Changeset 7b91c0e

doc/bibliography/pl.bib

-                      r5869cea
+                      r7b91c0e
     title       = {Asynchronous Exception Propagation in Blocked Tasks},
     booktitle   = {4th International Workshop on Exception Handling (WEH.08)},
     organization= {16th International Symposium on the Foundations of Software Engineering (FSE 16)},
+    optorganization= {16th International Symposium on the Foundations of Software Engineering (FSE 16)},
     address     = {Atlanta, U.S.A},
     month       = nov,
 …
 @inproceedings{Edelson92,
     keywords    = {persistence, pointers},
+    keywords    = {persistence, smart pointers},
     contributer = {pabuhr@plg},
     author      = {Daniel R. Edelson},
 …
     year        = 1992,
     pages       = {1-19},
+}
+@incollection{smartpointers,
+    keywords    = {smart pointers},
+    contributer = {pabuhr@plg},
+    author      = {Andrei Alexandrescu},
+    title       = {Smart Pointers},
+    booktitle   = {Modern C++ Design: Generic Programming and Design Patterns Applied},
+    publisher   = {Addison-Wesley},
+    year        = 2001,
+    chapter     = 7,
+    optpages    = {?-?},
+}
 …
+}
+@misc{vistorpattern,
+    keywords    = {visitor pattern},
+    contributer = {pabuhr@plg},
+    key         = {vistor pattern},
+    title       = {vistor pattern},
+    year        = 2020,
+    note        = {WikipediA},
+    howpublished= {\href{https://en.wikipedia.org/wiki/Visitor\_pattern}
+                  {https://\-en.wikipedia.org/\-wiki/\-Visitor\_pattern}},
+}
 % W

doc/theses/andrew_beach_MMath/Makefile

-                      r5869cea
+                      r7b91c0e
 ### Makefile for Andrew Beach's Masters Thesis
 DOC=thesis.pdf
+DOC=uw-ethesis.pdf
 BUILD=out
 TEXSRC=$(wildcard *.tex)
 …
 STYSRC=$(wildcard *.sty)
 CLSSRC=$(wildcard *.cls)
 TEXLIB= .:${BUILD}:
+TEXLIB= .:../../LaTeXmacros:${BUILD}:
 BIBLIB= .:../../bibliography
 …
         ${LATEX} ${BASE}
         ${BIBTEX} ${BUILD}/${BASE}
+        ${LATEX} ${BASE}
         ${GLOSSARY} ${BUILD}/${BASE}
         ${LATEX} ${BASE}

doc/theses/andrew_beach_MMath/existing.tex

-                      r5869cea
+                      r7b91c0e
+\chapter{\CFA{} Existing Features}
+\chapter{\CFA Existing Features}
+\CFA (C-for-all)~\cite{Cforall} is an open-source project extending ISO C with modern safety and productivity features, while still ensuring backwards compatibility with C and its programmers.
+\CFA is designed to have an orthogonal feature-set based closely on the C programming paradigm (non-object-oriented) and these features can be added incrementally to an existing C code-base allowing programmers to learn \CFA on an as-needed basis.
 \section{Overloading and extern}
 Cforall has overloading, allowing multiple definitions of the same name to
 be defined.
+be defined.~\cite{Moss18}
 This also adds name mangling so that the assembly symbols are unique for
 …
 The syntax for disabling mangling is:
 \begin{lstlisting}
+\begin{cfa}
 extern "C" {
     ...
+}
 \end{lstlisting}
+\end{cfa}
 To re-enable mangling once it is disabled the syntax is:
 \begin{lstlisting}
+\begin{cfa}
 extern "Cforall" {
     ...
+}
 \end{lstlisting}
+\end{cfa}
 Both should occur at the declaration level and effect all the declarations
 in \texttt{...}. Neither care about the state of mangling when they begin
+in @...@. Neither care about the state of mangling when they begin
 and will return to that state after the group is finished. So re-enabling
 is only used to nest areas of mangled and unmangled declarations.
 …
 \section{References}
 \CFA adds references to C. These are auto-dereferencing pointers and use the
 same syntax as pointers except they use ampersand (\codeCFA{\&}) instead of
 the asterisk (\codeCFA{*}). They can also be constaint or mutable, if they
+same syntax as pointers except they use ampersand (@&@) instead of
+the asterisk (@*@). They can also be constaint or mutable, if they
 are mutable they may be assigned to by using the address-of operator
 (\codeCFA\&) which converts them into a pointer.
+(@&@) which converts them into a pointer.
 \section{Constructors and Destructors}
 …
 functions with special names. The special names are used to define them and
 may be used to call the functions expicately. The \CFA special names are
 constructed by taking the tokens in the operators and putting \texttt{?} where
 the arguments would go. So multiplication is \texttt{?*?} while dereference
 is \texttt{*?}. This also make it easy to tell the difference between
 pre-fix operations (such as \texttt{++?}) and post-fix operations
 (\texttt{?++}).
 The special name for contructors is \texttt{?\{\}}, which comes from the
+constructed by taking the tokens in the operators and putting @?@ where
+the arguments would go. So multiplication is @?*?@ while dereference
+is @*?@. This also make it easy to tell the difference between
+pre-fix operations (such as @++?@) and post-fix operations
+(@?++@).
+The special name for contructors is @?{}@, which comes from the
 initialization syntax in C. The special name for destructors is
 \texttt{\^{}?\{\}}. % I don't like the \^{} symbol but $^\wedge$ isn't better.
+@^{}@. % I don't like the \^{} symbol but $^\wedge$ isn't better.
 Any time a type T goes out of scope the destructor matching
 \codeCFA{void ^?\{\}(T \&);} is called. In theory this is also true of
 primitive types such as \codeCFA{int}, but in practice those are no-ops and
+@void ^?{}(T &);@ is called. In theory this is also true of
+primitive types such as @int@, but in practice those are no-ops and
 are usually omitted for optimization.
 \section{Polymorphism}
 \CFA uses polymorphism to create functions and types that are defined over
 different types. \CFA polymorphic declarations serve the same role as \CPP
+different types. \CFA polymorphic declarations serve the same role as \CC
 templates or Java generics.
 …
 except that you may use the names introduced by the forall clause in them.
 Forall clauses are written \codeCFA{forall( ... )} where \codeCFA{...} becomes
+Forall clauses are written @forall( ... )@ where @...@ becomes
 the list of polymorphic variables (local type names) and assertions, which
 repersent required operations on those types.
 \begin{lstlisting}
+\begin{cfa}
 forall(dtype T | { void do_once(T &); })
 void do_twice(T & value) {
 …
     do_once(value);
+}
 \end{lstlisting}
+\end{cfa}
 A polymorphic function can be used in the same way normal functions are.
 …
 the the call site.
 As an example, even if no function named \codeCFA{do_once} is not defined
 near the definition of \codeCFA{do_twice} the following code will work.
 \begin{lstlisting}
+As an example, even if no function named @do_once@ is not defined
+near the definition of @do_twice@ the following code will work.
+\begin{cfa}
 int quadruple(int x) {
     void do_once(int & y) {
 …
     return x;
+}
 \end{lstlisting}
+\end{cfa}
 This is not the recommended way to implement a quadruple function but it
 does work. The complier will deduce that \codeCFA{do_twice}'s T is an
+does work. The complier will deduce that @do_twice@'s T is an
 integer from the argument. It will then look for a definition matching the
 assertion which is the \codeCFA{do_once} defined within the function. That
 function will be passed in as a function pointer to \codeCFA{do_twice} and
+assertion which is the @do_once@ defined within the function. That
+function will be passed in as a function pointer to @do_twice@ and
 called within it.
 …
 traits which collect assertions into convenent packages that can then be used
 in assertion lists instead of all of their components.
 \begin{lstlisting}
+\begin{cfa}
 trait done_once(dtype T) {
     void do_once(T &);
+}
 \end{lstlisting}
+\end{cfa}
 After this the forall list in the previous example could instead be written
 with the trait instead of the assertion itself.
 \begin{lstlisting}
+\begin{cfa}
 forall(dtype T | done_once(T))
 \end{lstlisting}
+\end{cfa}
 Traits can have arbitrary number of assertions in them and are usually used to
 …
 are now used in field declaractions instead of parameters and local variables.
 \begin{lstlisting}
+\begin{cfa}
 forall(dtype T)
 struct node {
 …
     T * data;
+}
 \end{lstlisting}
 The \codeCFA{node(T)} is a use of a polymorphic structure. Polymorphic types
+\end{cfa}
+The @node(T)@ is a use of a polymorphic structure. Polymorphic types
 must be provided their polymorphic parameters.
 …
 \section{Concurrency}
 \CFA has a number of concurrency features, \codeCFA{thread}s,
 \codeCFA{monitor}s and \codeCFA{mutex} parameters, \codeCFA{coroutine}s and
 \codeCFA{generator}s. The two features that interact with the exception system
 are \codeCFA{thread}s and \codeCFA{coroutine}s; they and their supporting
+\CFA has a number of concurrency features, @thread@s,
+@monitor@s and @mutex@ parameters, @coroutine@s and
+@generator@s. The two features that interact with the exception system
+are @thread@s and @coroutine@s; they and their supporting
 constructs will be described here.
 …
 library.
 In \CFA coroutines are created using the \codeCFA{coroutine} keyword which
 works just like \codeCFA{struct} except that the created structure will be
 modified by the compiler to satify the \codeCFA{is_coroutine} trait.
+In \CFA coroutines are created using the @coroutine@ keyword which
+works just like @struct@ except that the created structure will be
+modified by the compiler to satify the @is_coroutine@ trait.
 These structures act as the interface between callers and the coroutine,
 the fields are used to pass information in and out. Here is a simple example
 where the single field is used to pass the next number in a sequence out.
 \begin{lstlisting}
+\begin{cfa}
 coroutine CountUp {
     unsigned int next;
+}
 \end{lstlisting}
+\end{cfa}
 The routine part of the coroutine is a main function for the coroutine. It
 …
 function it continue from that same suspend statement instead of at the top
 of the function.
 \begin{lstlisting}
+\begin{cfa}
 void main(CountUp & this) {
     unsigned int next = 0;
 …
+    }
+}
 \end{lstlisting}
+\end{cfa}
 Control is passed to the coroutine with the resume function. This includes the
 …
 return value is for easy access to communication variables. For example the
 next value from a count-up can be generated and collected in a single
 expression: \codeCFA{resume(count).next}.
+expression: @resume(count).next@.
 \subsection{Monitors and Mutex}
 …
 parameters.
 Function parameters can have the \codeCFA{mutex} qualifiers on reference
 arguments, for example \codeCFA{void example(a_monitor & mutex arg);}. When the
+Function parameters can have the @mutex@ qualifiers on reference
+arguments, for example @void example(a_monitor & mutex arg);@. When the
 function is called it will acquire the lock on all of the mutex parameters.
 …
 Threads are created like coroutines except the keyword is changed:
 \begin{lstlisting}
+\begin{cfa}
 thread StringWorker {
     const char * input;
 …
     this.result = result;
+}
 \end{lstlisting}
+\end{cfa}
 The main function will start executing after the fork operation and continue
 executing until it is finished. If another thread joins with this one it will
 …
 From the outside this is the creation and destruction of the thread object.
 Fork happens after the constructor is run and join happens before the
 destructor runs. Join also happens during the \codeCFA{join} function which
+destructor runs. Join also happens during the @join@ function which
 can be used to join a thread earlier. If it is used the destructor does not
 join as that has already been completed.

