
+                        pointer->base = nullptr;
                         delete pointer;
                         return ret;
+                }
                 return Mutator::mutate( pointer );
+                return pointer;
+        }
 } // namespace GenPoly

src/GenPoly/ScrubTyVars.h

-                      r326cd2b
+                      r9cdfb4d0
 #include <cassert>            // for assert
+#include "Common/PassVisitor.h"
 #include "GenPoly.h"          // for TyVarMap, isPolyType, isDynType
 #include "SynTree/Mutator.h"  // for Mutator
 …
 namespace GenPoly {
         class ScrubTyVars : public Mutator {
+        struct ScrubTyVars : public WithVisitorRef<ScrubTyVars>, public WithShortCircuiting, public WithGuards {
                 /// Whether to scrub all type variables from the provided map, dynamic type variables from the provided map, or all type variables
                 enum ScrubMode { FromMap, DynamicFromMap, All };
 …
                 static SynTreeClass *scrubAll( SynTreeClass *target );
+                virtual Type* mutate( TypeInstType *typeInst );
+                virtual Type* mutate( StructInstType *structInst );
+                virtual Type* mutate( UnionInstType *unionInst );
+                virtual Expression* mutate( SizeofExpr *szeof );
+                virtual Expression* mutate( AlignofExpr *algnof );
+                virtual Type* mutate( PointerType *pointer );
+                /// determine if children should be visited based on whether base type should be scrubbed.
+                void primeBaseScrub( Type * );
+                void premutate( TypeInstType * ) { visit_children = false; }
+                void premutate( StructInstType * ) { visit_children = false; }
+                void premutate( UnionInstType * ) { visit_children = false; }
+                void premutate( SizeofExpr * szeof ) { primeBaseScrub( szeof->type ); }
+                void premutate( AlignofExpr * algnof ) { primeBaseScrub( algnof->type ); }
+                void premutate( PointerType * pointer ) { primeBaseScrub( pointer->base ); }
+                Type * postmutate( TypeInstType * typeInst );
+                Type * postmutate( StructInstType * structInst );
+                Type * postmutate( UnionInstType * unionInst );
+                Expression * postmutate( SizeofExpr * szeof );
+                Expression * postmutate( AlignofExpr * algnof );
+                Type * postmutate( PointerType * pointer );
           private:
 …
                 const TyVarMap *tyVars;  ///< Type variables to scrub
                 ScrubMode mode;          ///< which type variables to scrub? [FromMap]
+                Type * dynType = nullptr; ///< result of shouldScrub
         };
         template< typename SynTreeClass >
         SynTreeClass * ScrubTyVars::scrub( SynTreeClass *target, const TyVarMap &tyVars ) {
                 ScrubTyVars scrubber( tyVars );
+                PassVisitor<ScrubTyVars> scrubber( tyVars );
                 return static_cast< SynTreeClass * >( target->acceptMutator( scrubber ) );
+        }
 …
         template< typename SynTreeClass >
         SynTreeClass * ScrubTyVars::scrubDynamic( SynTreeClass *target, const TyVarMap &tyVars ) {
                 ScrubTyVars scrubber( tyVars, ScrubTyVars::DynamicFromMap );
+                PassVisitor<ScrubTyVars> scrubber( tyVars, ScrubTyVars::DynamicFromMap );
                 return static_cast< SynTreeClass * >( target->acceptMutator( scrubber ) );
+        }
 …
         template< typename SynTreeClass >
         SynTreeClass * ScrubTyVars::scrubAll( SynTreeClass *target ) {
                 ScrubTyVars scrubber;
+                PassVisitor<ScrubTyVars> scrubber;
                 return static_cast< SynTreeClass * >( target->acceptMutator( scrubber ) );
+        }

src/InitTweak/FixInit.cc

-                      r326cd2b
+                      r9cdfb4d0
 // file "LICENCE" distributed with Cforall.
 //
 // FixInit.h --
+// FixInit.cc --
 //
 // Author           : Rob Schluntz
 …
                 };
                 struct GenStructMemberCalls final : public WithGuards, public WithShortCircuiting, public WithIndexer {
+                struct GenStructMemberCalls final : public WithGuards, public WithShortCircuiting, public WithIndexer, public WithVisitorRef<GenStructMemberCalls> {
                         /// generate default/copy ctor and dtor calls for user-defined struct ctor/dtors
                         /// for any member that is missing a corresponding ctor/dtor call.
 …
                         static void generate( std::list< Declaration * > & translationUnit );
+                        void previsit( FunctionDecl * funcDecl );
+                        void postvisit( FunctionDecl * funcDecl );
+                        void previsit( MemberExpr * memberExpr );
+                        void previsit( ApplicationExpr * appExpr );
+                        void premutate( FunctionDecl * funcDecl );
+                        DeclarationWithType * postmutate( FunctionDecl * funcDecl );
+                        void premutate( MemberExpr * memberExpr );
+                        void premutate( ApplicationExpr * appExpr );
+                        /// Note: this post mutate used to be in a separate visitor. If this pass breaks, one place to examine is whether it is
+                        /// okay for this part of the recursion to occur alongside the rest.
+                        Expression * postmutate( UntypedExpr * expr );
                         SemanticError errors;
 …
                         bool isCtor = false; // true if current function is a constructor
                         StructDecl * structDecl = nullptr;
-                };
-                // very simple resolver-like mutator class - used to
-                // resolve UntypedExprs that are found within newly
-                // generated constructor/destructor calls
-                class MutatingResolver final : public Mutator {
-                  public:
-                        MutatingResolver( SymTab::Indexer & indexer ) : indexer( indexer ) {}
-                        using Mutator::mutate;
-                        virtual DeclarationWithType* mutate( ObjectDecl *objectDecl ) override;
-                        virtual Expression* mutate( UntypedExpr *untypedExpr ) override;
-                  private:
-                        SymTab::Indexer & indexer;
                 };
 …
                 void GenStructMemberCalls::generate( std::list< Declaration * > & translationUnit ) {
                         PassVisitor<GenStructMemberCalls> warner;
                         acceptAll( translationUnit, warner );
+                        mutateAll( translationUnit, warner );
+                }
 …
                         // arrays are not copy constructed, so this should always be an ExprStmt
                         ImplicitCtorDtorStmt * stmt = genCtorDtor( fname, var, cpArg );
+                        ExprStmt * exprStmt = strict_dynamic_cast< ExprStmt * >( stmt->get_callStmt() );
+                        assertf( stmt, "ResolveCopyCtors: genCtorDtor returned nullptr: %s / %s / %s", fname.c_str(), toString( var ).c_str(), toString( cpArg ).c_str() );
+                        ExprStmt * exprStmt = strict_dynamic_cast< ExprStmt * >( stmt->callStmt );
                         Expression * resolved = exprStmt->expr;
                         exprStmt->expr = nullptr; // take ownership of expr
 …
                         } // if
                         delete stmt;
+                        if ( TupleAssignExpr * assign = dynamic_cast< TupleAssignExpr * >( resolved ) ) {
+                                // fix newly generated StmtExpr
+                                postvisit( assign->stmtExpr );
+                        }
                         return resolved;
+                }
 …
                                 static UniqueName retNamer("_tmp_stmtexpr_ret");
-                                // create variable that will hold the result of the stmt expr
                                 result = result->clone();
                                 env->apply( result );
+                                if ( ! InitTweak::isConstructable( result ) ) {
+                                        delete result;
+                                        return;
+                                }
+                                // create variable that will hold the result of the stmt expr
                                 ObjectDecl * ret = ObjectDecl::newObject( retNamer.newName(), result, nullptr );
                                 ret->get_type()->set_const( false );
                                 stmtExpr->get_returnDecls().push_front( ret );
+                                ret->type->set_const( false );
+                                stmtExpr->returnDecls.push_front( ret );
                                 // must have a non-empty body, otherwise it wouldn't have a result
                                 CompoundStmt * body = stmtExpr->get_statements();
+                                CompoundStmt * body = stmtExpr->statements;
                                 assert( ! body->get_kids().empty() );
                                 // must be an ExprStmt, otherwise it wouldn't have a result
                                 ExprStmt * last = strict_dynamic_cast< ExprStmt * >( body->get_kids().back() );
                                 last->set_expr( makeCtorDtor( "?{}", ret, last->get_expr() ) );
                                 stmtExpr->get_dtors().push_front( makeCtorDtor( "^?{}", ret ) );
+                                last->expr = makeCtorDtor( "?{}", ret, last->get_expr() );
+                                stmtExpr->dtors.push_front( makeCtorDtor( "^?{}", ret ) );
                         } // if
+                }
 …
                         // to the outer context, rather than inside of the statement expression.
                         visit_children = false;
                         std::list< Statement * > & stmts = stmtExpr->get_statements()->get_kids();
+                        std::list< Statement * > & stmts = stmtExpr->statements->get_kids();
                         for ( Statement *& stmt : stmts ) {
                                 stmt = stmt->acceptMutator( *visitor );
                         } // for
                         assert( stmtExpr->get_result() );
                         Type * result = stmtExpr->get_result();
+                        assert( stmtExpr->result );
+                        Type * result = stmtExpr->result;
                         if ( ! result->isVoid() ) {
                                 for ( ObjectDecl * obj : stmtExpr->get_returnDecls() ) {
+                                for ( ObjectDecl * obj : stmtExpr->returnDecls ) {
                                         stmtsToAddBefore.push_back( new DeclStmt( obj ) );
                                 } // for
                                 // add destructors after current statement
                                 for ( Expression * dtor : stmtExpr->get_dtors() ) {
+                                for ( Expression * dtor : stmtExpr->dtors ) {
                                         stmtsToAddAfter.push_back( new ExprStmt( dtor ) );
                                 } // for
                                 // must have a non-empty body, otherwise it wouldn't have a result
                                 assert( ! stmts.empty() );
+                                assert( ! stmtExpr->get_returnDecls().empty() );
+                                stmts.push_back( new ExprStmt( new VariableExpr( stmtExpr->get_returnDecls().front() ) ) );
+                                stmtExpr->get_returnDecls().clear();
+                                stmtExpr->get_dtors().clear();
+                        }
+                        assert( stmtExpr->get_returnDecls().empty() );
+                        assert( stmtExpr->get_dtors().empty() );
+                                assertf( ! stmtExpr->returnDecls.empty() || stmtExpr->dtors.empty(), "StmtExpr returns non-void, but no return decls: %s", toString( stmtExpr ).c_str() );
+                                // if there is a return decl, add a use as the last statement; will not have return decl on non-constructable returns
+                                if ( ! stmtExpr->returnDecls.empty() ) {
+                                        stmts.push_back( new ExprStmt( new VariableExpr( stmtExpr->returnDecls.front() ) ) );
+                                }
+                                stmtExpr->returnDecls.clear();
+                                stmtExpr->dtors.clear();
+                        }
+                        assert( stmtExpr->returnDecls.empty() );
+                        assert( stmtExpr->dtors.empty() );
+                }
 …
+                }
                 void GenStructMemberCalls::previsit( FunctionDecl * funcDecl ) {
+                void GenStructMemberCalls::premutate( FunctionDecl * funcDecl ) {
                         GuardValue( function );
                         GuardValue( unhandled );
 …
+                }
                 void GenStructMemberCalls::postvisit( FunctionDecl * funcDecl ) {
+                DeclarationWithType * GenStructMemberCalls::postmutate( FunctionDecl * funcDecl ) {
                         // remove the unhandled objects from usedUninit, because a call is inserted
                         // to handle them - only objects that are later constructed are used uninitialized.
 …
                                                 Statement * callStmt = stmt.front();
-                                                MutatingResolver resolver( indexer );
                                                 try {
                                                         callStmt->acceptMutator( resolver );
+                                                        callStmt->acceptMutator( *visitor );
                                                         if ( isCtor ) {
                                                                 function->get_statements()->push_front( callStmt );
 …
                                 throw errors;
+                        }
+                        return funcDecl;
+                }
 …
+                }
                 void GenStructMemberCalls::previsit( ApplicationExpr * appExpr ) {
+                void GenStructMemberCalls::premutate( ApplicationExpr * appExpr ) {
                         if ( ! checkWarnings( function ) ) {
                                 visit_children = false;
 …
+                }
                 void GenStructMemberCalls::previsit( MemberExpr * memberExpr ) {
+                void GenStructMemberCalls::premutate( MemberExpr * memberExpr ) {
                         if ( ! checkWarnings( function ) || ! isCtor ) {
                                 visit_children = false;
 …
+                }
+                DeclarationWithType * MutatingResolver::mutate( ObjectDecl * objectDecl ) {
+                        // add object to the indexer assumes that there will be no name collisions
+                        // in generated code. If this changes, add mutate methods for entities with
+                        // scope and call {enter,leave}Scope explicitly.
+                        indexer.addId( objectDecl );
+                        return objectDecl;
+                }
+                Expression * MutatingResolver::mutate( UntypedExpr * untypedExpr ) {
+                Expression * GenStructMemberCalls::postmutate( UntypedExpr * untypedExpr ) {
                         Expression * newExpr = untypedExpr;
                         ResolvExpr::findVoidExpression( newExpr, indexer );

src/InitTweak/GenInit.cc

-                      r326cd2b
+                      r9cdfb4d0
 #include "InitTweak.h"             // for isConstExpr, InitExpander, checkIn...
 #include "Parser/LinkageSpec.h"    // for isOverridable, C
+#include "ResolvExpr/Resolver.h"
 #include "SymTab/Autogen.h"        // for genImplicitCall, SizeType
 #include "SymTab/Mangler.h"        // for Mangler
 …
 #include "SynTree/Type.h"          // for Type, ArrayType, Type::Qualifiers
 #include "SynTree/Visitor.h"       // for acceptAll, maybeAccept
+#include "Tuples/Tuples.h"         // for maybeImpure
 namespace InitTweak {
 …
         };
         struct HoistArrayDimension final : public WithDeclsToAdd, public WithShortCircuiting, public WithGuards {
+        struct HoistArrayDimension final : public WithDeclsToAdd, public WithShortCircuiting, public WithGuards, public WithIndexer {
                 /// hoist dimension from array types in object declaration so that it uses a single
                 /// const variable of type size_t, so that side effecting array dimensions are only
 …
                 void premutate( FunctionType * ) { visit_children = false; }
+                // need this so that enumerators are added to the indexer, due to premutate(AggregateDecl *)
+                void premutate( EnumDecl * ) {}
                 void hoist( Type * type );
 …
                                 if ( varExpr->var == retVal ) return;
+                        }
+                        stmtsToAddBefore.push_back( genCtorDtor( "?{}", retVal, returnStmt->get_expr() ) );
+                        Statement * stmt = genCtorDtor( "?{}", retVal, returnStmt->expr );
+                        assertf( stmt, "ReturnFixer: genCtorDtor returned nullptr: %s / %s", toString( retVal ).c_str(), toString( returnStmt->expr ).c_str() );
+                        stmtsToAddBefore.push_back( stmt );
                         // return the retVal object
 …
                         if ( ! arrayType->get_dimension() ) return; // xxx - recursive call to hoist?
+                        // don't need to hoist dimension if it's a constexpr - only need to if there's potential for side effects.
+                        if ( isConstExpr( arrayType->get_dimension() ) ) return;
+                        // need to resolve array dimensions in order to accurately determine if constexpr
+                        ResolvExpr::findSingleExpression( arrayType->dimension, SymTab::SizeType->clone(), indexer );
+                        // array is variable-length when the dimension is not constexpr
+                        arrayType->isVarLen = ! isConstExpr( arrayType->dimension );
+                        // don't need to hoist dimension if it's definitely pure - only need to if there's potential for side effects.
+                        if ( ! Tuples::maybeImpure( arrayType->dimension ) ) return;
                         ObjectDecl * arrayDimension = new ObjectDecl( dimensionName.newName(), storageClasses, LinkageSpec::C, 0, SymTab::SizeType->clone(), new SingleInit( arrayType->get_dimension() ) );
 …
         void HoistArrayDimension::premutate( FunctionDecl * ) {
                 GuardValue( inFunction );
+                inFunction = true;
+        }
 …
                 Type * type = objDecl->get_type();
                 while ( ArrayType * at = dynamic_cast< ArrayType * >( type ) ) {
+                        // must always construct VLAs with an initializer, since this is an error in C
+                        if ( at->isVarLen && objDecl->init ) return true;
                         type = at->get_base();
+                }

src/InitTweak/InitTweak.cc

-                      r326cd2b
+                      r9cdfb4d0
                 void previsit( ConstantExpr * ) {}
+                void previsit( VariableExpr * varExpr ) {
+                        visit_children = false;
+                        if ( EnumInstType * inst = dynamic_cast< EnumInstType * >( varExpr->result ) ) {
+                                long long int value;
+                                if ( inst->baseEnum->valueOf( varExpr->var, value ) ) {
+                                        // enumerators are const expr
+                                        return;
+                                }
+                        }
+                        isConstExpr = false;
+                }
                 bool isConstExpr = true;
         };

src/Parser/DeclarationNode.cc

-                      r326cd2b
+                      r9cdfb4d0
         DeclarationNode * newnode = new DeclarationNode;
         newnode->type = new TypeData( kind == OperKinds::PointTo ? TypeData::Pointer : TypeData::Reference );
+        if ( kind == OperKinds::And ) {
+                // T && is parsed as 'And' operator rather than two references => add a second reference type
+                TypeData * td = new TypeData( TypeData::Reference );
+                td->base = newnode->type;
+                newnode->type = td;
+        }
         if ( qualifiers ) {
                 return newnode->addQualifiers( qualifiers );

src/Parser/parser.yy

-                      r326cd2b
+                      r9cdfb4d0
 // Created On       : Sat Sep  1 20:22:55 2001
 // Last Modified By : Peter A. Buhr
 // Last Modified On : Mon Nov 27 17:23:35 2017
 // Update Count     : 2992
+// Last Modified On : Thu Dec 21 11:32:56 2017
+// Update Count     : 2996
 //
 …
 %type<decl> cfa_typedef_declaration cfa_variable_declaration cfa_variable_specifier
 %type<decl> c_declaration
+%type<decl> c_declaration static_assert
 %type<decl> KR_function_declarator KR_function_no_ptr KR_function_ptr KR_function_array
 %type<decl> KR_declaration_list KR_declaration_list_opt
 …
         | exception_statement
         | asm_statement
+        ;
 labeled_statement:
 …
         c_declaration pop ';'
         | cfa_declaration pop ';'                                                       // CFA
+        | STATICASSERT '(' constant_expression ',' string_literal ')' ';' // C11
+        | static_assert
+        ;
+static_assert:
+        STATICASSERT '(' constant_expression ',' string_literal ')' ';' // C11
                 { throw SemanticError("Static assert is currently unimplemented."); $$ = nullptr; }     // FIX ME
+        ;
 // C declaration syntax is notoriously confusing and error prone. Cforall provides its own type, variable and function
 …
                         $$ = distAttr( $2, $3 );
+                }
+        | static_assert
+        ;