doc/theses/andrew_beach_MMath/features.tex

-                      r5869cea
+                      r7b91c0e
 returns a reference to the virtual table instance. Defining this function
 also establishes the virtual type and virtual table pair to the resolver
 and promises that \codeCFA{exceptT} is a virtual type and a child of the
+and promises that @exceptT@ is a virtual type and a child of the
 base exception type.
 One odd thing about \codeCFA{get_exception_vtable} is that it should always
+One odd thing about @get_exception_vtable@ is that it should always
 be a constant function, returning the same value regardless of its argument.
 A pointer or reference to the virtual table instance could be used instead,
 …
 Also note the use of the word ``promise" in the trait description. \CFA
 cannot currently check to see if either \codeCFA{exceptT} or
 \codeCFA{virtualT} match the layout requirements. Currently this is
 considered part of \codeCFA{get_exception_vtable}'s correct implementation.
+cannot currently check to see if either @exceptT@ or
+@virtualT@ match the layout requirements. Currently this is
+considered part of @get_exception_vtable@'s correct implementation.
 \begin{lstlisting}
 …
 Finally there are three additional macros that can be used to refer to the
 these traits. They are \codeCFA{IS_EXCEPTION},
 \codeCFA{IS_TERMINATION_EXCEPTION} and \codeCFA{IS_RESUMPTION_EXCEPTION}.
+these traits. They are @IS_EXCEPTION@,
+@IS_TERMINATION_EXCEPTION@ and @IS_RESUMPTION_EXCEPTION@.
 Each takes the virtual type's name and, for polymorphic types only, the
 parenthesized list of polymorphic arguments. These do the name mangling to
 …
 The expression must evaluate to a reference to a termination exception. A
 termination exception is any exception with a
 \codeCFA{void defaultTerminationHandler(T &);} (the default handler) defined
+@void defaultTerminationHandler(T &);@ (the default handler) defined
 on it. The handler is taken from the call sight with \CFA's trait system and
 passed into the exception system along with the exception itself.
 …
 You can also re-throw the most recent termination exception with
 \codeCFA{throw;}. % This is terrible and you should never do it.
+@throw;@. % This is terrible and you should never do it.
 This can be done in a handler or any function that could be called from a
 handler.
 …
 The result of EXPRESSION must be a resumption exception type. A resumption
 exception type is any type that satisfies the assertion
 \codeCFA{void defaultResumptionHandler(T &);} (the default handler). When the
+@void defaultResumptionHandler(T &);@ (the default handler). When the
 statement is executed the expression is evaluated and the result is thrown.
 …
 \paragraph{Re-Throwing}
 You may also re-throw resumptions with a \codeCFA{throwResume;} statement.
 This can only be done from inside of a \codeCFA{catchResume} block.
+You may also re-throw resumptions with a @throwResume;@ statement.
+This can only be done from inside of a @catchResume@ block.
 Outside of any side effects of any code already run in the handler this will
 …
 \section{Finally Clauses}
 A \codeCFA{finally} clause may be placed at the end of a try statement after
+A @finally@ clause may be placed at the end of a try statement after
 all the handler clauses. In the simply case, with no handlers, it looks like
 this:
 …
 Because of this local control flow out of the finally block is forbidden.
 The compiler rejects uses of \codeCFA{break}, \codeCFA{continue},
 \codeCFA{fallthru} and \codeCFA{return} that would cause control to leave
+The compiler rejects uses of @break@, @continue@,
+@fallthru@ and @return@ that would cause control to leave
 the finally block. Other ways to leave the finally block - such as a long
 jump or termination - are much harder to check, at best requiring additional
 …
 There is no special statement for starting a cancellation, instead you call
 the standard library function \codeCFA{cancel\_stack} which takes an exception.
+the standard library function @cancel\_stack@ which takes an exception.
 Unlike in a throw this exception is not used in control flow but is just there
 to pass information about why the cancellation happened.
 …
 \item Thread Stack:
 Thread stacks are those created \codeCFA{thread} or otherwise satisfy the
 \codeCFA{is\_thread} trait.
+Thread stacks are those created @thread@ or otherwise satisfy the
+@is\_thread@ trait.
 Threads only have two structural points of communication that must happen,
 …
 and wait for another thread to join with it. The other thread, when it joins,
 checks for a cancellation. If so it will throw the resumption exception
 \codeCFA{ThreadCancelled}.
 There is a difference here in how explicate joins (with the \codeCFA{join}
+@ThreadCancelled@.
+There is a difference here in how explicate joins (with the @join@
 function) and implicate joins (from a destructor call). Explicate joins will
 take the default handler (\codeCFA{defaultResumptionHandler}) from the context
+take the default handler (@defaultResumptionHandler@) from the context
 and use like a regular through does if the exception is not caught. The
 implicate join does a program abort instead.
 …
 \item Coroutine Stack:
 Coroutine stacks are those created with \codeCFA{coroutine} or otherwise
 satisfy the \codeCFA{is\_coroutine} trait.
+Coroutine stacks are those created with @coroutine@ or otherwise
+satisfy the @is\_coroutine@ trait.
 A coroutine knows of two other coroutines, its starter and its last resumer.
 …
 After the stack is unwound control goes to the last resumer.
 Resume will resume throw a \codeCFA{CoroutineCancelled} exception, which is
+Resume will resume throw a @CoroutineCancelled@ exception, which is
 polymorphic over the coroutine type and has a pointer to the coroutine being
 canceled and the canceling exception. The resume function also has an
 assertion that the \codeCFA{defaultResumptionHandler} for the exception. So it
+assertion that the @defaultResumptionHandler@ for the exception. So it
 will use the default handler like a regular throw.

doc/theses/andrew_beach_MMath/future.tex

-                      r5869cea
+                      r7b91c0e
 \section{Additional Throws}
 Several other kinds of throws, beyond the termination throw (\codeCFA{throw}),
 the resumption throw (\codeCFA{throwResume}) and the re-throws, were considered.
+Several other kinds of throws, beyond the termination throw (@throw@),
+the resumption throw (@throwResume@) and the re-throws, were considered.
 None were as useful as the core throws but they would likely be worth
 revising.
 …
 is no reason not to allow it. It is however a small improvement; giving a bit
 of flexibility to the user in what style they want to use.
 \item Enabling local control flow (by \codeCFA{break}, \codeCFA{return} and
+\item Enabling local control flow (by @break@, @return@ and
 similar statements) out of a termination handler. The current set-up makes
 this very difficult but the catch function that runs the handler after it has

doc/theses/andrew_beach_MMath/implement.tex

-                      r5869cea
+                      r7b91c0e
 All of this is accessed through a field inserted at the beginning of every
 virtual type. Currently it is called \codeC{virtual_table} but it is not
+virtual type. Currently it is called @virtual_table@ but it is not
 ment to be accessed by the user. This field is a pointer to the type's
 virtual table instance. It is assigned once during the object's construction
 …
 using that to calculate the mangled name of the parent's virtual table type.
 There are two special fields that are included like normal fields but have
 special initialization rules: the \codeC{size} field is the type's size and is
 initialized with a sizeof expression, the \codeC{align} field is the type's
+special initialization rules: the @size@ field is the type's size and is
+initialized with a sizeof expression, the @align@ field is the type's
 alignment and uses an alignof expression. The remaining fields are resolved
 to a name matching the field's name and type using the normal visibility
 …
 The declarations include the virtual type definition and forward declarations
 of the virtual table instance, constructor, message function and
 \codeCFA{get_exception_vtable}. The definition includes the storage and
+@get_exception_vtable@. The definition includes the storage and
 initialization of the virtual table instance and the bodies of the three
 functions.
 …
 from the per-instance information. The virtual table type and most of the
 functions are polymorphic so they are all part of the core. The virtual table
 instance and the \codeCFA{get_exception_vtable} function.
 Coroutines and threads need instances of \codeCFA{CoroutineCancelled} and
 \codeCFA{ThreadCancelled} respectively to use all of their functionality.
 When a new data type is declared with \codeCFA{coroutine} or \codeCFA{thread}
+instance and the @get_exception_vtable@ function.
+Coroutines and threads need instances of @CoroutineCancelled@ and
+@ThreadCancelled@ respectively to use all of their functionality.
+When a new data type is declared with @coroutine@ or @thread@
 the forward declaration for the instance is created as well. The definition
 of the virtual table is created at the definition of the main function.
 …
 function.
 The function is \codeC{__cfa__virtual_cast} and it is implemented in the
+The function is @__cfa__virtual_cast@ and it is implemented in the
 standard library. It takes a pointer to the target type's virtual table and
 the object pointer being cast. The function is very simple, getting the
 …
 For the generated code a forward decaration of the virtual works as follows.
 There is a forward declaration of \codeC{__cfa__virtual_cast} in every cfa
+There is a forward declaration of @__cfa__virtual_cast@ in every cfa
 file so it can just be used. The object argument is the expression being cast
 so that is just placed in the argument list.
 …
 often across functions.
 At a very basic level this can be done with \codeC{setjmp} \& \codeC{longjmp}
+At a very basic level this can be done with @setjmp@ \& @longjmp@
 which simply move the top of the stack, discarding everything on the stack
 above a certain point. However this ignores all the clean-up code that should
 …
 both of these problems.
 Libunwind, provided in \texttt{unwind.h} on most platorms, is a C library
+Libunwind, provided in @unwind.h@ on most platorms, is a C library
 that provides \CPP style stack unwinding. Its operation is divided into two
 phases. The search phase -- phase 1 -- is used to scan the stack and decide
 …
 GCC will generate an LSDA and attach its personality function with the
 \texttt{-fexceptions} flag. However this only handles the cleanup attribute.
+@-fexceptions@ flag. However this only handles the cleanup attribute.
 This attribute is used on a variable and specifies a function that should be
 run when the variable goes out of scope. The function is passed a pointer to
 …
 messages for special cases (some of which should never be used by the
 personality function) and error codes but unless otherwise noted the
 personality function should always return \codeC{_URC_CONTINUE_UNWIND}.
 The \codeC{version} argument is the verson of the implementation that is
+personality function should always return @_URC_CONTINUE_UNWIND@.
+The @version@ argument is the verson of the implementation that is
 calling the personality function. At this point it appears to always be 1 and
 it will likely stay that way until a new version of the API is updated.
 The \codeC{action} argument is set of flags that tell the personality
+The @action@ argument is set of flags that tell the personality
 function when it is being called and what it must do on this invocation.
 The flags are as follows:
 \begin{itemize}
 \item\codeC{_UA_SEARCH_PHASE}: This flag is set whenever the personality
+\item@_UA_SEARCH_PHASE@: This flag is set whenever the personality
 function is called during the search phase. The personality function should
 decide if unwinding will stop in this function or not. If it does then the
 personality function should return \codeC{_URC_HANDLER_FOUND}.
 \item\codeC{_UA_CLEANUP_PHASE}: This flag is set whenever the personality
+personality function should return @_URC_HANDLER_FOUND@.
+\item@_UA_CLEANUP_PHASE@: This flag is set whenever the personality
 function is called during the cleanup phase. If no other flags are set this
 means the entire frame will be unwound and all cleanup code should be run.
 \item\codeC{_UA_HANDLER_FRAME}: This flag is set during the cleanup phase
+\item@_UA_HANDLER_FRAME@: This flag is set during the cleanup phase
 on the function frame that found the handler. The personality function must
 prepare to return to normal code execution and return
 \codeC{_URC_INSTALL_CONTEXT}.
 \item\codeC{_UA_FORCE_UNWIND}: This flag is set if the personality function
+@_URC_INSTALL_CONTEXT@.
+\item@_UA_FORCE_UNWIND@: This flag is set if the personality function
 is called through a forced unwind call. Forced unwind only performs the
 cleanup phase and uses a different means to decide when to stop. See its
 …
 \end{itemize}
 The \codeC{exception_class} argument is a copy of the \codeC{exception}'s
 \codeC{exception_class} field.
 The \codeC{exception} argument is a pointer to the user provided storage
+The @exception_class@ argument is a copy of the @exception@'s
+@exception_class@ field.
+The @exception@ argument is a pointer to the user provided storage
 object. It has two public fields, the exception class which is actually just
 a number that identifies the exception handling mechanism that created it and
 …
 exception needs to
 The \codeC{context} argument is a pointer to an opaque type. This is passed
+The @context@ argument is a pointer to an opaque type. This is passed
 to the many helper functions that can be called inside the personality
 function.
 …
 functions traversing the stack new-to-old until a function finds a handler or
 the end of the stack is reached. In the latter case raise exception will
 return with \codeC{_URC_END_OF_STACK}.
+return with @_URC_END_OF_STACK@.
 Once a handler has been found raise exception continues onto the the cleanup
 …
 If an error is encountered raise exception will return either
 \codeC{_URC_FATAL_PHASE1_ERROR} or \codeC{_URC_FATAL_PHASE2_ERROR} depending
+@_URC_FATAL_PHASE1_ERROR@ or @_URC_FATAL_PHASE2_ERROR@ depending
 on when the error occured.
 …
 been unwound.
 Each time it is called the stop function should return \codeC{_URC_NO_REASON}
+Each time it is called the stop function should return @_URC_NO_REASON@
 or transfer control directly to other code outside of libunwind. The
 framework does not provide any assistance here.
 Its arguments are the same as the paired personality function.
 The actions \codeC{_UA_CLEANUP_PHASE} and \codeC{_UA_FORCE_UNWIND} are always
+The actions @_UA_CLEANUP_PHASE@ and @_UA_FORCE_UNWIND@ are always
 set when it is called. By the official standard that is all but both GCC and
 Clang add an extra action on the last call at the end of the stack:
 \codeC{_UA_END_OF_STACK}.
+@_UA_END_OF_STACK@.
 \section{Exception Context}
 …
 Each stack has its own exception context. In a purely sequental program, using
 only core Cforall, there is only one stack and the context is global. However
 if the library \texttt{libcfathread} is linked then there can be multiple
+if the library @libcfathread@ is linked then there can be multiple
 stacks so they will each need their own.
 To handle this code always gets the exception context from the function
 \codeC{this_exception_context}. The main exception handling code is in
 \texttt{libcfa} and that library also defines the function as a weak symbol
 so it acts as a default. Meanwhile in \texttt{libcfathread} the function is
+@this_exception_context@. The main exception handling code is in
+@libcfa@ and that library also defines the function as a weak symbol
+so it acts as a default. Meanwhile in @libcfathread@ the function is
 defined as a strong symbol that replaces it when the libraries are linked
 together.
 The version of the function defined in \texttt{libcfa} is very simple. It
+The version of the function defined in @libcfa@ is very simple. It
 returns a pointer to a global static variable. With only one stack this
 global instance is associated with the only stack.
 The version of the function defined in \texttt{libcfathread} has to handle
+The version of the function defined in @libcfathread@ has to handle
 more as there are multiple stacks. The exception context is included as
 part of the per-stack data stored as part of coroutines. In the cold data
 section, stored at the base of each stack, is the exception context for that
 stack. The \codeC{this_exception_context} uses the concurrency library to get
+stack. The @this_exception_context@ uses the concurrency library to get
 the current coroutine and through it the cold data section and the exception
 context.
 …
 to store the exception. Macros with pointer arthritic and type cast are
 used to move between the components or go from the embedded
 \codeC{_Unwind_Exception} to the entire node.
+@_Unwind_Exception@ to the entire node.
 All of these nodes are strung together in a linked list. One linked list per
 …
 C which is what the \CFA compiler outputs so a work-around is used.
 This work around is a function called \codeC{__cfaehm_try_terminate} in the
+This work around is a function called @__cfaehm_try_terminate@ in the
 standard library. The contents of a try block and the termination handlers
 are converted into functions. These are then passed to the try terminate
 …
 These nested functions and all other functions besides
 \codeC{__cfaehm_try_terminate} in \CFA use the GCC personality function and
 the \texttt{-fexceptions} flag to generate the LSDA. This allows destructors
+@__cfaehm_try_terminate@ in \CFA use the GCC personality function and
+the @-fexceptions@ flag to generate the LSDA. This allows destructors
 to be implemented with the cleanup attribute.
 …
 The handler function does both the matching and catching. It tries each
 the condition of \codeCFA{catchResume} in order, top-to-bottom and until it
+the condition of @catchResume@ in order, top-to-bottom and until it
 finds a handler that matches. If no handler matches then the function returns
 false. Otherwise the matching handler is run, if it completes successfully
 the function returns true. Rethrows, through the \codeCFA{throwResume;}
+the function returns true. Rethrows, through the @throwResume;@
 statement, cause the function to return true.
+% Recursive Resumption Stuff:
+Blocking out part of the stack is accomplished by updating the front of the
+list as the search continues. Before the handler at a node is called the head
+of the list is updated to the next node of the current node. After the search
+is complete, successful or not, the head of the list is reset.
+This means the current handler and every handler that has already been
+checked are not on the list while a handler is run. If a resumption is thrown
+during the handling of another resumption the active handlers and all the
+other handler checked up to this point will not be checked again.
+This structure also supports new handler added while the resumption is being
+handled. These are added to the front of the list, pointing back along the
+stack -- the first one will point over all the checked handlers -- and the
+ordering is maintained.
 \subsection{Libunwind Compatibility}
 …
 Cancellation also uses libunwind to do its stack traversal and unwinding,
 however it uses a different primary function \codeC{_Unwind_ForcedUnwind}.
+however it uses a different primary function @_Unwind_ForcedUnwind@.
 Details of its interface can be found in the unwind section.

doc/theses/andrew_beach_MMath/unwinding.tex

-                      r5869cea
+                      r7b91c0e
 Even this is fairly simple if nothing needs to happen when the stack unwinds.
 Traditional C can unwind the stack by saving and restoring state (with
 \codeC{setjmp} \& \codeC{longjmp}). However many languages define actions that
+@setjmp@ \& @longjmp@). However many languages define actions that
 have to be taken when something is removed from the stack, such as running
 a variable's destructor or a \codeCFA{try} statement's \codeCFA{finally}
+a variable's destructor or a @try@ statement's @finally@
 clause. Handling this requires walking the stack going through each stack
 frame.
 …
 \CFA uses two primary functions in libunwind to create most of its
 exceptional control-flow: \codeC{_Unwind_RaiseException} and
 \codeC{_Unwind_ForcedUnwind}.
+exceptional control-flow: @_Unwind_RaiseException@ and
+@_Unwind_ForcedUnwind@.
 Their operation is divided into two phases: search and clean-up. The search
 phase -- phase 1 -- is used to scan the stack but not unwinding it. The
 …
 A personality function performs three tasks, although not all have to be
 present. The tasks performed are decided by the actions provided.
 \codeC{_Unwind_Action} is a bitmask of possible actions and an argument of
+@_Unwind_Action@ is a bitmask of possible actions and an argument of
 this type is passed into the personality function.
 \begin{itemize}
 \item\codeC{_UA_SEARCH_PHASE} is passed in search phase and tells the
+\item@_UA_SEARCH_PHASE@ is passed in search phase and tells the
 personality function to check for handlers. If there is a handler in this
 stack frame, as defined by the language, the personality function should
 return \codeC{_URC_HANDLER_FOUND}. Otherwise it should return
 \codeC{_URC_CONTINUE_UNWIND}.
 \item\codeC{_UA_CLEANUP_PHASE} is passed in during the clean-up phase and
+return @_URC_HANDLER_FOUND@. Otherwise it should return
+@_URC_CONTINUE_UNWIND@.
+\item@_UA_CLEANUP_PHASE@ is passed in during the clean-up phase and
 means part or all of the stack frame is removed. The personality function
 should do whatever clean-up the language defines
 (such as running destructors/finalizers) and then generally returns
 \codeC{_URC_CONTINUE_UNWIND}.
 \item\codeC{_UA_HANDLER_FRAME} means the personality function must install
+@_URC_CONTINUE_UNWIND@.
+\item@_UA_HANDLER_FRAME@ means the personality function must install
 a handler. It is also passed in during the clean-up phase and is in addition
 to the clean-up action. libunwind provides several helpers for the personality
 function here. Once it is done, the personality function must return
 \codeC{_URC_INSTALL_CONTEXT}.
+@_URC_INSTALL_CONTEXT@.
 \end{itemize}
 The personality function is given a number of other arguments. Some are for
 compatability and there is the \codeC{struct _Unwind_Context} pointer which
+compatability and there is the @struct _Unwind_Context@ pointer which
 passed to many helpers to get information about the current stack frame.
 …
 raise-exception but with some extras.
 The first it passes in an extra action to the personality function on each
 stack frame, \codeC{_UA_FORCE_UNWIND}, which means a handler cannot be
+stack frame, @_UA_FORCE_UNWIND@, which means a handler cannot be
 installed.
 …
 stack frames have been removed. By the standard API this is marked by setting
 the stack pointer inside the context passed to the stop function. However both
 GCC and Clang add an extra action for this case \codeC{_UA_END_OF_STACK}.
+GCC and Clang add an extra action for this case @_UA_END_OF_STACK@.
 Each time function the stop function is called it can do one or two things.
 When it is not the end of the stack it can return \codeC{_URC_NO_REASON} to
+When it is not the end of the stack it can return @_URC_NO_REASON@ to
 continue unwinding.
 % Is there a reason that NO_REASON is used instead of CONTINUE_UNWIND?
 …
 The stop function is very simple. It checks the end of stack flag to see if
 it is finished unwinding. If so, it calls \codeC{exit} to end the process,
+it is finished unwinding. If so, it calls @exit@ to end the process,
 otherwise it returns with no-reason to continue unwinding.
 % Yeah, this is going to have to change.
 …
 location of the instruction pointer and stack layout, which varies with
 compiler and optimization levels. So for frames where there are only
 destructors, GCC's attribute cleanup with the \texttt{-fexception} flag is
+destructors, GCC's attribute cleanup with the @-fexception@ flag is
 sufficient to handle unwinding.
 The only functions that require more than that are those that contain
+\codeCFA{try} statements. A \codeCFA{try} statement has a \codeCFA{try}
 clause, some number of \codeCFA{catch} clauses and \codeCFA{catchResume}
 clauses and may have a \codeCFA{finally} clause. Of these only \codeCFA{try}
 statements with \codeCFA{catch} clauses need to be transformed and only they
 and the \codeCFA{try} clause are involved.
+@try@ statements. A @try@ statement has a @try@
+clause, some number of @catch@ clauses and @catchResume@
+clauses and may have a @finally@ clause. Of these only @try@
+statements with @catch@ clauses need to be transformed and only they
+and the @try@ clause are involved.
 The \codeCFA{try} statement is converted into a series of closures which can
+The @try@ statement is converted into a series of closures which can
 access other parts of the function according to scoping rules but can be
 passed around. The \codeCFA{try} clause is converted into the try functions,
 almost entirely unchanged. The \codeCFA{catch} clauses are converted into two
+passed around. The @try@ clause is converted into the try functions,
+almost entirely unchanged. The @catch@ clauses are converted into two
 functions; the match function and the catch function.
 …
 runs the handler's body.
 These three functions are passed to \codeC{try_terminate}. This is an
+These three functions are passed to @try_terminate@. This is an
 % Maybe I shouldn't quote that, it isn't its actual name.
 internal hand-written function that has its own personality function and
 …
 handler was found in this frame. If it was then the personality function
 installs the handler, which is setting the instruction pointer in
 \codeC{try_terminate} to an otherwise unused section that calls the catch
+@try_terminate@ to an otherwise unused section that calls the catch
 function, passing it the current exception and handler index.
 \codeC{try_terminate} returns as soon as the catch function returns.
+@try_terminate@ returns as soon as the catch function returns.
 At this point control has returned to normal control flow.