src/ResolvExpr/CastCost.cc

-                      r326cd2b
+                      r9cdfb4d0
 namespace ResolvExpr {
         class CastCost : public ConversionCost {
+        struct CastCost : public ConversionCost {
           public:
                 CastCost( Type *dest, const SymTab::Indexer &indexer, const TypeEnvironment &env );
+                CastCost( Type *dest, const SymTab::Indexer &indexer, const TypeEnvironment &env, CostFunction costFunc );
+                virtual void visit( BasicType *basicType );
+                virtual void visit( PointerType *pointerType );
+                using ConversionCost::previsit;
+                using ConversionCost::postvisit;
+                void postvisit( BasicType * basicType );
+                void postvisit( PointerType * pointerType );
         };
 …
                                 // all typedefs should be gone by this point
                                 TypeDecl *type = strict_dynamic_cast< TypeDecl* >( namedType );
                                 if ( type->get_base() ) {
                                         return castCost( src, type->get_base(), indexer, env ) + Cost::safe;
+                                if ( type->base ) {
+                                        return castCost( src, type->base, indexer, env ) + Cost::safe;
                                 } // if
                         } // if
 …
                 } else if ( ReferenceType * refType = dynamic_cast< ReferenceType * > ( dest ) ) {
                         PRINT( std::cerr << "conversionCost: dest is reference" << std::endl; )
                         return convertToReferenceCost( src, refType, indexer, env, [](Type * t1, Type * t2, const TypeEnvironment & env, const SymTab::Indexer & indexer) {
+                        return convertToReferenceCost( src, refType, indexer, env, [](Type * t1, Type * t2, const SymTab::Indexer & indexer, const TypeEnvironment & env ) {
                                 return ptrsCastable( t1, t2, env, indexer );
                         });
                 } else {
                         CastCost converter( dest, indexer, env );
+                        PassVisitor<CastCost> converter( dest, indexer, env, castCost );
                         src->accept( converter );
                         if ( converter.get_cost() == Cost::infinity ) {
+                        if ( converter.pass.get_cost() == Cost::infinity ) {
                                 return Cost::infinity;
                         } else {
                                 // xxx - why are we adding cost 0 here?
                                 return converter.get_cost() + Cost::zero;
+                                return converter.pass.get_cost() + Cost::zero;
                         } // if
                 } // if
+        }
         CastCost::CastCost( Type *dest, const SymTab::Indexer &indexer, const TypeEnvironment &env )
                 : ConversionCost( dest, indexer, env ) {
+        CastCost::CastCost( Type *dest, const SymTab::Indexer &indexer, const TypeEnvironment &env, CostFunction costFunc )
+                : ConversionCost( dest, indexer, env, costFunc ) {
+        }
         void CastCost::visit( BasicType *basicType ) {
+        void CastCost::postvisit( BasicType *basicType ) {
                 PointerType *destAsPointer = dynamic_cast< PointerType* >( dest );
                 if ( destAsPointer && basicType->isInteger() ) {
 …
+        }
         void CastCost::visit( PointerType *pointerType ) {
+        void CastCost::postvisit( PointerType *pointerType ) {
                 if ( PointerType *destAsPtr = dynamic_cast< PointerType* >( dest ) ) {
                         if ( pointerType->get_qualifiers() <= destAsPtr->get_qualifiers() && typesCompatibleIgnoreQualifiers( pointerType->get_base(), destAsPtr->get_base(), indexer, env ) ) {
+                        if ( pointerType->get_qualifiers() <= destAsPtr->get_qualifiers() && typesCompatibleIgnoreQualifiers( pointerType->base, destAsPtr->base, indexer, env ) ) {
                                 cost = Cost::safe;
                         } else {
                                 TypeEnvironment newEnv( env );
                                 newEnv.add( pointerType->get_forall() );
                                 newEnv.add( pointerType->get_base()->get_forall() );
                                 int castResult = ptrsCastable( pointerType->get_base(), destAsPtr->get_base(), newEnv, indexer );
+                                newEnv.add( pointerType->forall );
+                                newEnv.add( pointerType->base->forall );
+                                int castResult = ptrsCastable( pointerType->base, destAsPtr->base, newEnv, indexer );
                                 if ( castResult > 0 ) {
                                         cost = Cost::safe;

src/ResolvExpr/CommonType.cc

-                      r326cd2b
+                      r9cdfb4d0
                         // special case where one type has a reference depth of 1 larger than the other
                         if ( diff > 0 || diff < 0 ) {
+                                // std::cerr << "reference depth diff: " << diff << std::endl;
                                 Type * result = nullptr;
+                                if ( ReferenceType * ref1 = dynamic_cast< ReferenceType * >( type1 ) ) {
+                                ReferenceType * ref1 = dynamic_cast< ReferenceType * >( type1 );
+                                ReferenceType * ref2 = dynamic_cast< ReferenceType * >( type2 );
+                                if ( diff > 0 ) {
+                                        // deeper on the left
+                                        assert( ref1 );
+                                        result = handleReference( ref1->base, type2, widenFirst, widenSecond, indexer, env, openVars );
+                                } else {
+                                        // deeper on the right
+                                        assert( ref2 );
+                                        result = handleReference( type1, ref2->base, widenFirst, widenSecond, indexer, env, openVars );
+                                }
+                                if ( result && ref1 ) {
                                         // formal is reference, so result should be reference
+                                        result = handleReference( ref1->base, type2, widenFirst, widenSecond, indexer, env, openVars );
+                                        if ( result ) result = new ReferenceType( ref1->get_qualifiers(), result );
+                                } else {
+                                        // formal is value, so result should be value
+                                        ReferenceType * ref2 = strict_dynamic_cast< ReferenceType * > ( type2 );
+                                        result = handleReference( type1, ref2->base, widenFirst, widenSecond, indexer, env, openVars );
+                                        // std::cerr << "formal is reference; result should be reference" << std::endl;
+                                        result = new ReferenceType( ref1->get_qualifiers(), result );
+                                }
                                 // std::cerr << "common type of reference [" << type1 << "] and [" << type2 << "] is [" << result << "]" << std::endl;
 …
+        }
         void CommonType::visit( __attribute((unused)) VoidType *voidType ) {}
+        void CommonType::visit( VoidType * ) {}
         void CommonType::visit( BasicType *basicType ) {
 …
+        }
         void CommonType::visit( __attribute((unused)) ArrayType *arrayType ) {}
+        void CommonType::visit( ArrayType * ) {}
         void CommonType::visit( ReferenceType *refType ) {
 …
+        }
         void CommonType::visit( __attribute((unused)) FunctionType *functionType ) {}
         void CommonType::visit( __attribute((unused)) StructInstType *aggregateUseType ) {}
         void CommonType::visit( __attribute((unused)) UnionInstType *aggregateUseType ) {}
+        void CommonType::visit( FunctionType * ) {}
+        void CommonType::visit( StructInstType * ) {}
+        void CommonType::visit( UnionInstType * ) {}
         void CommonType::visit( EnumInstType *enumInstType ) {
 …
+        }
         void CommonType::visit( __attribute((unused)) TraitInstType *aggregateUseType ) {
+        void CommonType::visit( TraitInstType * ) {
+        }
 …
+        }
         void CommonType::visit( __attribute((unused)) TupleType *tupleType ) {}
         void CommonType::visit( __attribute((unused)) VarArgsType *varArgsType ) {}
+        void CommonType::visit( TupleType * ) {}
+        void CommonType::visit( VarArgsType * ) {}
         void CommonType::visit( ZeroType *zeroType ) {