doc/theses/fangren_yu_COOP_F20/Report.tex

-                      r5869cea
+                      r7b91c0e
 \usepackage[usenames]{color}
 \input{common}                                          % common CFA document macros
 \usepackage[dvips,plainpages=false,pdfpagelabels,pdfpagemode=UseNone,colorlinks=true,pagebackref=true,linkcolor=blue,citecolor=blue,urlcolor=blue,pagebackref=true,breaklinks=true]{hyperref}
+\usepackage[dvips,plainpages=false,pdfpagelabels,pdfpagemode=UseNone,colorlinks=true,linkcolor=blue,citecolor=blue,urlcolor=blue,pagebackref=true,breaklinks=true]{hyperref}
 \usepackage{breakurl}
 \urlstyle{sf}
 …
 \renewcommand{\subsectionmark}[1]{\markboth{\thesubsection\quad #1}{\thesubsection\quad #1}}
 \pagenumbering{roman}
 \linenumbers                                            % comment out to turn off line numbering
+%\linenumbers                                            % comment out to turn off line numbering
 \maketitle
 \pdfbookmark[1]{Contents}{section}
+\tableofcontents
+\clearpage
 \thispagestyle{plain}
 \pagenumbering{arabic}
 \begin{abstract}
+\CFA is an evolutionary extension to the C programming language, featuring a parametric type system, and is currently under active development. The reference compiler for \CFA language, @cfa-cc@, has some of its major components dated back to early 2000s, and is based on inefficient data structures and algorithms. Some improvements targeting the expression resolution algorithm, suggested by a recent prototype experiment on a simplified model, are implemented in @cfa-cc@ to support the full \CFA language. These optimizations speed up the compiler significantly by a factor of 20 across the existing \CFA codebase, bringing the compilation time of a mid-sized \CFA source file down to 10-second level. A few cases derived from realistic code examples that causes trouble to the compiler are analyzed in detail, with proposed solutions. This step of \CFA project development is critical to its eventual goal to be used alongside C for large software systems.
+\CFA is an evolutionary, non-object-oriented extension of the C programming language, featuring a parametric type-system, and is currently under active development. The reference compiler for the \CFA language, @cfa-cc@, has some of its major components dated back to the early 2000s, which are based on inefficient data structures and algorithms. This report introduces improvements targeting the expression resolution algorithm, suggested by a recent prototype experiment on a simplified model, which are implemented in @cfa-cc@ to support the full \CFA language. These optimizations speed up the compiler by a factor of 20 across the existing \CFA codebase, bringing the compilation time of a mid-sized \CFA source file down to the 10-second level. A few problem cases derived from realistic code examples are analyzed in detail, with proposed solutions. This work is a critical step in the \CFA project development to achieve its eventual goal of being used alongside C for large software systems.
 \end{abstract}
+\clearpage
+\section*{Acknowledgements}
+\begin{sloppypar}
+I would like to thank everyone in the \CFA team for their contribution towards this project. Programming language design and development is a tough subject and requires a lot of teamwork. Without the collaborative efforts from the team, this project could not have been a success. Specifically, I would like to thank Andrew Beach for introducing me to the \CFA codebase, Thierry Delisle for maintaining the test and build automation framework, Michael Brooks for providing example programs of various experimental language and type system features, and most importantly, Professor Martin Karsten for recommending me to the \CFA team, and my supervisor, Professor Peter Buhr for encouraging me to explore deeply into intricate compiler algorithms. Finally, I gratefully acknowledge the help from Aaron Moss, former graduate from the team and the author of the precedent thesis work, to participate in the \CFA team's virtual conferences and email correspondence, and provide many critical arguments and suggestions. 2020 had been an unusually challenging year for everyone and we managed to keep a steady pace.
+\end{sloppypar}
+\clearpage
+\tableofcontents
+\clearpage
 \section{Introduction}
 \CFA language, developed by the Programming Language Group at University of Waterloo, has a long history, with the first proof-of-concept compiler built in 2003 by Richard Bilson~\cite{Bilson03}. Many new features are added to the language over time, but the core of \CFA, parametric functions introduced by the @forall@ clause (hence the name of the language), with the type system supporting parametric overloading, remains mostly unchanged.
 The current \CFA reference compiler @cfa-cc@ still includes many parts taken directly from the original Bilson's implementation, and serves as a starting point for the enhancement work to the type system. Unfortunately, it does not provide the efficiency required for the language to be used practically: a \CFA source file of approximately 1000 lines of code can take a few minutes to compile. The cause of the problem is that the old compiler used inefficient data structures and algorithms for expression resolution, which involved a lot of copying and redundant work.
 This paper presents a series of optimizations to the performance-critical parts of the resolver, with a major rework of the data structure used by the compiler, using a functional programming approach to reduce memory complexity. Subsequent improvements are mostly suggested by running the compiler builds with a performance profiler against the \CFA standard library source code and a test suite to find the most underperforming components in the compiler algorithm.
 The \CFA team endorses a pragmatic philosophy in work that mostly focuses on practical implications of language design and implementation, rather than the theoretical limits. In particular, the compiler is designed to work on production \CFA code efficiently and keep type safety, while sometimes making compromises to expressiveness in extreme corner cases. However, when these corner cases do appear in actual usage, they need to be thoroughly investigated. Analysis presented in this paper, therefore, are conducted on a case-by-case basis. Some of them eventually point to certain weaknesses in the language design and solutions are proposed based on experimental results.
 \section{Completed work}
+\CFA language, developed by the Programming Language Group at the University of Waterloo, has a long history, with the initial language design in 1992 by Glen Ditchfield~\cite{Ditchfield92} and the first proof-of-concept compiler built in 2003 by Richard Bilson~\cite{Bilson03}. Many new features have been added to the language over time, but the core of \CFA's type-system --- parametric functions introduced by the @forall@ clause (hence the name of the language) providing parametric overloading --- remains mostly unchanged.
+The current \CFA reference compiler, @cfa-cc@, is designed using the visitor pattern~\cite{vistorpattern} over an abstract syntax tree (AST), where multiple passes over the AST modify it for subsequent passes. @cfa-cc@ still includes many parts taken directly from the original Bilson implementation, which served as the starting point for this enhancement work to the type system. Unfortunately, the prior implementation did not provide the efficiency required for the language to be practical: a \CFA source file of approximately 1000 lines of code can take a multiple minutes to compile. The cause of the problem is that the old compiler used inefficient data structures and algorithms for expression resolution, which involved significant copying and redundant work.
+This report presents a series of optimizations to the performance-critical parts of the resolver, with a major rework of the compiler data-structures using a functional-programming approach to reduce memory complexity. The improvements were suggested by running the compiler builds with a performance profiler against the \CFA standard-library source-code and a test suite to find the most underperforming components in the compiler algorithm.
+The \CFA team endorses a pragmatic philosophy that focuses on practical implications of language design and implementation rather than theoretical limits. In particular, the compiler is designed to be expressive with respect to code reuse while maintaining type safety, but compromise theoretical soundness in extreme corner cases. However, when these corner cases do appear in actual usage, they need to be thoroughly investigated. A case-by-case analysis is presented for several of these corner cases, some of which point to certain weaknesses in the language design with solutions proposed based on experimental results.
+\section{AST restructuring}
 \subsection{Memory model with sharing}
+A major rework of the abstract syntax tree (AST) data structure in the compiler is completed as the first step of the project. The majority of work were documented in the reference manual of the compiler~\cite{cfa-cc}. To summarize:
+\begin{itemize}
+\item
+AST nodes (and therefore subtrees) can be shared without copying when reused.
+\item
+Modifications apply the functional programming principle, making copies for local changes without affecting the original data shared by other owners. In-place mutations are permitted as a special case when sharing does not happen. The logic is implemented by reference counting.
+\item
+Memory allocation and freeing are performed automatically using smart pointers.
+\end{itemize}
+The resolver algorithm designed for overload resolution naturally introduces a significant amount of reused intermediate representations, especially in the following two places:
+\begin{itemize}
+\item
+Function overload candidates are computed by combining the argument candidates bottom-up, with many of them being a common term. For example, if $n$ overloads of a function @f@ all take an integer for the first parameter but different types for the second (@f( int, int )@, @f( int, double )@, etc.) the first term is reused $n$ times for each of the generated candidate expressions. This effect is particularly bad for deep expression trees.
+\item
+In the unification algorithm and candidate elimination step, actual types are obtained by substituting the type parameters by their bindings. Let $n$ be the complexity (\ie number of nodes in representation) of the original type, $m$ be the complexity of bound type for parameters, and $k$ be the number of occurrences of type parameters in the original type. If everything needs to be deep-copied, the substitution step takes $O(n+mk)$ time and memory, while using shared nodes it is reduced to $O(n)$ time and $O(k)$ memory.
+\end{itemize}
+One of the worst examples for the old compiler is a long chain of I/O operations
+\begin{cfa}
+sout | 1 | 2 | 3 | 4 | ...
+\end{cfa}
+The pipe operator is overloaded by \CFA I/O library for every primitive type in C language, as well as I/O manipulators defined by the library. In total there are around 50 overloads for the output stream operation. On resolving the $n$-th pipe operator in the sequence, the first term, which is the result of sub-expression containing $n-1$ pipe operators, is reused to resolve every overload. Therefore at least $O(n^2)$ copies of expression nodes are made during resolution, not even counting type unification cost; combined with two large factors from number of overloads of pipe operators, and that the ``output stream type'' in \CFA is a trait with 27 assertions (which adds to complexity of the pipe operator's type) this makes compiling a long output sequence extremely slow. In new AST representation only $O(n)$ copies are required and type of pipe operator is not copied at all.
+Reduction in space complexity is especially important, as preliminary profiling result on the old compiler build shows that over half of time spent in expression resolution are on memory allocations.
+A major rework of the AST data-structure in the compiler was completed as the first step of the project. The majority of this work is documented in my prior report documenting the compiler reference-manual~\cite{cfa-cc}. To summarize:
+\begin{itemize}
+\item
+AST nodes (and therefore subtrees) can be shared without copying.
+\item
+Modifications are performed using functional-programming principles, making copies for local changes without affecting the original data shared by other owners. In-place mutations are permitted as a special case when there is no sharing. The logic is implemented by reference counting.
+\item
+Memory allocation and freeing are performed automatically using smart pointers~\cite{smartpointers}.
+\end{itemize}
+The resolver algorithm, designed for overload resolution, uses a significant amount of reused, and hence copying, for the intermediate representations, especially in the following two places:
+\begin{itemize}
+\item
+Function overload candidates are computed by combining the argument candidates bottom-up, with many being a common term. For example, if $n$ overloads of a function @f@ all take an integer for the first parameter but different types for the second, \eg @f( int, int )@, @f( int, double )@, etc., the first term is copied $n$ times for each of the generated candidate expressions. This copying is particularly bad for deep expression trees.
+\item
+In the unification algorithm and candidate elimination step, actual types are obtained by substituting the type parameters by their bindings. Let $n$ be the complexity (\ie number of nodes in representation) of the original type, $m$ be the complexity of the bound type for parameters, and $k$ be the number of occurrences of type parameters in the original type. If every substitution needs to be deep-copied, these copy step takes $O(n+mk)$ time and memory, while using shared nodes it is reduced to $O(n)$ time and $O(k)$ memory.
+\end{itemize}
+One of the worst examples for the old compiler is a long chain of I/O operations:
+\begin{cfa}
+sout | 1 | 2 | 3 | 4 | ...;   // print integer constants
+\end{cfa}
+The pipe operator is overloaded by the \CFA I/O library for every primitive type in the C language, as well as I/O manipulators defined by the library. In total, there are around 50 overloads for the output stream operation. On resolving the $n$-th pipe operator in the sequence, the first term, which is the result of sub-expression containing $n-1$ pipe operators, is reused to resolve every overload. Therefore at least $O(n^2)$ copies of expression nodes are made during resolution, not even counting type unification cost; combined with the two large factors from number of overloads of pipe operators, and that the ``output stream type'' in \CFA is a trait with 27 assertions (which adds to complexity of the pipe operator's type) this makes compiling a long output sequence extremely slow. In the new AST representation, only $O(n)$ copies are required and the type of the pipe operator is not copied at all.
+Reduction in space complexity is especially important, as preliminary profiling results on the old compiler build showed over half of the time spent in expression resolution is on memory allocations.
+Since the compiler codebase is large and the new memory model mostly benefits expression resolution, some of the old data structures are still kept, and a conversion pass happens before and after the general resolve phase. Rewriting every compiler module will take longer, and whether the new model is correct was unknown when this project started, therefore only the resolver is currently implemented with the new data structure.
 \subsection{Merged resolver calls}
 The pre-resolve phase of compilation, inadequately called ``validate'' in the compiler source code, does more than just simple syntax validation, as it also normalizes input program. Some of them, however, requires type information on expressions and therefore needs to call the resolver before the general resolve phase. There are three notable places where the resolver is invoked:
 \begin{itemize}
 \item
+Attempt to generate default constructor, copy constructor and destructor for user-defined @struct@ types
 \item
 Resolve @with@ statements (the same as in Python, which introduces fields of a structure directly in scope)
+The pre-resolve phase of compilation, inappropriately called ``validate'' in the compiler source code, has a number of passes that do more than simple syntax and semantic validation; some passes also normalizes the input program. A few of these passes require type information for expressions, and therefore, need to call the resolver before the general resolve phase. There are three notable places where the resolver is invoked:
+\begin{itemize}
+\item
+Generate default constructor, copy constructor and destructor for user-defined @struct@ types.
+\item
+Resolve @with@ statements (the same as in Pascal~\cite{pascal}), which introduces fields of a structure directly into a scope.
 \item
 Resolve @typeof@ expressions (cf. @decltype@ in \CC); note that this step may depend on symbols introduced by @with@ statements.
 \end{itemize}
+Since the compiler codebase is large and the new memory model mostly only benefits expression resolution, the old data structure is still kept, and a conversion pass happens before and after resolve phase. Rewriting every compiler module will take a long time, and whether the new model is correct is still unknown when started, therefore only the resolver is implemented with the new data structure.
+Since the constructor calls were one of the most expensive to resolve (reason will be shown in the next section), pre-resolve phase were taking more time after resolver moves to the more efficient new implementation. To better facilitate the new resolver, every step that requires type information are reintegrated as part of resolver.
+A by-product of this work is that the reversed dependence of @with@ statement and @typeof@ can now be handled. Previously, the compiler is unable to handle cases such as
+Since the constructor calls are one of the most expensive to resolve (reason given in~\VRef{s:SpecialFunctionLookup}), this pre-resolve phase was taking a large amount of time even after the resolver was changed to the more efficient new implementation. The problem is that multiple resolutions repeat a significant amount of work. Therefore, to better facilitate the new resolver, every step that requires type information should be integrated as part of the general resolver phase.
+A by-product of this work is that reversed dependence between @with@ statement and @typeof@ can now be handled. Previously, the compiler was unable to handle cases such as:
 \begin{cfa}
 struct S { int x; };
 S foo();
 typeof( foo() ) s; // type is S
 with (s) {
+with (s) {
         x; // refers to s.x
+}
 \end{cfa}
 since type of @s@ is still unresolved when handling @with@ expressions. Instead, the new (and correct) approach is to evaluate @typeof@ expressions when the declaration is first seen, and it suffices because of the declaration-before-use rule.
+since the type of @s@ is unresolved when handling @with@ expressions because the @with@ pass follows the @typeof@ pass (interchanging passes only interchanges the problem). Instead, the new (and correct) approach is to evaluate @typeof@ expressions when the declaration is first seen during resolution, and it suffices because of the declaration-before-use rule.
 \subsection{Special function lookup}
+Reducing the number of functions looked up for overload resolution is an effective way to gain performance when there are many overloads but most of them are trivially wrong. In practice, most functions have few (if any) overloads but there are notable exceptions. Most importantly, constructor @?{}@, destructor @^?{}@, and assignment @?=?@ are generated for every user-defined type, and in a large source file there can be hundreds of them. Furthermore, many calls to them are generated for initializing variables and passing arguments. This fact makes them the most overloaded and most called functions.
+In an object-oriented programming language, object has methods declared with their types, so a call such as @obj.f()@ only needs to perform lookup in the method table corresponding to type of @obj@. \CFA on the other hand, does not have methods, and all types are open (\ie new operations can be defined on them), so a similar approach will not work in general. However, the ``big 3'' operators have a unique property enforced by the language rules, such that the first parameter must have a reference type. Since \CFA does not have class inheritance, reference type must always match exactly. Therefore, argument-dependent lookup can be implemented for these operators, by using a dedicated symbol table.
+The lookup key used for the special functions is the mangled type name of the first parameter, which acts as the @this@ parameter in an object-oriented language. To handle generic types, the type parameters are stripped off, and only the base type is matched. Note that a constructor (destructor, assignment operator) taking arbitrary @this@ argument, for example @forall( dtype T ) void ?{}( T & );@ is not allowed, and it guarantees that if the @this@ type is known, all possible overloads can be found by searching with the given type. In case that the @this@ argument itself is overloaded, it is resolved first and all possible result types are used for lookup.
+Note that for the generated expressions, the particular variable for @this@ argument is fully known, without overloads, so the majority of constructor call resolutions only need to check for one given object type. Explicit constructor calls and assignment statements sometimes may require lookup for multiple types. In the extremely rare case that type of @this@ argument is yet unbound, everything will have to be checked, just like without the argument-dependent lookup algorithm; fortunately, this case almost never happens in practice. An example is found in the library function @new@:
+\label{s:SpecialFunctionLookup}
+Reducing the number of function looked ups for overload resolution is an effective way to gain performance when there are many overloads but most of them are trivially wrong. In practice, most functions have few (if any) overloads but there are notable exceptions. Most importantly, constructor @?{}@, destructor @^?{}@, and assignment @?=?@ are generated for every user-defined type (@struct@ and @union@ in C), and in a large source file there can be hundreds of them. Furthermore, many calls are generated for initializing variables, passing arguments and copying values. This fact makes them the most overloaded and most called functions.
+In an object-oriented programming language, the object-method types are scoped within a class, so a call such as @obj.f()@ only needs to perform lookup in the method table corresponding to the type of @obj@. \CFA on the other hand, does not have methods, and all types are open, \ie new operations can be defined on them without inheritance; at best a \CFA type can be constrained by a translation unit. However, the ``big 3'' operators have a unique property enforced by the language rules: the first parameter must be a reference to its associated type, which acts as the @this@ parameter in an object-oriented language. Since \CFA does not have class inheritance, the reference type must always match exactly. Therefore, argument-dependent lookup can be implemented for these operators by using a dedicated, fast symbol-table.
+The lookup key for the special functions is the mangled type name of the first parameter. To handle generic types, the type parameters are stripped off, and only the base type is matched. Note a constructor (destructor, assignment operator) may not take an arbitrary @this@ argument, \eg @forall( dtype T ) void ?{}( T & )@, thus guaranteeing that if the @this@ type is known, all possible overloads can be found by searching with this given type. In the case where the @this@ argument itself is overloaded, it is resolved first and all possible result types are used for lookup.
+Note that for a generated expression, the particular variable for the @this@ argument is fully known, without overloads, so the majority of constructor-call resolutions only need to check for one given object type. Explicit constructor calls and assignment statements sometimes require lookup for multiple types. In the extremely rare case that the @this@-argument type is unbound, all necessary types are guaranteed to be checked, as for the previous lookup without the argument-dependent lookup; fortunately, this complex case almost never happens in practice. An example is found in the library function @new@:
 \begin{cfa}
 forall( dtype T | sized( T ), ttype TT | { void ?{}( T &, TT ); } )
 T * new( TT p ) { return &(*malloc()){ p }; }
 \end{cfa}
 as @malloc@ may return a pointer to any type, depending on context.
 Interestingly, this particular line of code actually caused another complicated issue, where the unusually massive work of checking every constructor in presence makes the case even worse. Section~\ref{s:TtypeResolutionInfiniteRecursion} presents a detailed analysis for the problem.
 The ``callable'' operator @?()@ (cf. @operator()@ in \CC) could also be included in the special operator list, as it is usually only on user-defined types, and the restriction that first argument must be a reference seems reasonable in this case.
+as @malloc@ may return a pointer to any type, depending on context.
+Interestingly, this particular declaration actually causes another complicated issue, making the complex checking of every constructor even worse. \VRef[Section]{s:TtypeResolutionInfiniteRecursion} presents a detailed analysis of this problem.
+The ``callable'' operator @?()@ (cf. @operator()@ in \CC) can also be included in this special operator list, as it is usually only on user-defined types, and the restriction that the first argument must be a reference seems reasonable in this case.
 \subsection{Improvement of function type representation}
+Since substituting type parameters with their bound types is one fundamental operation in many parts of resolver algorithm (particularly unification and environment binding), making as few copies of type nodes as possible helps reducing memory complexity. Even with the new memory management model, allocation is still a significant factor of resolver performance. Conceptually, operations on type nodes of AST should be performed in functional programming style, treating the data structure as immutable and only copy when necessary. The in-place mutation is a mere optimization that does not change logic of operations.
+The model was broken on function types by an inappropriate design. Function types require some special treatment due to the existence of assertions. In particular, it must be able to distinguish two different kinds of type parameter usage:
+Since substituting type parameters with their bound types is one fundamental operation in many parts of resolver algorithm (particularly unification and environment binding), making as few copies of type nodes as possible helps reducing memory complexity. Even with the new memory management model, allocation is still a significant factor of resolver performance. Conceptually, operations on type nodes of the AST should be performed in functional-programming style, treating the data structure as immutable and only copying when necessary. The in-place mutation is a mere optimization that does not change the logic for operations.
+However, the model was broken for function types by an inappropriate design. Function types require special treatment due to the existence of assertions that constrain the types it supports. Specifically, it must be possible to distinguish two different kinds of type parameter usage:
 \begin{cfa}
 forall( dtype T ) void foo( T * t ) {
         forall( dtype U ) void bar( T * t, U * u ) { ... }
+}
 \end{cfa}
 Here, only @U@ is a free parameter in declaration of @bar@, as it appears in the function's own forall clause; while @T@ is not free.
 Moreover, the resolution algorithm also has to distinguish type bindings of multiple calls to the same function, for example with
+        forall( dtype U ) void bar( @T@ * t, @U@ * u ) { ... }
+}
+\end{cfa}
+Here, only @U@ is a free parameter in the nested declaration of function @bar@, as @T@ must be bound at the call site when resolving @bar@.
+Moreover, the resolution algorithm also has to distinguish type bindings of multiple calls to the same function, \eg:
 \begin{cfa}
 forall( dtype T ) int foo( T x );
 foo( foo( 1.0 ) );
 \end{cfa}
 The inner call has binding (T: double) while the outer call has binding (T: int). Therefore a unique representation of free parameters in each expression is required. This was previously done by creating a copy of the parameter declarations inside function type, and fixing references afterwards. However, fixing references is an inherently deep operation that does not work well with functional programming model, as it must be evaluated eagerly on the entire syntax tree representing the function type.
 The revised approach generates a unique ID value for each function call expression instance and represents an occurrence of free parameter type with a pair of generated ID and the original parameter declaration, so that references do not need to be fixed, and a shallow copy of function type is possible.
 Note that after the change, all declaration nodes in syntax tree representation maps one-to-one with the actual declarations in the program, and therefore are guaranteed to be unique. Such property can potentially enable more optimizations, and some related ideas are presented after Section~\ref{s:SharedSub-ExpressionCaseUniqueExpressions}.
+int i = foo( foo( 1.0 ) );
+\end{cfa}
+The inner call has binding (T: double) while the outer call has binding (T: int). Therefore a unique representation for the free parameters is required in each expression. This type binding was previously done by creating a copy of the parameter declarations inside the function type and fixing references afterwards. However, fixing references is an inherently deep operation that does not work well with the functional-programming style, as it forces eager evaluation on the entire syntax tree representing the function type.
+The revised approach generates a unique ID value for each function call expression instance and represents an occurrence of a free-parameter type with a pair of generated ID and original parameter declaration, so references are unique and a shallow copy of the function type is possible.
+Note that after the change, all declaration nodes in the syntax-tree representation now map one-to-one with the actual declarations in the program, and therefore are guaranteed to be unique. This property can potentially enable more optimizations, and some related ideas are presented at the end of \VRef{s:SharedSub-ExpressionCaseUniqueExpressions}.
 \subsection{Improvement of pruning steps}
 A minor improvement for candidate elimination is to skip the step on the function overloads themselves and only perform on results of function application. As function calls are usually by name, the name resolution rule dictates that every function candidate necessarily has a different type; indirect function calls are rare, and when they do appear, they usually will not have many possible interpretations, and those rarely matches exactly in argument type. Since function types have a much more complex representation than data types (with multiple parameters and assertions), checking equality on them also takes longer.
 A brief test of this approach shows that the number of function overloads considered in expression resolution increases by a negligible amount of less than 1 percent, while type comparisons in candidate elimination are cut by more than half. Improvement is consistent over all \CFA source files in the test suite.
+A minor improvement for candidate elimination is to skip the step on the function overloads and only check the results of function application. As function calls are usually by name (versus pointers to functions), the name resolution rule dictates that every function candidate necessarily has a different type; indirect function calls are rare, and when they do appear, there are even fewer cases with multiple interpretations, and these rarely match exactly in argument type. Since function types have a much more complex representation (with multiple parameters and assertions) than data types, checking equality on them also takes longer.
+A brief test of this approach shows that the number of function overloads considered in expression resolution increases by an amount of less than 1 percent, while type comparisons in candidate elimination are reduced by more than half. This improvement is consistent over all \CFA source files in the test suite.
 …
 \label{s:SharedSub-ExpressionCaseUniqueExpressions}
 Unique expression denotes an expression that must be evaluated only once, to prevent unwanted side effects. It is currently only a compiler artifact, generated on tuple member expression of the form
+Unique expression denotes an expression evaluated only once to prevent unwanted side effects. It is currently only a compiler artifact, generated for tuple-member expression of the form:
 \begin{cfa}
 struct S { int a; int b; };
 …
 s.[a, b]; // tuple member expression, type is [int, int]
 \end{cfa}
 If the aggregate expression contains function calls, it cannot be evaluated multiple times:
+If the aggregate expression is function call, it cannot be evaluated multiple times:
 \begin{cfa}
 S makeS();
 makeS().[a, b]; // this should only make one S
+makeS().[a, b]; // this should only generate a unique S
 \end{cfa}
 Before code generation, the above expression is internally represented as
 …
 \end{cfa}
 at code generation, where @_unique_var@ and @_unique_var_evaluated@ are generated variables whose scope covers all appearances of the same expression.
+Note that although the unique expression is only used for tuple expansion now, it is a generally useful construction, and can be seen in other languages, such as Scala's @lazy val@~\cite{Scala}; therefore it could be worthwhile to introduce the unique expression to a broader context in \CFA and even make it directly available to programmers.
+In the compiler's visitor pattern, however, this creates a problem where multiple paths to a logically unique expression exist, so it may be modified more than once and become ill-formed; some specific intervention is required to ensure that unique expressions are only visited once. Furthermore, a unique expression appearing in more than one places will be copied on mutation so its representation is no longer unique. Some hacks are required to keep it in sync, and the methods are different when mutating the unique expression instance itself or its underlying expression.