src/ResolvExpr/ConversionCost.cc

-                      r326cd2b
+                      r9cdfb4d0
                         EqvClass eqvClass;
                         NamedTypeDecl *namedType;
                         PRINT( std::cerr << "type inst " << destAsTypeInst->get_name(); )
                         if ( env.lookup( destAsTypeInst->get_name(), eqvClass ) ) {
+                        PRINT( std::cerr << "type inst " << destAsTypeInst->name; )
+                        if ( env.lookup( destAsTypeInst->name, eqvClass ) ) {
                                 if ( eqvClass.type ) {
                                         return conversionCost( src, eqvClass.type, indexer, env );
 …
                                         return Cost::infinity;
+                                }
                         } else if ( ( namedType = indexer.lookupType( destAsTypeInst->get_name() ) ) ) {
+                        } else if ( ( namedType = indexer.lookupType( destAsTypeInst->name ) ) ) {
                                 PRINT( std::cerr << " found" << std::endl; )
                                 TypeDecl *type = dynamic_cast< TypeDecl* >( namedType );
                                 // all typedefs should be gone by this point
                                 assert( type );
                                 if ( type->get_base() ) {
                                         return conversionCost( src, type->get_base(), indexer, env ) + Cost::safe;
+                                if ( type->base ) {
+                                        return conversionCost( src, type->base, indexer, env ) + Cost::safe;
                                 } // if
                         } // if
 …
                 } else if ( ReferenceType * refType = dynamic_cast< ReferenceType * > ( dest ) ) {
                         PRINT( std::cerr << "conversionCost: dest is reference" << std::endl; )
                         return convertToReferenceCost( src, refType, indexer, env, [](Type * t1, Type * t2, const TypeEnvironment & env, const SymTab::Indexer &){
+                        return convertToReferenceCost( src, refType, indexer, env, [](Type * t1, Type * t2, const SymTab::Indexer &, const TypeEnvironment & env ){
                                 return ptrsAssignable( t1, t2, env );
                         });
                 } else {
                         ConversionCost converter( dest, indexer, env );
+                        PassVisitor<ConversionCost> converter( dest, indexer, env, conversionCost );
                         src->accept( converter );
                         if ( converter.get_cost() == Cost::infinity ) {
+                        if ( converter.pass.get_cost() == Cost::infinity ) {
                                 return Cost::infinity;
                         } else {
                                 return converter.get_cost() + Cost::zero;
+                                return converter.pass.get_cost() + Cost::zero;
                         } // if
                 } // if
 …
         Cost convertToReferenceCost( Type * src, Type * dest, int diff, const SymTab::Indexer & indexer, const TypeEnvironment & env, PtrsFunction func ) {
                 PRINT( std::cerr << "convert to reference cost... diff " << diff << std::endl; )
+                PRINT( std::cerr << "convert to reference cost... diff " << diff << " " << src << " / " << dest << std::endl; )
                 if ( diff > 0 ) {
                         // TODO: document this
                         Cost cost = convertToReferenceCost( strict_dynamic_cast< ReferenceType * >( src )->get_base(), dest, diff-1, indexer, env, func );
+                        Cost cost = convertToReferenceCost( strict_dynamic_cast< ReferenceType * >( src )->base, dest, diff-1, indexer, env, func );
                         cost.incReference();
                         return cost;
                 } else if ( diff < -1 ) {
                         // TODO: document this
                         Cost cost = convertToReferenceCost( src, strict_dynamic_cast< ReferenceType * >( dest )->get_base(), diff+1, indexer, env, func );
+                        Cost cost = convertToReferenceCost( src, strict_dynamic_cast< ReferenceType * >( dest )->base, diff+1, indexer, env, func );
                         cost.incReference();
                         return cost;
 …
                         if ( srcAsRef && destAsRef ) { // pointer-like conversions between references
                                 PRINT( std::cerr << "converting between references" << std::endl; )
+                                if ( srcAsRef->get_base()->get_qualifiers() <= destAsRef->get_base()->get_qualifiers() && typesCompatibleIgnoreQualifiers( srcAsRef->get_base(), destAsRef->get_base(), indexer, env ) ) {
+                                        return Cost::safe;
+                                Type::Qualifiers tq1 = srcAsRef->base->get_qualifiers();
+                                Type::Qualifiers tq2 = destAsRef->base->get_qualifiers();
+                                if ( tq1 <= tq2 && typesCompatibleIgnoreQualifiers( srcAsRef->base, destAsRef->base, indexer, env ) ) {
+                                        PRINT( std::cerr << " :: compatible and good qualifiers" << std::endl; )
+                                        if ( tq1 == tq2 ) {
+                                                // types are the same
+                                                return Cost::zero;
+                                        } else {
+                                                // types are the same, except otherPointer has more qualifiers
+                                                return Cost::safe;
+                                        }
                                 } else {  // xxx - this discards reference qualifiers from consideration -- reducing qualifiers is a safe conversion; is this right?
                                         int assignResult = func( srcAsRef->get_base(), destAsRef->get_base(), env, indexer );
+                                        int assignResult = func( srcAsRef->base, destAsRef->base, indexer, env );
                                         PRINT( std::cerr << "comparing references: " << assignResult << " " << srcAsRef << " " << destAsRef << std::endl; )
                                         if ( assignResult > 0 ) {
 …
                         } else {
                                 PRINT( std::cerr << "reference to rvalue conversion" << std::endl; )
                                 ConversionCost converter( dest, indexer, env );
+                                PassVisitor<ConversionCost> converter( dest, indexer, env, conversionCost );
                                 src->accept( converter );
                                 return converter.get_cost();
+                                return converter.pass.get_cost();
                         } // if
                 } else {
 …
                         assert( diff == -1 && destAsRef );
                         PRINT( std::cerr << "dest is: " << dest << " / src is: " << src << std::endl; )
                         if ( typesCompatibleIgnoreQualifiers( src, destAsRef->get_base(), indexer, env ) ) {
+                        if ( typesCompatibleIgnoreQualifiers( src, destAsRef->base, indexer, env ) ) {
                                 PRINT( std::cerr << "converting compatible base type" << std::endl; )
                                 if ( src->get_lvalue() ) {
 …
+                                        )
                                         // lvalue-to-reference conversion:  cv lvalue T => cv T &
                                         if ( src->get_qualifiers() == destAsRef->get_base()->get_qualifiers() ) {
+                                        if ( src->get_qualifiers() == destAsRef->base->get_qualifiers() ) {
                                                 return Cost::reference; // cost needs to be non-zero to add cast
                                         } if ( src->get_qualifiers() < destAsRef->get_base()->get_qualifiers() ) {
+                                        } if ( src->get_qualifiers() < destAsRef->base->get_qualifiers() ) {
                                                 return Cost::safe; // cost needs to be higher than previous cast to differentiate adding qualifiers vs. keeping same
                                         } else {
                                                 return Cost::unsafe;
                                         } // if
                                 } else if ( destAsRef->get_base()->get_const() ) {
+                                } else if ( destAsRef->base->get_const() ) {
                                         PRINT( std::cerr << "rvalue to const ref conversion" << std::endl; )
                                         // rvalue-to-const-reference conversion: T => const T &
 …
+        }
         ConversionCost::ConversionCost( Type *dest, const SymTab::Indexer &indexer, const TypeEnvironment &env )
                 : dest( dest ), indexer( indexer ), cost( Cost::infinity ), env( env ) {
+        ConversionCost::ConversionCost( Type *dest, const SymTab::Indexer &indexer, const TypeEnvironment &env, CostFunction costFunc )
+                : dest( dest ), indexer( indexer ), cost( Cost::infinity ), env( env ), costFunc( costFunc ) {
+        }
 …
         };
         void ConversionCost::visit( __attribute((unused)) VoidType *voidType ) {
+        void ConversionCost::postvisit( VoidType * ) {
                 cost = Cost::infinity;
+        }
         void ConversionCost::visit(BasicType *basicType) {
+        void ConversionCost::postvisit(BasicType *basicType) {
                 if ( BasicType *destAsBasic = dynamic_cast< BasicType* >( dest ) ) {
                         int tableResult = costMatrix[ basicType->get_kind() ][ destAsBasic->get_kind() ];
 …
+        }
         void ConversionCost::visit( PointerType * pointerType ) {
+        void ConversionCost::postvisit( PointerType * pointerType ) {
                 if ( PointerType *destAsPtr = dynamic_cast< PointerType* >( dest ) ) {
+                        PRINT( std::cerr << pointerType << " ===> " << destAsPtr; )
+                        Type::Qualifiers tq1 = pointerType->get_base()->get_qualifiers();
+                        Type::Qualifiers tq2 = destAsPtr->get_base()->get_qualifiers();
+                        if ( tq1 <= tq2 && typesCompatibleIgnoreQualifiers( pointerType->get_base(), destAsPtr->get_base(), indexer, env ) ) {
+                        PRINT( std::cerr << pointerType << " ===> " << destAsPtr << std::endl; )
+                        Type::Qualifiers tq1 = pointerType->base->get_qualifiers();
+                        Type::Qualifiers tq2 = destAsPtr->base->get_qualifiers();
+                        if ( tq1 <= tq2 && typesCompatibleIgnoreQualifiers( pointerType->base, destAsPtr->base, indexer, env ) ) {
+                                PRINT( std::cerr << " :: compatible and good qualifiers" << std::endl; )
                                 if ( tq1 == tq2 ) {
                                         // types are the same
 …
                                 } else {
                                         // types are the same, except otherPointer has more qualifiers
-                                        PRINT( std::cerr << " :: compatible and good qualifiers" << std::endl; )
                                         cost = Cost::safe;
+                                }
                         } else {  // xxx - this discards qualifiers from consideration -- reducing qualifiers is a safe conversion; is this right?
+                        } else {
                                 int assignResult = ptrsAssignable( pointerType->base, destAsPtr->base, env );
                                 PRINT( std::cerr << " :: " << assignResult << std::endl; )
+                                if ( assignResult > 0 && pointerType->get_base()->get_qualifiers() <= destAsPtr->get_qualifiers() ) {
+                                        cost = Cost::safe;
+                                if ( assignResult > 0 && tq1 <= tq2 ) {
+                                        // xxx - want the case where qualifiers are added to be more expensive than the case where qualifiers are the same. Is 1 safe vs. 2 safe correct?
+                                        if ( tq1 == tq2 ) {
+                                                cost = Cost::safe;
+                                        } else if ( tq1 < tq2 ) {
+                                                cost = Cost::safe+Cost::safe;
+                                        }
                                 } else if ( assignResult < 0 ) {
                                         cost = Cost::unsafe;
 …
+        }
         void ConversionCost::visit( ArrayType * ) {}
         void ConversionCost::visit( ReferenceType * refType ) {
+        void ConversionCost::postvisit( ArrayType * ) {}
+        void ConversionCost::postvisit( ReferenceType * refType ) {
                 // Note: dest can never be a reference, since it would have been caught in an earlier check
                 assert( ! dynamic_cast< ReferenceType * >( dest ) );
 …
                 // recursively compute conversion cost from T1 to T2.
                 // cv can be safely dropped because of 'implicit dereference' behavior.
                 refType->base->accept( *this );
+                cost = costFunc( refType->base, dest, indexer, env );
                 if ( refType->base->get_qualifiers() == dest->get_qualifiers() ) {
                         cost.incReference();  // prefer exact qualifiers
 …
+        }
         void ConversionCost::visit( FunctionType * ) {}
         void ConversionCost::visit( StructInstType * inst ) {
+        void ConversionCost::postvisit( FunctionType * ) {}
+        void ConversionCost::postvisit( StructInstType * inst ) {
                 if ( StructInstType *destAsInst = dynamic_cast< StructInstType* >( dest ) ) {
                         if ( inst->name == destAsInst->name ) {
 …
+        }
         void ConversionCost::visit( UnionInstType * inst ) {
+        void ConversionCost::postvisit( UnionInstType * inst ) {
                 if ( UnionInstType *destAsInst = dynamic_cast< UnionInstType* >( dest ) ) {
                         if ( inst->name == destAsInst->name ) {
 …
+        }
         void ConversionCost::visit( EnumInstType * ) {
+        void ConversionCost::postvisit( EnumInstType * ) {
                 static Type::Qualifiers q;
                 static BasicType integer( q, BasicType::SignedInt );
                 integer.accept( *this );  // safe if dest >= int
+                cost = costFunc( &integer, dest, indexer, env );  // safe if dest >= int
                 if ( cost < Cost::unsafe ) {
                         cost.incSafe();
 …
+        }
         void ConversionCost::visit( TraitInstType * ) {}
         void ConversionCost::visit( TypeInstType *inst ) {
+        void ConversionCost::postvisit( TraitInstType * ) {}
+        void ConversionCost::postvisit( TypeInstType *inst ) {
                 EqvClass eqvClass;
                 NamedTypeDecl *namedType;
                 if ( env.lookup( inst->get_name(), eqvClass ) ) {
                         cost = conversionCost( eqvClass.type, dest, indexer, env );
+                if ( env.lookup( inst->name, eqvClass ) ) {
+                        cost = costFunc( eqvClass.type, dest, indexer, env );
                 } else if ( TypeInstType *destAsInst = dynamic_cast< TypeInstType* >( dest ) ) {
                         if ( inst->get_name() == destAsInst->get_name() ) {
+                        if ( inst->name == destAsInst->name ) {
                                 cost = Cost::zero;
+                        }
                 } else if ( ( namedType = indexer.lookupType( inst->get_name() ) ) ) {
+                } else if ( ( namedType = indexer.lookupType( inst->name ) ) ) {
                         TypeDecl *type = dynamic_cast< TypeDecl* >( namedType );
                         // all typedefs should be gone by this point
                         assert( type );
                         if ( type->get_base() ) {
                                 cost = conversionCost( type->get_base(), dest, indexer, env ) + Cost::safe;
                         } // if
                 } // if
+        }
         void ConversionCost::visit( TupleType * tupleType ) {
+                        if ( type->base ) {
+                                cost = costFunc( type->base, dest, indexer, env ) + Cost::safe;
+                        } // if
+                } // if
+        }
+        void ConversionCost::postvisit( TupleType * tupleType ) {
                 Cost c = Cost::zero;
                 if ( TupleType * destAsTuple = dynamic_cast< TupleType * >( dest ) ) {
                         std::list< Type * >::const_iterator srcIt = tupleType->get_types().begin();
                         std::list< Type * >::const_iterator destIt = destAsTuple->get_types().begin();
                         while ( srcIt != tupleType->get_types().end() && destIt != destAsTuple->get_types().end() ) {
                                 Cost newCost = conversionCost( *srcIt++, *destIt++, indexer, env );
+                        std::list< Type * >::const_iterator srcIt = tupleType->types.begin();
+                        std::list< Type * >::const_iterator destIt = destAsTuple->types.begin();
+                        while ( srcIt != tupleType->types.end() && destIt != destAsTuple->types.end() ) {
+                                Cost newCost = costFunc( *srcIt++, *destIt++, indexer, env );
                                 if ( newCost == Cost::infinity ) {
                                         return;
 …
                                 c += newCost;
                         } // while
                         if ( destIt != destAsTuple->get_types().end() ) {
+                        if ( destIt != destAsTuple->types.end() ) {
                                 cost = Cost::infinity;
                         } else {
 …
+        }
         void ConversionCost::visit( VarArgsType * ) {
+        void ConversionCost::postvisit( VarArgsType * ) {
                 if ( dynamic_cast< VarArgsType* >( dest ) ) {
                         cost = Cost::zero;
 …
+        }
         void ConversionCost::visit( ZeroType * ) {
+        void ConversionCost::postvisit( ZeroType * ) {
                 if ( dynamic_cast< ZeroType * >( dest ) ) {
                         cost = Cost::zero;
 …
+        }
         void ConversionCost::visit( OneType * ) {
+        void ConversionCost::postvisit( OneType * ) {
                 if ( dynamic_cast< OneType * >( dest ) ) {
                         cost = Cost::zero;

src/ResolvExpr/ConversionCost.h

-                      r326cd2b
+                      r9cdfb4d0
 #include "Cost.h"             // for Cost
+#include "Common/PassVisitor.h"
 #include "SynTree/Visitor.h"  // for Visitor
 #include "SynTree/SynTree.h"  // for Visitor Nodes
 …
         class TypeEnvironment;
+        class ConversionCost : public Visitor {
+        typedef std::function<Cost(Type *, Type *, const SymTab::Indexer &, const TypeEnvironment &)> CostFunction;
+        struct ConversionCost : public WithShortCircuiting {
           public:
                 ConversionCost( Type *dest, const SymTab::Indexer &indexer, const TypeEnvironment &env );
+                ConversionCost( Type *dest, const SymTab::Indexer &indexer, const TypeEnvironment &env, CostFunction );
                 Cost get_cost() const { return cost; }
+                virtual void visit(VoidType *voidType);
+                virtual void visit(BasicType *basicType);
+                virtual void visit(PointerType *pointerType);
+                virtual void visit(ArrayType *arrayType);
+                virtual void visit(ReferenceType *refType);
+                virtual void visit(FunctionType *functionType);
+                virtual void visit(StructInstType *aggregateUseType);
+                virtual void visit(UnionInstType *aggregateUseType);
+                virtual void visit(EnumInstType *aggregateUseType);
+                virtual void visit(TraitInstType *aggregateUseType);
+                virtual void visit(TypeInstType *aggregateUseType);
+                virtual void visit(TupleType *tupleType);
+                virtual void visit(VarArgsType *varArgsType);
+                virtual void visit(ZeroType *zeroType);
+                virtual void visit(OneType *oneType);
+                void previsit( BaseSyntaxNode * ) { visit_children = false; }
+                void postvisit( VoidType * voidType );
+                void postvisit( BasicType * basicType );
+                void postvisit( PointerType * pointerType );
+                void postvisit( ArrayType * arrayType );
+                void postvisit( ReferenceType * refType );
+                void postvisit( FunctionType * functionType );
+                void postvisit( StructInstType * aggregateUseType );
+                void postvisit( UnionInstType * aggregateUseType );
+                void postvisit( EnumInstType * aggregateUseType );
+                void postvisit( TraitInstType * aggregateUseType );
+                void postvisit( TypeInstType * aggregateUseType );
+                void postvisit( TupleType * tupleType );
+                void postvisit( VarArgsType * varArgsType );
+                void postvisit( ZeroType * zeroType );
+                void postvisit( OneType * oneType );
           protected:
                 Type *dest;
 …
                 Cost cost;
                 const TypeEnvironment &env;
+                CostFunction costFunc;
         };
         typedef std::function<int(Type *, Type *, const TypeEnvironment &, const SymTab::Indexer &)> PtrsFunction;
+        typedef std::function<int(Type *, Type *, const SymTab::Indexer &, const TypeEnvironment &)> PtrsFunction;
         Cost convertToReferenceCost( Type * src, ReferenceType * dest, const SymTab::Indexer & indexer, const TypeEnvironment & env, PtrsFunction func );
 } // namespace ResolvExpr

src/ResolvExpr/CurrentObject.cc

-                      r326cd2b
+                      r9cdfb4d0
 namespace ResolvExpr {
-        long long int getConstValue( ConstantExpr * constExpr ) {
-                if ( BasicType * basicType = dynamic_cast< BasicType * >( constExpr->get_result() ) ) {
-                        if ( basicType->isInteger() ) {
-                                return constExpr->get_constant()->get_ival();
-                        } else {
-                                assertf( false, "Non-integer constant expression in getConstValue %s", toString( constExpr ).c_str() ); // xxx - might be semantic error
+                        }
-                } else if ( dynamic_cast< OneType * >( constExpr->get_result() ) ) {
-                        return 1;
-                } else if ( dynamic_cast< ZeroType * >( constExpr->get_result() ) ) {
-                        return 0;
-                } else {
-                        assertf( false, "unhandled type on getConstValue %s", toString( constExpr->get_result() ).c_str() ); // xxx - might be semantic error
+                }
+        }
         template< typename AggrInst >
         TypeSubstitution makeGenericSubstitution( AggrInst * inst ) {
 …
                         base = at->get_base();
                         memberIter = createMemberIterator( base );
+                        if ( at->isVarLen ) throw SemanticError( "VLA initialization does not support @=", at );
                         setSize( at->get_dimension() );
+                }
 …
                 void setSize( Expression * expr ) {
                         if ( ConstantExpr * constExpr = dynamic_cast< ConstantExpr * >( expr ) ) {
+                                size = getConstValue( constExpr );
+                                PRINT( std::cerr << "array type with size: " << size << std::endl; )
+                                try {
+                                        size = constExpr->intValue();
+                                        PRINT( std::cerr << "array type with size: " << size << std::endl; )
+                                } catch ( SemanticError & ) {
+                                        throw SemanticError( "Constant expression of non-integral type in array dimension: ", expr );
+                                }
                         }       else if ( CastExpr * castExpr = dynamic_cast< CastExpr * >( expr ) ) {
                                 setSize( castExpr->get_arg() ); // xxx - need to perform the conversion specified by the cast
+                        } else if ( VariableExpr * varExpr = dynamic_cast< VariableExpr * >( expr ) ) {
+                                if ( EnumInstType * inst = dynamic_cast< EnumInstType * > ( varExpr->result ) ) {
+                                        long long int value;
+                                        if ( inst->baseEnum->valueOf( varExpr->var, value ) ) {
+                                                size = value;
+                                        }
+                                }
                         } else {
                                 assertf( false, "unhandled expression in setSize: %s", toString( expr ).c_str() ); // xxx - if not a constant expression, it's not simple to determine how long the array actually is, which is necessary for initialization to be done correctly -- fix this
 …
                         // need to permit integer-constant-expressions, including: integer constants, enumeration constants, character constants, sizeof expressions, _Alignof expressions, cast expressions
                         if ( ConstantExpr * constExpr = dynamic_cast< ConstantExpr * >( expr ) ) {
+                                index = getConstValue( constExpr );
+                                try {
+                                        index = constExpr->intValue();
+                                } catch( SemanticError & ) {
+                                        throw SemanticError( "Constant expression of non-integral type in array designator: ", expr );
+                                }
                         } else if ( CastExpr * castExpr = dynamic_cast< CastExpr * >( expr ) ) {
                                 setPosition( castExpr->get_arg() );
                         } else if ( VariableExpr * varExpr = dynamic_cast< VariableExpr * >( expr ) ) {
+                                assertf( dynamic_cast<EnumInstType *> ( varExpr->get_result() ), "ArrayIterator given variable that isn't an enum constant : %s", toString( expr ).c_str() );
+                                index = 0; // xxx - get actual value of enum constant
+                                EnumInstType * inst = dynamic_cast<EnumInstType *>( varExpr->get_result() );
+                                assertf( inst, "ArrayIterator given variable that isn't an enum constant : %s", toString( expr ).c_str() );
+                                long long int value;
+                                if ( inst->baseEnum->valueOf( varExpr->var, value ) ) {
+                                        index = value;
+                                }
                         } else if ( dynamic_cast< SizeofExpr * >( expr ) || dynamic_cast< AlignofExpr * >( expr ) ) {
                                 index = 0; // xxx - get actual sizeof/alignof value?