+Example when mutating the underlying expression (visit-once guard)
+The conditional check ensures a single call to @makeS()@ even though there are logically multiple calls because of the tuple field expansion.
+Note that although the unique expression is only used for tuple expansion now, it is a generally useful construction, and is seen in other programming languages, such as Scala's @lazy val@~\cite{Scala}; therefore it may be worthwhile to introduce the unique expression to a broader context in \CFA and even make it directly available to programmers.
+In the compiler's visitor pattern, however, this creates a problem where multiple paths to a logically unique expression exist, so it may be modified more than once and become ill-formed; some specific intervention is required to ensure unique expressions are only visited once. Furthermore, a unique expression appearing in more than one places is copied on mutation so its representation is no longer unique.
+Currently, special cases are required to keep everything synchronized, and the methods are different when mutating the unique expression instance itself or its underlying expression:
+\begin{itemize}
+\item
+When mutating the underlying expression (visit-once guard)
 \begin{cfa}
 void InsertImplicitCalls::previsit( const ast::UniqueExpr * unqExpr ) {
         if ( visitedIds.count( unqExpr->id ) ) visit_children = false;
+        @if ( visitedIds.count( unqExpr->id ) ) visit_children = false;@
         else visitedIds.insert( unqExpr->id );
+}
 \end{cfa}
+Example when mutating the unique instance itself, which actually creates copies
+\item
+When mutating the unique instance itself, which actually creates copies
 \begin{cfa}
 auto mutExpr = mutate( unqExpr ); // internally calls copy when shared
+if ( ! unqMap.count( unqExpr->id ) ) {
+@if ( ! unqMap.count( unqExpr->id ) ) {@
         ...
 } else {
 …
+}
 \end{cfa}
+Such workaround seems difficult to be fit into a common visitor template. This suggests the memory model may need different kinds of nodes to accurately represent the syntax tree.
+Together with the fact that declaration nodes are always unique, it is possible that AST nodes can be classified by three different types:
+\begin{itemize}
+\item
+\textbf{Strictly unique} with only one owner (declarations);
+\item
+\textbf{Logically unique} with (possibly) many owners but should not be copied (unique expression example presented here);
+\item
+\textbf{Shared} by functional programming model, which assume immutable data structure and are copied on mutation.
+\end{itemize}
+Such workarounds are difficult to fit into the common visitor pattern, which suggests the memory model may need different kinds of nodes to accurately represent this feature in the AST.
+Given that declaration nodes are unique, it is possible for AST nodes to be divided into three different types:
+\begin{itemize}
+\item
+\textbf{Singleton} with only one owner (declarations);
+\item
+\textbf{No-copy} with multiple owners but cannot be copied (unique expression example presented here);
+\item
+\textbf{Copy} by functional-programming style, which assumes immutable data structures that are copied on mutation.
 \end{itemize}
 The boilerplate code can potentially handle these three cases differently.
 …
 \section{Analysis of resolver algorithm complexity}
 The focus of this chapter is to identify and analyze some realistic cases that cause resolver algorithm to have an exponential run time. As previous work has shown [3], the overload resolution problem in \CFA has worst-case exponential complexity; however, only few specific patterns can trigger the exponential complexity in practice. Implementing heuristic-based optimization for those selected cases is helpful to alleviate the problem.
+The focus of this section is to identify and analyze some realistic cases that cause the resolver algorithm to have an exponential runtime. As previous work has shown~\cite[\S~4.2.1]{Moss19}, the overload resolution problem in \CFA has worst-case exponential complexity; however, only few specific patterns can trigger the exponential complexity in practice. Implementing heuristic-based optimization for those selected cases is helpful to alleviate the problem.
 …
 \label{s:UnboundReturnType}
 The interaction of return type overloading and polymorphic functions creates this problem of function calls with unbound return type, and is further complicated by the presence of assertions.
+The interaction of return-type overloading and polymorphic functions creates function calls with unbounded return-type, and is further complicated by the presence of assertions.
 The prime example of a function with unbound return type is the type-safe version of C @malloc@:
 \begin{cfa}
+// size deduced from type, so no need to provide the size argument
+forall( dtype T | sized( T ) ) T * malloc( void );
+\end{cfa}
+Unbound return type can be problematic in resolver algorithm complexity because a single match of function call with unbound return type may create multiple candidates. In the worst case, consider a function declared to return any @otype@:
+forall( dtype T | sized( T ) )
+T * malloc( void ) { return (T *)malloc( sizeof(T) ); } // call C malloc
+int * i = malloc();  // type deduced from left-hand size $\Rightarrow$ no size argument or return cast
+\end{cfa}
+An unbound return-type is problematic in resolver complexity because a single match of a function call with an unbound return type may create multiple candidates. In the worst case, consider a function declared that returns any @otype@ (defined \VPageref{otype}):
 \begin{cfa}
 forall( otype T ) T anyObj( void );
 \end{cfa}
 As the resolver attempts to satisfy the otype constraint on @T@, a single call to @anyObj()@ without the result type known creates at least as many candidates as the number of complete types currently in scope; with generic types it becomes even worse, for example, assuming a declaration of generic pair is available at that point:
+As the resolver attempts to satisfy the otype constraint on @T@, a call to @anyObj()@ in an expression, without the result type known, creates at least as many candidates as the number of complete types currently in scope; with generic types it becomes even worse, \eg assuming a declaration of a generic @pair@ is available at that point:
 \begin{cfa}
 forall( otype T, otype U ) struct pair { T first; U second; };
 \end{cfa}
 Then an @anyObj()@ call can result in arbitrarily complex types, such as @pair( pair( int,int ), pair( int,int ) )@, and the depth can grow indefinitely until the specified parameter depth limit, thus creating exponentially many candidates. However, the expected types allowed by parent expressions are practically very few, so most of those interpretations are invalid; if the result type is never bound up to top level, by the semantic rules it is ambiguous if there are more than one valid bindings, and resolution can fail fast. It is therefore reasonable to delay resolving assertions on an unbound parameter in return type; however, with the current cost model, such behavior may further cause irregularities in candidate selection, such that the presence of assertions can change the preferred candidate, even when order of expression costs are supposed to stay the same. Detailed analysis of this issue will be presented later, in the correctness part.
+Then an @anyObj()@ call can result in arbitrarily complex types, such as @pair( pair( int, int ), pair( int, int ) )@, and the depth can grow indefinitely until a specified parameter-depth limit, thus creating exponentially many candidates. However, the expected types allowed by parent expressions are practically very few, so most of those interpretations are invalid; if the result type is never bound up to the top level, by the semantic rules it is ambiguous if there is more than one valid binding and resolution fails quickly. It is therefore reasonable to delay resolving assertions on an unbound parameter in a return type; however, with the current cost model, such behavior may further cause irregularities in candidate selection, such that the presence of assertions can change the preferred candidate, even when order of expression costs are supposed to stay the same. A detailed analysis of this issue is presented in \VRef{s:AnalysisTypeSystemCorrectness}.
 …
 \label{s:TtypeResolutionInfiniteRecursion}
 @ttype@ (``tuple type'') is a relatively new addition to the language that attempts to provide type-safe variadic argument semantics. Unlike regular @dtype@ parameters, @ttype@ is only valid in function parameter list, and may only appear once as the type of last parameter. At the call site, a @ttype@ parameter is bound to the tuple type of all remaining function call arguments.
+@ttype@ (``tuple type'') is a relatively new addition to the language that attempts to provide type-safe variadic argument semantics. Unlike regular @dtype@ parameters, @ttype@ is only valid in a function parameter-list, and may only appear once as the last parameter type. At the call site, a @ttype@ parameter is bound to the tuple type of all remaining function-call arguments.
 There are two kinds of idiomatic @ttype@ usage: one is to provide flexible argument forwarding, similar to the variadic template in \CC (\lstinline[language=C++]|template<typename... args>|), as shown below in the implementation of @unique_ptr@
 …
         T * data;
 };
 forall( dtype T | sized( T ), ttype Args | { void ?{}( T &, Args ); })
 void ?{}( unique_ptr( T ) & this, Args args ) {
         this.data = new( args );
+}
 \end{cfa}
 the other is to implement structural recursion in the first-rest manner:
 \begin{cfa}
 forall( otype T, ttype Params | { void process( T ); void func( Params ); })
+forall( dtype T | sized( T ), @ttype Args@ | { void ?{}( T &, Args ); })
+void ?{}( unique_ptr( T ) & this, Args @args@ ) {
+        this.data = new( @args@ );  // forward constructor arguments to dynamic allocator
+}
+\end{cfa}
+The other usage is to implement structural recursion in the first-rest pattern:
+\begin{cfa}
+forall( otype T, @ttype Params@ | { void process( T ); void func( Params ); })
 void func( T arg1, Params p ) {
         process( arg1 );
+        func( p );
+}
+\end{cfa}
+For the second use case, it is important that the number of parameters in the recursive call go down, since the call site must deduce all assertion candidates, and that is only possible if by just looking at argument types (and not their values), the recursion is known to be completed in a finite number of steps.
+In recent experiments, however, some flaw in the type binding rules can lead to the first kind of @ttype@ use case produce an invalid candidate that the resolver enters an infinite loop.
+This bug was discovered in an attempt to raise assertion recursive depth limit and one of the library program takes exponentially longer time to compile. The cause of the problem is identified to be the following set of functions.
+File @memory.cfa@ contains
+\begin{cfa}
+#include "memory.hfa"
+#include "stdlib.hfa"
+\end{cfa}
+where file @memory.hfa@ contains the @unique_ptr@ declaration above, and two other similar functions with @ttype@ parameter:
+\begin{cfa}
+forall( dtype T | sized( T ), ttype Args | { void ?{}( T &, Args ); }) {
+        func( @p@ );  // recursive call until base case of one argument
+}
+\end{cfa}
+For the second use case, it is imperative the number of parameters in the recursive call goes down, since the call site must deduce all assertion candidates, and that is only possible if by observation of the argument types (and not their values), the recursion is known to be completed in a finite number of steps.
+In recent experiments, however, a flaw in the type-binding rules can lead to the first kind of @ttype@ use case producing an invalid candidate and the resolver enters an infinite loop.
+This bug was discovered in an attempt to raise the assertion recursive-depth limit and one of the library programs took exponentially longer to compile. The cause of the problem is the following set of functions:
+\begin{cfa}
+// unique_ptr  declaration from above
+forall( dtype T | sized( T ), ttype Args | { void ?{}( T &, Args ); } ) { // distribute forall clause
         void ?{}( counter_data( T ) & this, Args args );
         void ?{}( counter_ptr( T ) & this, Args args );
         void ?{}( unique_ptr( T ) & this, Args args );
+}
+\end{cfa}
+File @stdlib.hfa@ contains
+\begin{cfa}
 forall( dtype T | sized( T ), ttype TT | { void ?{}( T &, TT ); } )
+T * new( TT p ) { return &(*malloc()){ p }; }
+\end{cfa}
+In the expression @(*malloc()){p}@, the type of object being constructed is yet unknown, since the return type information is not immediately provided. That caused every constructor to be searched, and while normally a bound @ttype@ cannot be unified with any free parameter, it is possible with another free @ttype@. Therefore in addition to the correct option provided by assertion, 3 wrong options are examined, each of which again requires the same assertion, for an unknown base type T and @ttype@ arguments, and that becomes an infinite loop, until the specified recursion limit and resolution is forced to fail. Moreover, during the recursion steps, number of candidates grows exponentially, since there are always 3 options at each step.
+Unfortunately, @ttype@ to @ttype@ binding is necessary, to allow calling the function provided by assertion indirectly.
+\begin{cfa}
+forall( dtype T | sized( T ), ttype Args | { void ?{}( T &, Args ); })
+void ?{}( unique_ptr( T ) & this, Args args ) { this.data = (T * )new( args ); }
+\end{cfa}
+Here the constructor assertion is used for the @new( args )@ call.
+T * new( TT p ) { return @&(*malloc()){ p };@ }
+\end{cfa}
+In the expression @(*malloc()){p}@, the type of the object being constructed is unknown, since the return-type information is not immediately available. That causes every constructor to be searched, and while normally a bound @ttype@ cannot be unified with any free parameter, it is possible with another free @ttype@. Therefore, in addition to the correct option provided by the assertion, 3 wrong options are examined, each of which again requires the same assertion, for an unknown base-type @T@ and @ttype@ argument, which becomes an infinite loop until the specified recursion limit and resolution is fails. Moreover, during the recursion steps, the number of candidates grows exponentially, since there are always 3 options at each step.
+Unfortunately, @ttype@ to @ttype@ binding is necessary, to allow indirectly calling a function provided in an assertion.
+\begin{cfa}
+forall( dtype T | sized( T ), ttype Args | { @void ?{}( T &, Args );@ })
+void ?{}( unique_ptr( T ) & this, Args args ) { this.data = (T *)@new( args )@; } // constructor call
+\end{cfa}
+Here the constructor assertion is used by the @new( args )@ call to indirectly call the constructor on the allocated storage.
 Therefore, it is hard, perhaps impossible, to solve this problem by tweaking the type binding rules. An assertion caching algorithm can help improve this case by detecting cycles in recursion.
 Meanwhile, without the caching algorithm implemented, some changes in the \CFA source code are enough to eliminate this problem, at least in the current codebase. Note that the issue only happens with an overloaded variadic function, which rarely appears in practice, since the idiomatic use cases are for argument forwarding and self-recursion. The only overloaded @ttype@ function so far discovered in all of \CFA standard library code is the constructor, and by utilizing the argument-dependent lookup process described in Section~\ref{s:UnboundReturnType}, adding a cast before constructor call gets rid of the issue.
 \begin{cfa}
 T * new( TT p ) { return &(*(T * )malloc()){ p }; }
+Meanwhile, without a caching algorithm implemented, some changes in the \CFA source code are enough to eliminate this problem, at least in the current codebase. Note that the issue only happens with an overloaded variadic function, which rarely appears in practice, since the idiomatic use cases are for argument forwarding and self-recursion. The only overloaded @ttype@ function so far discovered in all of \CFA standard library is the constructor, and by utilizing the argument-dependent lookup process described in \VRef{s:UnboundReturnType}, adding a cast before the constructor call removes the issue.
+\begin{cfa}
+T * new( TT p ) { return &(*@(T * )@malloc()){ p }; }
 \end{cfa}
 …
 \subsection{Reused assertions in nested generic type}
 The following test of deeply nested dynamic generic type reveals that locally caching reused assertions is necessary, rather than just a resolver optimization, because recomputing assertions can result in bloated generated code size:
+The following test of deeply nested, dynamic generic type reveals that locally caching reused assertions is necessary, rather than just a resolver optimization, because recomputing assertions can result in bloated generated code size:
 \begin{cfa}
 struct nil {};
 …
 int main() {
         #if   N==0
         nil x;
+        nil @x@;
         #elif N==1
         cons( size_t, nil ) x;
+        cons( size_t, nil ) @x@;
         #elif N==2
         cons( size_t, cons( size_t, nil ) ) x;
+        cons( size_t, cons( size_t, nil ) ) @x@;
         #elif N==3
         cons( size_t, cons( size_t, cons( size_t, nil ) ) ) x;