src/ResolvExpr/FindOpenVars.cc

-                      r326cd2b
+                      r9cdfb4d0
 #include <map>                    // for map<>::mapped_type
+#include "Common/PassVisitor.h"
 #include "SynTree/Declaration.h"  // for TypeDecl, DeclarationWithType (ptr ...
 #include "SynTree/Type.h"         // for Type, Type::ForallList, ArrayType
-#include "SynTree/Visitor.h"      // for Visitor
 namespace ResolvExpr {
+        class FindOpenVars : public Visitor {
+          public:
+        struct FindOpenVars : public WithGuards {
                 FindOpenVars( OpenVarSet &openVars, OpenVarSet &closedVars, AssertionSet &needAssertions, AssertionSet &haveAssertions, bool firstIsOpen );
+          private:
+                virtual void visit(PointerType *pointerType);
+                virtual void visit(ArrayType *arrayType);
+                virtual void visit(FunctionType *functionType);
+                virtual void visit(TupleType *tupleType);
+                void previsit( PointerType * pointerType );
+                void previsit( ArrayType * arrayType );
+                void previsit( FunctionType * functionType );
+                void previsit( TupleType * tupleType );
                 void common_action( Type *type );
 …
         void findOpenVars( Type *type, OpenVarSet &openVars, OpenVarSet &closedVars, AssertionSet &needAssertions, AssertionSet &haveAssertions, bool firstIsOpen ) {
                 FindOpenVars finder( openVars, closedVars, needAssertions, haveAssertions, firstIsOpen );
+                PassVisitor<FindOpenVars> finder( openVars, closedVars, needAssertions, haveAssertions, firstIsOpen );
                 type->accept( finder );
+        }
 …
                         } // for
                 } // if
 ///   std::cout << "type is ";
 ///   type->print( std::cout );
 ///   std::cout << std::endl << "need is" << std::endl;
 ///   printAssertionSet( needAssertions, std::cout );
 ///   std::cout << std::endl << "have is" << std::endl;
 ///   printAssertionSet( haveAssertions, std::cout );
+///   std::cerr << "type is ";
+///   type->print( std::cerr );
+///   std::cerr << std::endl << "need is" << std::endl;
+///   printAssertionSet( needAssertions, std::cerr );
+///   std::cerr << std::endl << "have is" << std::endl;
+///   printAssertionSet( haveAssertions, std::cerr );
+        }
         void FindOpenVars::visit(PointerType *pointerType) {
+        void FindOpenVars::previsit(PointerType *pointerType) {
                 common_action( pointerType );
-                Visitor::visit( pointerType );
+        }
         void FindOpenVars::visit(ArrayType *arrayType) {
+        void FindOpenVars::previsit(ArrayType *arrayType) {
                 common_action( arrayType );
-                Visitor::visit( arrayType );
+        }
         void FindOpenVars::visit(FunctionType *functionType) {
+        void FindOpenVars::previsit(FunctionType *functionType) {
                 common_action( functionType );
                 nextIsOpen = ! nextIsOpen;
+                Visitor::visit( functionType );
+                nextIsOpen = ! nextIsOpen;
+                GuardAction( [this](){ nextIsOpen = ! nextIsOpen; } );
+        }
         void FindOpenVars::visit(TupleType *tupleType) {
+        void FindOpenVars::previsit(TupleType *tupleType) {
                 common_action( tupleType );
-                Visitor::visit( tupleType );
+        }
 } // namespace ResolvExpr

src/ResolvExpr/Occurs.cc

-                      r326cd2b
+                      r9cdfb4d0
 // file "LICENCE" distributed with Cforall.
 //
 // Occurs.cc --
+// Occurs.cc --
 //
 // Author           : Richard C. Bilson
 …
 #include <string>             // for string
+#include "Common/PassVisitor.h"
 #include "SynTree/Type.h"     // for TypeInstType, Type
-#include "SynTree/Visitor.h"  // for Visitor
 #include "TypeEnvironment.h"  // for EqvClass, TypeEnvironment
 namespace ResolvExpr {
+        class Occurs : public Visitor {
+          public:
+        struct Occurs : public WithVisitorRef<Occurs> {
                 Occurs( std::string varName, const TypeEnvironment &env );
+                bool get_result() const { return result; }
+                virtual void visit( TypeInstType *typeInst );
+          private:
+                void previsit( TypeInstType * typeInst );
                 bool result;
                 std::set< std::string > eqvVars;
                 const TypeEnvironment &env;
+                const TypeEnvironment &tenv;
         };
         bool occurs( Type *type, std::string varName, const TypeEnvironment &env ) {
                 Occurs occur( varName, env );
+                PassVisitor<Occurs> occur( varName, env );
                 type->accept( occur );
                 return occur.get_result();
+                return occur.pass.result;
+        }
         Occurs::Occurs( std::string varName, const TypeEnvironment &env ) : result( false ), env( env ) {
+        Occurs::Occurs( std::string varName, const TypeEnvironment & env ) : result( false ), tenv( env ) {
                 EqvClass eqvClass;
                 if ( env.lookup( varName, eqvClass ) ) {
+                if ( tenv.lookup( varName, eqvClass ) ) {
                         eqvVars = eqvClass.vars;
                 } else {
 …
+        }
         void Occurs::visit( TypeInstType *typeInst ) {
+        void Occurs::previsit( TypeInstType * typeInst ) {
                 EqvClass eqvClass;
 ///   std::cout << "searching for vars: ";
 ///   std::copy( eqvVars.begin(), eqvVars.end(), std::ostream_iterator< std::string >( std::cout, " " ) );
 ///   std::cout << std::endl;
+///   std::cerr << "searching for vars: ";
+///   std::copy( eqvVars.begin(), eqvVars.end(), std::ostream_iterator< std::string >( std::cerr, " " ) );
+///   std::cerr << std::endl;
                 if ( eqvVars.find( typeInst->get_name() ) != eqvVars.end() ) {
                         result = true;
                 } else if ( env.lookup( typeInst->get_name(), eqvClass ) ) {
+                } else if ( tenv.lookup( typeInst->get_name(), eqvClass ) ) {
                         if ( eqvClass.type ) {
 ///       std::cout << typeInst->get_name() << " is bound to";
 ///       eqvClass.type->print( std::cout );
 ///       std::cout << std::endl;
                                 eqvClass.type->accept( *this );
+///       std::cerr << typeInst->get_name() << " is bound to";
+///       eqvClass.type->print( std::cerr );
+///       std::cerr << std::endl;
+                                eqvClass.type->accept( *visitor );
                         } // if
                 } // if

src/ResolvExpr/PolyCost.cc

-                      r326cd2b
+                      r9cdfb4d0
 //
+#include "Common/PassVisitor.h"
 #include "SymTab/Indexer.h"   // for Indexer
 #include "SynTree/Type.h"     // for TypeInstType, Type
-#include "SynTree/Visitor.h"  // for Visitor
 #include "TypeEnvironment.h"  // for EqvClass, TypeEnvironment
 namespace ResolvExpr {
+        class PolyCost : public Visitor {
+          public:
+        struct PolyCost {
                 PolyCost( const TypeEnvironment &env, const SymTab::Indexer &indexer );
+                int get_result() const { return result; }
+          private:
+                virtual void visit(TypeInstType *aggregateUseType);
+                void previsit( TypeInstType * aggregateUseType );
                 int result;
                 const TypeEnvironment &env;
+                const TypeEnvironment &tenv;
                 const SymTab::Indexer &indexer;
         };
         int polyCost( Type *type, const TypeEnvironment & env, const SymTab::Indexer &indexer ) {
                 PolyCost coster( env, indexer );
+                PassVisitor<PolyCost> coster( env, indexer );
                 type->accept( coster );
                 return coster.get_result();
+                return coster.pass.result;
+        }
         PolyCost::PolyCost( const TypeEnvironment & env, const SymTab::Indexer & indexer ) : result( 0 ), env( env ), indexer( indexer ) {
+        PolyCost::PolyCost( const TypeEnvironment & env, const SymTab::Indexer & indexer ) : result( 0 ), tenv( env ), indexer( indexer ) {
+        }
         void PolyCost::visit(TypeInstType * typeInst) {
+        void PolyCost::previsit(TypeInstType * typeInst) {
                 EqvClass eqvClass;
                 if ( env.lookup( typeInst->name, eqvClass ) ) {
+                if ( tenv.lookup( typeInst->name, eqvClass ) ) {
                         if ( eqvClass.type ) {
                                 if ( TypeInstType * otherTypeInst = dynamic_cast< TypeInstType* >( eqvClass.type ) ) {

src/ResolvExpr/PtrsAssignable.cc

r326cd2b	r9cdfb4d0
67	67	PtrsAssignable::PtrsAssignable( Type *dest, const TypeEnvironment &env ) : dest( dest ), result( 0 ), env( env ) {}
68	68
69		void PtrsAssignable::visit( ~~__attribute((unused)) VoidType *voidType~~ ) {
	69	void PtrsAssignable::visit( VoidType * ) {
70	70	// T * = void * is disallowed - this is a change from C, where any
71	71	// void * can be assigned or passed to a non-void pointer without a cast.