+        cons( size_t, cons( size_t, cons( size_t, nil ) ) ) @x@;
         // similarly for N=4,5,6
         #endif
+}
 \end{cfa}
 At the declaration of @x@, it is implicitly initialized by generated constructor call, whose signature is given by
+At the declaration of @x@, it is implicitly initialized by generated constructor call, with signature:
 \begin{cfa}
 forall( otype L, otype R ) void ?{}( cons( L, R ) & );
 \end{cfa}
+Note that the @otype@ constraint contains 4 assertions:
+where the @otype@ constraint contains the 4 assertions:\label{otype}
 \begin{cfa}
 void ?{}( L & ); // default constructor
 …
 L & ?=?( L &, L & ); // assignment
 \end{cfa}
+Now since the right hand side of outermost cons is again a cons, recursive assertions are required. When the compiler cannot cache and reuse already resolved assertions, it becomes a problem, as each of those 4 pending assertions again asks for 4 more assertions one level below. Without any caching, number of resolved assertions grows exponentially, while that is obviously unnecessary since there are only $n+1$ different types involved. Even worse, this causes exponentially many wrapper functions generated later at the codegen step, and results in huge compiled binary.
 \begin{table}[h]
+\begin{table}[htb]
+\centering
 \caption{Compilation results of nested cons test}
+\label{t:NestedConsTest}
 \begin{tabular}{|r|r|r|}
 \hline
 …
 \end{table}
+As the local functions are implemented by emitting executable code on the stack~\cite{gcc-nested-func}, it eventually means that compiled code also has exponential run time. This problem has evident practical implications, as nested collection types are frequently used in real production code.
+Now since the right hand side of outermost cons is again a cons, recursive assertions are required. \VRef[Table]{t:NestedConsTest} shows when the compiler does not cache and reuse already resolved assertions, it becomes a problem, as each of these 4 pending assertions again asks for 4 more assertions one level below. Without caching, the number of resolved assertions grows exponentially, which is unnecessary since there are only $n+1$ different types involved. Even worse, this problem causes exponentially many wrapper functions to be generated at the backend, resulting in a huge binary. As the local functions are implemented by emitting executable code on the stack~\cite{gcc-nested-func}, it means that compiled code also has exponential run time. This problem has practical implications, as nested collection types are frequently used in real production code.
 \section{Analysis of type system correctness}
+\label{s:AnalysisTypeSystemCorrectness}
 In Moss' thesis~\cite[\S~4.1.2,~p.~45]{Moss19}, the author presents the following example:
 …
 From the set of candidates whose parameter and argument types have been unified and whose assertions have been satisfied, those whose sub-expression interpretations have the smallest total cost of conversion are selected ... The total cost of conversion for each of these candidates is then calculated based on the implicit conversions and polymorphism involved in adapting the types of the sub-expression interpretations to the formal parameter types.
 \end{quote}
+With this model, the algorithm picks @g1@ in resolving the @f( g( 42 ) )@ call, which seems to be undesirable.
+There are further evidence that shows the Bilson model is fundamentally incorrect, following the discussion of unbound return type in Section~\ref{s:UnboundReturnType}. By the conversion cost specification, a binding from a polymorphic type parameter to a concrete type incurs a polymorphic cost of 1. It remains unspecified \emph{when} the type parameters should become bound. When the parameterized types appear in the function parameters, they can be deduced from the argument type, and there is no ambiguity. In the unbound return case, however, the binding may happen at any stage in expression resolution, therefore it is impossible to define a unique local conversion cost. Note that type binding happens exactly once per parameter in resolving the entire expression, so the global binding cost is unambiguously 1.
+As per the current compiler implementation, it does have a notable inconsistency in handling such case. For any unbound parameter that does \emph{not} come with an associated assertion, it remains unbound to the parent expression; for those that does however, they are immediately bound in the assertion resolution step, and concrete result types are used in the parent expressions.
+With this model, the algorithm picks @g1@ in resolving the @f( g( 42 ) )@ call, which is undesirable.
+There is further evidence that shows the Bilson model is fundamentally incorrect, following the discussion of unbound return type in \VRef{s:UnboundReturnType}. By the conversion-cost specification, a binding from a polymorphic type-parameter to a concrete type incurs a polymorphic cost of 1. It remains unspecified \emph{when} the type parameters should become bound. When the parameterized types appear in function parameters, they can be deduced from the argument type, and there is no ambiguity. In the unbound return case, however, the binding may happen at any stage in expression resolution, therefore it is impossible to define a unique local conversion cost. Note that type binding happens exactly once per parameter in resolving the entire expression, so the global binding cost is unambiguously 1.
+In the current compiler implementation, there is a notable inconsistency in handling this case. For any unbound parameter that does \emph{not} come with an associated assertion, it remains unbound to the parent expression; for those that do, however, they are immediately bound in the assertion resolution step, and concrete result types are used in the parent expressions.
 Consider the following example:
 \begin{cfa}
 …
 void h( int * );
 \end{cfa}
 The expression @h( f() )@ eventually has a total cost of 1 from binding (T: int), but in the eager resolution model, the cost of 1 may occur either at call to @f@ or at call to @h@, and with the assertion resolution triggering a binding, the local cost of @f()@ is (0 poly, 0 spec) with no assertions, but (1 poly, -1 spec) with an assertion:
 \begin{cfa}
 forall( dtype T | { void g( T * ); } ) T * f( void );
+The expression @h( f() )@ eventually has a total cost of 1 from binding (T: int), but in the eager-resolution model, the cost of 1 may occur either at the call to @f@ or at call to @h@, and with the assertion resolution triggering a binding, the local cost of @f()@ is (0 poly, 0 spec) with no assertions, but (1 poly, -1 spec) with an assertion:
+\begin{cfa}
+forall( dtype T | @{ void g( T * ); }@ ) T * f( void );
 void g( int * );
 void h( int * );
 \end{cfa}
 and that contradicts the principle that adding assertions should make expression cost lower. Furthermore, the time at which type binding and assertion resolution happens is an implementation detail of the compiler, but not a part of language definition. That means two compliant \CFA compilers, one performing immediate assertion resolution at each step, and one delaying assertion resolution on unbound types, can produce different expression costs and therefore different candidate selection, making the language rule itself partially undefined and therefore unsound. By the above reasoning, the updated cost model using global sum of costs should be accepted as the standard. It also allows the compiler to freely choose when to resolve assertions, as the sum of total costs is independent of that choice; more optimizations regarding assertion resolution can also be implemented.
+and that contradicts the principle that adding assertions should make expression cost lower. Furthermore, the time at which type binding and assertion resolution happens is an implementation detail of the compiler, not part of the language definition. That means two compliant \CFA compilers, one performing immediate assertion resolution at each step, and one delaying assertion resolution on unbound types, can produce different expression costs and therefore different candidate selection, making the language rule itself partially undefined, and therefore, unsound. By the above reasoning, the updated cost model using global sum of costs should be accepted as the standard. It also allows the compiler to freely choose when to resolve assertions, as the sum of total costs is independent of that choice; more optimizations regarding assertion resolution can also be implemented.
 \section{Timing results}
+For the timing results presented here, the \CFA compiler is built with gcc 9.3.0, and tested on a server machine running Ubuntu 20.04, 64GB RAM and 32-core 2.2 GHz CPU, results reported by the time command, and using only 8 cores in parallel such that the time is close to the case with 100\% CPU utilization on a single thread.
+On the most recent build, the \CFA standard library (~1.3 MB of source code) compiles in 4 minutes 47 seconds total processor time (single thread equivalent), with the slowest file taking 13 seconds. The test suite (178 test cases, ~2.2MB of source code) completes within 25 minutes total processor time,\footnote{Including a few runtime tests; total time spent in compilation is approximately 21 minutes.} with the slowest file taking 23 seconds. In contrast, the library build on old compiler takes 85 minutes total, 5 minutes for the slowest file. Full test suite takes too long with old compiler build and is therefore not run, but the slowest test cases take approximately 5 minutes. Overall, the most recent build compared to old build in April 2020, before the project started, is consistently faster by a factor of 20.
+Additionally, 6 selected \CFA source files with distinct features from library and test suite are used to test compiler performance after each of the optimizations are implemented. Test files are from the most recent build and run through C preprocessor to eliminate the factor of header file changes. The selected tests are:
+\begin{itemize}
+\item
+@lib/fstream@ (112 KB)\footnote{File sizes are after preprocessing, with no line information (\lstinline|gcc -E -P|).}: implementation of I/O library
+For the timing results presented here, the \CFA compiler is built with gcc 9.3.0, and tested on a server machine running Ubuntu 20.04, 64GB RAM and 32-core 2.2 GHz CPU.
+Timing is reported by the @time@ command and an experiment is run using 8 cores, where each core is at 100\% CPU utilization.
+On the most recent build, the \CFA standard library ($\approx$1.3 MB of source code) compiles in 4 minutes 47 seconds total processor time (single thread equivalent), with the slowest file taking 13 seconds. The test suite (178 test cases, $\approx$2.2MB of source code) completes within 25 minutes total processor time,
+% PAB: I do not understand this footnote.
+%\footnote{Including a few runtime tests; total time spent in compilation is approximately 21 minutes.}
+with the slowest file taking 23 seconds. In contrast, the library build with the old compiler takes 85 minutes total, 5 minutes for the slowest file. The full test-suite takes too long with old compiler build and is therefore not run, but the slowest test cases take approximately 5 minutes. Overall, the most recent build compared to an old build is consistently faster by a factor of 20.
+Additionally, 6 selected \CFA source files with distinct features from the library and test suite are used to illustrate the compiler performance change after each of the implemented optimizations. Test files are from the most recent build and run through the C preprocessor to expand header file, perform macro expansions, but no line number information (@gcc -E -P@).
+\VRef[Table]{t:SelectedFileByCompilerBuild} shows the selected tests:
+\begin{itemize}
+\item
+@lib/fstream@ (112 KB)
 \item
 @lib/mutex@ (166 KB): implementation of concurrency primitive
 …
 @lib/stdlib@ (64 KB): type-safe wrapper to @void *@-based C standard library functions
 \item
 @test/ISO2@ (55 KB): application of I/O library
+@test/io2@ (55 KB): application of I/O library
 \item
 @test/thread@ (188 KB): application of threading library
 \end{itemize}
+The \CFA compiler builds are picked from git commit history that passed the test suite, and implement the optimizations incrementally:
+\begin{itemize}
+\item
+\#0 is the first working build of new AST data structure
+versus \CFA compiler builds picked from the git commit history that implement the optimizations incrementally:
+\begin{itemize}
+\item
+old resolver
+\item
+\#0 is the first working build of the new AST data structure
 \item
 \#1 implements special symbol table and argument-dependent lookup
 \item
+\#2 implements late assertion satisfaction
+\item
+\#3 implements revised function type representation
+\item
+\#4 skips pruning on expressions with function type (most recent build)
+\end{itemize}
+The old resolver with no memory sharing and none of the optimizations above is also tested.
+\begin{table}
+\#2 implements late assertion-satisfaction
+\item
+\#3 implements revised function-type representation
+\item
+\#4 skips pruning on expressions for function types (most recent build)
+\end{itemize}
+Reading left to right for a test shows the benefit of each optimization on the cost of compilation.
+\begin{table}[htb]
+\centering
 \caption{Compile time of selected files by compiler build, in seconds}
+\label{t:SelectedFileByCompilerBuild}
 \begin{tabular}{|l|r|r|r|r|r|r|}
 \hline
 …
 \end{table}
 \section{Conclusion}
 Over the course of 8 months of active research and development in \CFA type system and compiler algorithm, performance of the reference \CFA compiler, cfa-cc, has been greatly improved, allowing mid-sized \CFA programs to be compiled and built reasonably fast. As there are also ongoing efforts in the team on building a standard library, evaluating the runtime performance, and attempting to incorporate \CFA with existing software written in C, this project is especially meaningful for practical purposes.
 Analysis conducted in the project were based significantly on heuristics and practical evidence, as the theoretical bounds and average cases for the expression resolution problem differ. This approach was difficult at start to follow, with an unacceptably slow compiler, since running the program through debugger and validation tools (\eg @gdb@, @valgrind@) adds another order of magnitude to run time, which was already in minutes. However, near the end of the project, many significant improvements have already been made and new optimizations can be tested immediately. The positive feedback in development cycle benefits the \CFA team as a whole, more than just for the compiler optimizations.
 Some potential issues of the language that may happen frequently in practice have been identified. Due to the time constraint and complex nature of these problems, a handful of them remain unsolved, but some constructive proposals are made. Notably, introducing a local assertion cache in the resolver is a common solution for a few remaining problems, so that should be the focus of work soon.
 The \CFA team are planning on a public alpha release of the language as the compiler performance becomes promising, and other parts of the system, such as a standard library, are also being enhanced. Ideally, the remaining problems should be resolved before release, and the solutions will also be integral to drafting a formal specification.
+Over the course of 8 months of active research and development of the \CFA type system and compiler algorithms, performance of the reference \CFA compiler, cfa-cc, has been greatly improved. Now, mid-sized \CFA programs are compiled reasonably fast. Currently, there are ongoing efforts by the \CFA team to augment the standard library and evaluate its runtime performance, and incorporate \CFA with existing software written in C; therefore this project is especially meaningful for these practical purposes.
+Accomplishing this work was difficult. Analysis conducted in the project is based significantly on heuristics and practical evidence, as the theoretical bounds and average cases for the expression resolution problem differ. As well, the slowness of the initial compiler made attempts to understand why and where problems exist extremely difficult because both debugging and validation tools (\eg @gdb@, @valgrind@, @pref@) further slowed down compilation time. However, by the end of the project, I had found and fixed several significant problems and new optimizations are easier to introduce and test. The reduction in the development cycle benefits the \CFA team as a whole.
+Some potential issues of the language, which happen frequently in practice, have been identified. Due to the time constraint and complex nature of these problems, a handful of them remain unsolved, but some constructive proposals are made. Notably, introducing a local assertion cache in the resolver is a reasonable solution for a few remaining problems, so that should be the focus of future work.
+The \CFA team are planning on a public alpha release of the language as the compiler performance, given my recent improvements, is now useable. Other parts of the system, such as the standard library, have made significant gains due to the speed up in the development cycle. Ideally, the remaining problems should be resolved before release, and the solutions will also be integral to drafting a formal specification.
 \addcontentsline{toc}{section}{\refname}