src/ResolvExpr/Resolver.cc

-                      r326cd2b
+                      r9cdfb4d0
                 } // if
                 #endif
                 assertf( finder.get_alternatives().size() == 1, "findSingleExpression: must have exactly one alternative at the end." );
+                assertf( finder.get_alternatives().size() == 1, "findSingleExpression: must have exactly one alternative at the end: (%zd) %s", finder.get_alternatives().size(), toString( untyped ).c_str() );
                 Alternative &choice = finder.get_alternatives().front();
                 Expression *newExpr = choice.expr->clone();
 …
         namespace {
+                /// resolve `untyped` to the expression whose type satisfies `pred` with the lowest cost; kindStr is used for providing better error messages
+                template<typename Pred>
+                void findKindExpression(Expression *& untyped, const SymTab::Indexer & indexer, const std::string & kindStr, Pred pred) {
+                        TypeEnvironment env;
+                        AlternativeFinder finder( indexer, env );
+                        finder.findWithAdjustment( untyped );
+                        AltList candidates;
+                        for ( Alternative & alt : finder.get_alternatives() ) {
+                                if ( pred( alt.expr->result ) ) {
+                                        candidates.push_back( std::move( alt ) );
+                                }
+                        }
+                        // choose the lowest cost expression among the candidates
+                        AltList winners;
+                        findMinCost( candidates.begin(), candidates.end(), back_inserter( winners ) );
+                        if ( winners.size() == 0 ) {
+                                throw SemanticError( "No reasonable alternatives for " + kindStr + " expression: ", untyped );
+                        } else if ( winners.size() != 1 ) {
+                                std::ostringstream stream;
+                                stream << "Cannot choose between " << winners.size() << " alternatives for " + kindStr +  " expression\n";
+                                untyped->print( stream );
+                                stream << "Alternatives are:\n";
+                                printAlts( winners, stream, 1 );
+                                throw SemanticError( stream.str() );
+                        }
+                        // there is one unambiguous interpretation - move the expression into the with statement
+                        Alternative & alt = winners.front();
+                        finishExpr( alt.expr, alt.env, untyped->env );
+                        delete untyped;
+                        untyped = alt.expr;
+                        alt.expr = nullptr;
+                }
                 bool isIntegralType( Type *type ) {
                         if ( dynamic_cast< EnumInstType * >( type ) ) {
 …
                 void findIntegralExpression( Expression *& untyped, const SymTab::Indexer &indexer ) {
+                        TypeEnvironment env;
+                        AlternativeFinder finder( indexer, env );
+                        finder.find( untyped );
+#if 0
+                        if ( finder.get_alternatives().size() != 1 ) {
+                                std::cout << "untyped expr is ";
+                                untyped->print( std::cout );
+                                std::cout << std::endl << "alternatives are:";
+                                for ( std::list< Alternative >::const_iterator i = finder.get_alternatives().begin(); i != finder.get_alternatives().end(); ++i ) {
+                                        i->print( std::cout );
+                                } // for
+                        } // if
+#endif
+                        Expression *newExpr = 0;
+                        const TypeEnvironment *newEnv = 0;
+                        for ( AltList::const_iterator i = finder.get_alternatives().begin(); i != finder.get_alternatives().end(); ++i ) {
+                                if ( i->expr->get_result()->size() == 1 && isIntegralType( i->expr->get_result() ) ) {
+                                        if ( newExpr ) {
+                                                throw SemanticError( "Too many interpretations for case control expression", untyped );
+                                        } else {
+                                                newExpr = i->expr->clone();
+                                                newEnv = &i->env;
+                                        } // if
+                                } // if
+                        } // for
+                        if ( ! newExpr ) {
+                                throw SemanticError( "No interpretations for case control expression", untyped );
+                        } // if
+                        finishExpr( newExpr, *newEnv, untyped->env );
+                        delete untyped;
+                        untyped = newExpr;
+                }
+                        findKindExpression( untyped, indexer, "condition", isIntegralType );
+                }
+        }
 …
         void Resolver::previsit( IfStmt *ifStmt ) {
                 findSingleExpression( ifStmt->condition, indexer );
+                findIntegralExpression( ifStmt->condition, indexer );
+        }
         void Resolver::previsit( WhileStmt *whileStmt ) {
                 findSingleExpression( whileStmt->condition, indexer );
+                findIntegralExpression( whileStmt->condition, indexer );
+        }
         void Resolver::previsit( ForStmt *forStmt ) {
                 if ( forStmt->condition ) {
                         findSingleExpression( forStmt->condition, indexer );
+                        findIntegralExpression( forStmt->condition, indexer );
                 } // if
 …
+        }
         void Resolver::previsit( WithStmt * withStmt ) {
                 for ( Expression *& expr : withStmt->exprs )  {
-                        TypeEnvironment env;
-                        AlternativeFinder finder( indexer, env );
-                        finder.findWithAdjustment( expr );
                         // only struct- and union-typed expressions are viable candidates
+                        AltList candidates;
+                        for ( Alternative & alt : finder.get_alternatives() ) {
+                                if ( isStructOrUnion( alt.expr->result ) ) {
+                                        candidates.push_back( std::move( alt ) );
+                                }
+                        }
+                        // choose the lowest cost expression among the candidates
+                        AltList winners;
+                        findMinCost( candidates.begin(), candidates.end(), back_inserter( winners ) );
+                        if ( winners.size() == 0 ) {
+                                throw SemanticError( "No reasonable alternatives for with statement expression: ", expr );
+                        } else if ( winners.size() != 1 ) {
+                                std::ostringstream stream;
+                                stream << "Cannot choose between " << winners.size() << " alternatives for with statement expression\n";
+                                expr->print( stream );
+                                stream << "Alternatives are:\n";
+                                printAlts( winners, stream, 1 );
+                                throw SemanticError( stream.str() );
+                        }
+                        // there is one unambiguous interpretation - move the expression into the with statement
+                        Alternative & alt = winners.front();
+                        finishExpr( alt.expr, alt.env, expr->env );
+                        delete expr;
+                        expr = alt.expr;
+                        alt.expr = nullptr;
+                        findKindExpression( expr, indexer, "with statement", isStructOrUnion );
                         // if with expression might be impure, create a temporary so that it is evaluated once

src/ResolvExpr/Resolver.h

r326cd2b	r9cdfb4d0
33	33	void findVoidExpression( Expression *& untyped, const SymTab::Indexer &indexer );
34	34	void findSingleExpression( Expression *& untyped, const SymTab::Indexer &indexer );
	35	void findSingleExpression( Expression & untyped, Type type, const SymTab::Indexer &indexer );
35	36	void resolveCtorInit( ConstructorInit * ctorInit, const SymTab::Indexer & indexer );
36	37	void resolveStmtExpr( StmtExpr * stmtExpr, const SymTab::Indexer & indexer );

src/SymTab/Indexer.cc

-                      r326cd2b
+                      r9cdfb4d0
+        }
+        void Indexer::addMembers( AggregateDecl * aggr, Expression * expr ) {
+                for ( Declaration * decl : aggr->members ) {
+                        if ( DeclarationWithType * dwt = dynamic_cast< DeclarationWithType * >( decl ) ) {
+                                addId( dwt, expr );
+                                if ( dwt->name == "" ) {
+                                        Type * t = dwt->get_type()->stripReferences();
+                                        if ( dynamic_cast< StructInstType * >( t ) || dynamic_cast< UnionInstType * >( t ) ) {
+                                                Expression * base = expr->clone();
+                                                ResolvExpr::referenceToRvalueConversion( base );
+                                                addMembers( t->getAggr(), new MemberExpr( dwt, base ) );
+                                        }
+                                }
+                        }
+                }
+        }
         void Indexer::addWith( WithStmt * stmt ) {
                 for ( Expression * expr : stmt->exprs ) {
 …
                                 assertf( aggr, "WithStmt expr has non-aggregate type: %s", toString( expr->result ).c_str() );
+                                for ( Declaration * decl : aggr->members ) {
+                                        if ( DeclarationWithType * dwt = dynamic_cast< DeclarationWithType * >( decl ) ) {
+                                                addId( dwt, expr );
+                                        }
+                                }
+                                addMembers( aggr, expr );
+                        }
+                }

src/SymTab/Indexer.h

-                      r326cd2b
+                      r9cdfb4d0
                 void addWith( WithStmt * );
+                /// adds all of the members of the Aggregate (addWith helper)
+                void addMembers( AggregateDecl * aggr, Expression * expr );
                 /// convenience function for adding a list of Ids to the indexer
                 void addIds( const std::list< DeclarationWithType * > & decls );

src/SymTab/Validate.cc

-                      r326cd2b
+                      r9cdfb4d0
                 template< typename AggDecl > void handleAggregate( AggDecl *aggregateDecl );
                 bool inStruct = false;
+                AggregateDecl * parentAggr = nullptr;
         };
 …
         template< typename AggDecl >
         void HoistStruct::handleAggregate( AggDecl *aggregateDecl ) {
                 if ( inStruct ) {
+                if ( parentAggr ) {
                         // Add elements in stack order corresponding to nesting structure.
                         declsToAddBefore.push_front( aggregateDecl );
                 } else {
                         GuardValue( inStruct );
                         inStruct = true;
+                        GuardValue( parentAggr );
+                        parentAggr = aggregateDecl;
                 } // if
                 // Always remove the hoisted aggregate from the inner structure.
 …
                                 throw SemanticError( "Cannot redefine typedef: " + tyDecl->name );
+                        }
+                        // cannot redefine VLA typedefs
+                        // Cannot redefine VLA typedefs. Note: this is slightly incorrect, because our notion of VLAs
+                        // at this point in the translator is imprecise. In particular, this will disallow redefining typedefs
+                        // with arrays whose dimension is an enumerator or a cast of a constant/enumerator. The effort required
+                        // to fix this corner case likely outweighs the utility of allowing it.
                         if ( isVariableLength( t1 ) || isVariableLength( t2 ) ) {
                                 throw SemanticError( "Cannot redefine typedef: " + tyDecl->name );

src/SynTree/AggregateDecl.cc

-                      r326cd2b
+                      r9cdfb4d0
 std::string TraitDecl::typeString() const { return "trait"; }
+namespace {
+        long long int getConstValue( Expression * expr ) {
+                if ( CastExpr * castExpr = dynamic_cast< CastExpr * > ( expr ) ) {
+                        return getConstValue( castExpr->arg );
+                } else if ( ConstantExpr * constExpr = dynamic_cast< ConstantExpr * >( expr ) ) {
+                        return constExpr->intValue();
+                // can be -1, +1, etc.
+                // } else if ( UntypedExpr * untypedExpr = dynamic_cast< UntypedExpr * >( expr ) ) {
+                //      if ( untypedExpr-> )
+                } else {
+                        assertf( false, "Unhandled expression type in getConstValue for enumerators: %s", toString( expr ).c_str() );
+                }
+        }
+}
+bool EnumDecl::valueOf( Declaration * enumerator, long long int & value ) {
+        if ( enumValues.empty() ) {
+                long long int currentValue = 0;
+                for ( Declaration * member : members ) {
+                        ObjectDecl * field = strict_dynamic_cast< ObjectDecl * >( member );
+                        if ( field->init ) {
+                                SingleInit * init = strict_dynamic_cast< SingleInit * >( field->init );
+                                currentValue = getConstValue( init->value );
+                        }
+                        assertf( enumValues.count( field->name ) == 0, "Enum %s has multiple members with the name %s", name.c_str(), field->name.c_str() );
+                        enumValues[ field->name ] = currentValue;
+                        ++currentValue;
+                }
+        }
+        if ( enumValues.count( enumerator->name ) ) {
+                value = enumValues[ enumerator->name ];
+                return true;
+        }
+        return false;
+}
 // Local Variables: //
 // tab-width: 4 //

src/SynTree/Declaration.h

-                      r326cd2b
+                      r9cdfb4d0
         EnumDecl( const EnumDecl &other ) : Parent( other ) {}
+        bool valueOf( Declaration * enumerator, long long int & value );
         virtual EnumDecl *clone() const override { return new EnumDecl( *this ); }
         virtual void accept( Visitor &v ) override { v.visit( this ); }
         virtual Declaration *acceptMutator( Mutator &m )  override { return m.mutate( this ); }
   private:
+        std::map< std::string, long long int > enumValues;
         virtual std::string typeString() const override;
 };