doc/theses/fangren_yu_COOP_S20/Report.tex

r5869cea	r7b91c0e
17	17	\usepackage[usenames]{color}
18	18	\input{common} % common CFA document macros
19		\usepackage[dvips,plainpages=false,pdfpagelabels,pdfpagemode=UseNone,colorlinks=true,~~pagebackref=true,~~linkcolor=blue,citecolor=blue,urlcolor=blue,pagebackref=true,breaklinks=true]{hyperref}
	19	\usepackage[dvips,plainpages=false,pdfpagelabels,pdfpagemode=UseNone,colorlinks=true,linkcolor=blue,citecolor=blue,urlcolor=blue,pagebackref=true,breaklinks=true]{hyperref}
20	20	\usepackage{breakurl}
21	21	\urlstyle{sf}

driver/cfa.cc

-                      r5869cea
+                      r7b91c0e
 // Created On       : Tue Aug 20 13:44:49 2002
 // Last Modified By : Peter A. Buhr
 // Last Modified On : Tue Nov 17 14:27:28 2020
 // Update Count     : 440
+// Last Modified On : Sat Jan 16 07:30:19 2021
+// Update Count     : 442
 //
 …
                 args[nargs++] = "-no-integrated-cpp";
                 args[nargs++] = "-Wno-deprecated";
+                args[nargs++] = "-Wno-strict-aliasing";                 // casting from one type to another
                 #ifdef HAVE_CAST_FUNCTION_TYPE
                 args[nargs++] = "-Wno-cast-function-type";

libcfa/prelude/builtins.c

-                      r5869cea
+                      r7b91c0e
 // type that wraps a pointer and a destructor-like function - used in generating implicit destructor calls for struct members in user-defined functions
 // Note: needs to occur early, because it is used to generate destructor calls during code generation
 forall(dtype T)
+forall(T &)
 struct __Destructor {
         T * object;
 …
 // defined destructor in the case that non-generated code wants to use __Destructor
 forall(dtype T)
+forall(T &)
 static inline void ^?{}(__Destructor(T) & x) {
         if (x.object && x.dtor) {
 …
 // easy interface into __Destructor's destructor for easy codegen purposes
 extern "C" {
         forall(dtype T)
+        forall(T &)
         static inline void __destroy_Destructor(__Destructor(T) * dtor) {
                 ^(*dtor){};
 …
 void abort( const char fmt[], ... ) __attribute__ (( format(printf, 1, 2), __nothrow__, __leaf__, __noreturn__ ));
 forall(dtype T)
+forall(T &)
 static inline T & identity(T & i) {
         return i;
 …
 static inline void ^?{}($generator &) {}
 trait is_generator(dtype T) {
+trait is_generator(T &) {
       void main(T & this);
       $generator * get_generator(T & this);
 };
 forall(dtype T | is_generator(T))
+forall(T & | is_generator(T))
 static inline T & resume(T & gen) {
         main(gen);
 …
 static inline {
         forall( dtype DT | { DT & ?+=?( DT &, one_t ); } )
+        forall( DT & | { DT & ?+=?( DT &, one_t ); } )
         DT & ++?( DT & x ) { return x += 1; }
         forall( dtype DT | sized(DT) | { void ?{}( DT &, DT ); void ^?{}( DT & ); DT & ?+=?( DT &, one_t ); } )
+        forall( DT & | sized(DT) | { void ?{}( DT &, DT ); void ^?{}( DT & ); DT & ?+=?( DT &, one_t ); } )
         DT & ?++( DT & x ) { DT tmp = x; x += 1; return tmp; }
         forall( dtype DT | { DT & ?-=?( DT &, one_t ); } )
+        forall( DT & | { DT & ?-=?( DT &, one_t ); } )
         DT & --?( DT & x ) { return x -= 1; }
         forall( dtype DT | sized(DT) | { void ?{}( DT &, DT ); void ^?{}( DT & ); DT & ?-=?( DT &, one_t ); } )
+        forall( DT & | sized(DT) | { void ?{}( DT &, DT ); void ^?{}( DT & ); DT & ?-=?( DT &, one_t ); } )
         DT & ?--( DT & x ) { DT tmp = x; x -= 1; return tmp; }
         forall( dtype DT | { int ?!=?( const DT &, zero_t ); } )
+        forall( DT & | { int ?!=?( const DT &, zero_t ); } )
         int !?( const DT & x ) { return !( x != 0 ); }
 } // distribution
 // universal typed pointer constant
 static inline forall( dtype DT ) DT * intptr( uintptr_t addr ) { return (DT *)addr; }
+static inline forall( DT & ) DT * intptr( uintptr_t addr ) { return (DT *)addr; }
 static inline forall( ftype FT ) FT * intptr( uintptr_t addr ) { return (FT *)addr; }
 …
 #define __CFA_EXP_OVERFLOW__()
 static inline forall( otype OT | { void ?{}( OT & this, one_t ); OT ?*?( OT, OT ); } ) {
+static inline forall( OT | { void ?{}( OT & this, one_t ); OT ?*?( OT, OT ); } ) {
         OT ?\?( OT ep, unsigned int y ) { __CFA_EXP__(); }
         OT ?\?( OT ep, unsigned long int y ) { __CFA_EXP__(); }

libcfa/prelude/prelude-gen.cc

r5869cea	r7b91c0e
159	159	int main() {
160	160	cout << "# 2 \"prelude.cfa\" // needed for error messages from this file" << endl;
161		cout << "trait sized(~~dtype T~~) {};" << endl;
	161	cout << "trait sized(T &) {};" << endl;
162	162
163	163	cout << "//////////////////////////" << endl;
…	…
264	264	for (auto cvq : qualifiersPair) {
265	265	for (auto is_vol : { " ", "volatile" }) {
266		cout << "forall(~~dtype DT~~) void ?{}(" << cvq.first << type << " * " << is_vol << " &, " << cvq.second << "DT *);" << endl;
	266	cout << "forall(DT &) void ?{}(" << cvq.first << type << " * " << is_vol << " &, " << cvq.second << "DT *);" << endl;
267	267	}
268	268	}
…	…
279	279	for (auto cvq : qualifiersSingle) {
280	280	for (auto is_vol : { " ", "volatile" }) {
281		cout << "forall(~~dtype DT~~) void ?{}(" << cvq << " DT" << " * " << is_vol << " &);" << endl;
	281	cout << "forall(DT &) void ?{}(" << cvq << " DT" << " * " << is_vol << " &);" << endl;
282	282	}
283	283	for (auto is_vol : { " ", "volatile" }) {
284		cout << "forall(~~dtype DT~~) void ^?{}(" << cvq << " DT" << " * " << is_vol << " &);" << endl;
	284	cout << "forall(DT &) void ^?{}(" << cvq << " DT" << " * " << is_vol << " &);" << endl;
285	285	}