src/SynTree/Expression.cc

-                      r326cd2b
+                      r9cdfb4d0
+}
+long long int ConstantExpr::intValue() const {
+        if ( BasicType * basicType = dynamic_cast< BasicType * >( result ) ) {
+                if ( basicType->isInteger() ) {
+                        return get_constant()->get_ival();
+                }
+        } else if ( dynamic_cast< OneType * >( result ) ) {
+                return 1;
+        } else if ( dynamic_cast< ZeroType * >( result ) ) {
+                return 0;
+        }
+        throw SemanticError( "Constant expression of non-integral type ", this );
+}
 VariableExpr::VariableExpr( DeclarationWithType *_var ) : Expression(), var( _var ) {
         assert( var );
 …
         //      assert( inst->baseEnum );
         //      EnumDecl * decl = inst->baseEnum;
+        //      for ( Declaration * member : decl->members ) {
+        //              if ( member == _var ) {
+        //                      type->set_lvalue( false );
+        //              }
+        //      long long int value;
+        //      if ( decl->valueOf( var, value ) ) {
+        //              type->set_lvalue( false );
         //      }
         // }
 …
                 } else {
                         // references have been removed, in which case dereference returns an lvalue of the base type.
                         ret->get_result()->set_lvalue( true );
+                        ret->result->set_lvalue( true );
+                }
+        }
 …
         if ( ! body.empty() ) {
                 if ( ExprStmt * exprStmt = dynamic_cast< ExprStmt * >( body.back() ) ) {
                         set_result( maybeClone( exprStmt->get_expr()->get_result() ) );
+                        result = maybeClone( exprStmt->expr->result );
+                }
+        }
         // ensure that StmtExpr has a result type
         if ( ! result ) {
                 set_result( new VoidType( Type::Qualifiers() ) );
+                result = new VoidType( Type::Qualifiers() );
+        }
+}

src/SynTree/Expression.h

-                      r326cd2b
+                      r9cdfb4d0
         Constant * get_constant() { return & constant; }
+        const Constant * get_constant() const { return & constant; }
         void set_constant( const Constant & newValue ) { constant = newValue; }
+        long long int intValue() const;
         virtual ConstantExpr * clone() const { return new ConstantExpr( * this ); }

src/SynTree/Label.h

r326cd2b	r9cdfb4d0
33	33	std::list< Attribute * >& get_attributes() { return attributes; }
34	34
35		operator std::string() { return name; }
	35	operator std::string() const { return name; }
36	36	bool empty() { return name.empty(); }
37	37	private:

src/SynTree/Type.h

r326cd2b	r9cdfb4d0
356	356	bool isTtype() const;
357	357
	358	bool isUnprototyped() const { return isVarArgs && parameters.size() == 0; }
	359
358	360	virtual FunctionType clone() const override { return new FunctionType( this ); }
359	361	virtual void accept( Visitor & v ) override { v.visit( this ); }

src/SynTree/TypeSubstitution.cc

-                      r326cd2b
+                      r9cdfb4d0
 void TypeSubstitution::normalize() {
+        PassVisitor<Substituter> sub( *this, true );
         do {
                 subCount = 0;
                 freeOnly = true;
+                sub.pass.subCount = 0;
+                sub.pass.freeOnly = true;
                 for ( TypeEnvType::iterator i = typeEnv.begin(); i != typeEnv.end(); ++i ) {
                         i->second = i->second->acceptMutator( *this );
+                        i->second = i->second->acceptMutator( sub );
+                }
         } while ( subCount );
+}
 Type * TypeSubstitution::mutate( TypeInstType *inst ) {
         BoundVarsType::const_iterator bound = boundVars.find( inst->get_name() );
+        } while ( sub.pass.subCount );
+}
+Type * TypeSubstitution::Substituter::postmutate( TypeInstType *inst ) {
+        BoundVarsType::const_iterator bound = boundVars.find( inst->name );
         if ( bound != boundVars.end() ) return inst;
         TypeEnvType::const_iterator i = typeEnv.find( inst->get_name() );
         if ( i == typeEnv.end() ) {
+        TypeEnvType::const_iterator i = sub.typeEnv.find( inst->get_name() );
+        if ( i == sub.typeEnv.end() ) {
                 return inst;
         } else {
 ///         std::cout << "found " << inst->get_name() << ", replacing with ";
 ///         i->second->print( std::cout );
 ///         std::cout << std::endl;
+///         std::cerr << "found " << inst->get_name() << ", replacing with ";
+///         i->second->print( std::cerr );
+///         std::cerr << std::endl;
                 subCount++;
                 Type *newtype = i->second->clone();
+                Type * newtype = i->second->clone();
                 newtype->get_qualifiers() |= inst->get_qualifiers();
                 delete inst;
 …
+}
 Expression * TypeSubstitution::mutate( NameExpr *nameExpr ) {
         VarEnvType::const_iterator i = varEnv.find( nameExpr->get_name() );
         if ( i == varEnv.end() ) {
+Expression * TypeSubstitution::Substituter::postmutate( NameExpr * nameExpr ) {
+        VarEnvType::const_iterator i = sub.varEnv.find( nameExpr->name );
+        if ( i == sub.varEnv.end() ) {
                 return nameExpr;
         } else {
 …
+}
+template< typename TypeClass >
+Type *TypeSubstitution::handleType( TypeClass *type ) {
+        ValueGuard<BoundVarsType> oldBoundVars( boundVars );
+void TypeSubstitution::Substituter::premutate( Type * type ) {
+        GuardValue( boundVars );
         // bind type variables from forall-qualifiers
         if ( freeOnly ) {
                 for ( Type::ForallList::const_iterator tyvar = type->get_forall().begin(); tyvar != type->get_forall().end(); ++tyvar ) {
                         boundVars.insert( (*tyvar )->get_name() );
+                for ( Type::ForallList::const_iterator tyvar = type->forall.begin(); tyvar != type->forall.end(); ++tyvar ) {
+                        boundVars.insert( (*tyvar)->name );
                 } // for
         } // if
-        Type *ret = Mutator::mutate( type );
-        return ret;
+}
 template< typename TypeClass >
 Type *TypeSubstitution::handleAggregateType( TypeClass *type ) {
         ValueGuard<BoundVarsType> oldBoundVars( boundVars );
+void TypeSubstitution::Substituter::handleAggregateType( TypeClass * type ) {
+        GuardValue( boundVars );
         // bind type variables from forall-qualifiers
         if ( freeOnly ) {
                 for ( Type::ForallList::const_iterator tyvar = type->get_forall().begin(); tyvar != type->get_forall().end(); ++tyvar ) {
                         boundVars.insert( (*tyvar )->get_name() );
+                for ( Type::ForallList::const_iterator tyvar = type->forall.begin(); tyvar != type->forall.end(); ++tyvar ) {
+                        boundVars.insert( (*tyvar)->name );
                 } // for
                 // bind type variables from generic type instantiations
                 std::list< TypeDecl* > *baseParameters = type->get_baseParameters();
                 if ( baseParameters && ! type->get_parameters().empty() ) {
+                if ( baseParameters && ! type->parameters.empty() ) {
                         for ( std::list< TypeDecl* >::const_iterator tyvar = baseParameters->begin(); tyvar != baseParameters->end(); ++tyvar ) {
                                 boundVars.insert( (*tyvar)->get_name() );
+                                boundVars.insert( (*tyvar)->name );
                         } // for
                 } // if
         } // if
+        Type *ret = Mutator::mutate( type );
+        return ret;
+}
+Type * TypeSubstitution::mutate( VoidType *voidType ) {
+        return handleType( voidType );
+}
+Type * TypeSubstitution::mutate( BasicType *basicType ) {
+        return handleType( basicType );
+}
+Type * TypeSubstitution::mutate( PointerType *pointerType ) {
+        return handleType( pointerType );
+}
+Type * TypeSubstitution::mutate( ArrayType *arrayType ) {
+        return handleType( arrayType );
+}
+Type * TypeSubstitution::mutate( FunctionType *functionType ) {
+        return handleType( functionType );
+}
+Type * TypeSubstitution::mutate( StructInstType *aggregateUseType ) {
+        return handleAggregateType( aggregateUseType );
+}
+Type * TypeSubstitution::mutate( UnionInstType *aggregateUseType ) {
+        return handleAggregateType( aggregateUseType );
+}
+Type * TypeSubstitution::mutate( EnumInstType *aggregateUseType ) {
+        return handleType( aggregateUseType );
+}
+Type * TypeSubstitution::mutate( TraitInstType *aggregateUseType ) {
+        return handleType( aggregateUseType );
+}
+Type * TypeSubstitution::mutate( TupleType *tupleType ) {
+        return handleType( tupleType );
+}
+Type * TypeSubstitution::mutate( VarArgsType *varArgsType ) {
+        return handleType( varArgsType );
+}
+Type * TypeSubstitution::mutate( ZeroType *zeroType ) {
+        return handleType( zeroType );
+}
+Type * TypeSubstitution::mutate( OneType *oneType ) {
+        return handleType( oneType );
+}
+void TypeSubstitution::Substituter::premutate( StructInstType * aggregateUseType ) {
+        handleAggregateType( aggregateUseType );
+}
+void TypeSubstitution::Substituter::premutate( UnionInstType *aggregateUseType ) {
+        handleAggregateType( aggregateUseType );
+}

src/SynTree/TypeSubstitution.h

-                      r326cd2b
+                      r9cdfb4d0
 #include "SynTree/Declaration.h"   // for TypeDecl, Declaration (ptr only)
 #include "SynTree/Expression.h"    // for Expression (ptr only), NameExpr (p...
-#include "SynTree/Mutator.h"       // for Mutator
 #include "SynTree/Type.h"          // for Type, ArrayType (ptr only), BasicT...
+class TypeSubstitution : public Mutator {
+        typedef Mutator Parent;
+class TypeSubstitution {
   public:
         TypeSubstitution();
 …
         TypeSubstitution *clone() const { return new TypeSubstitution( *this ); }
   private:
-        virtual Type* mutate(TypeInstType *aggregateUseType);
-        virtual Expression* mutate(NameExpr *nameExpr);
+        /// Records type variable bindings from forall-statements
+        template< typename TypeClass > Type *handleType( TypeClass *type );
+        /// Records type variable bindings from forall-statements and instantiations of generic types
+        template< typename TypeClass > Type *handleAggregateType( TypeClass *type );
+        virtual Type* mutate(VoidType *basicType);
+        virtual Type* mutate(BasicType *basicType);
+        virtual Type* mutate(PointerType *pointerType);
+        virtual Type* mutate(ArrayType *arrayType);
+        virtual Type* mutate(FunctionType *functionType);
+        virtual Type* mutate(StructInstType *aggregateUseType);

Context Navigation

Legend: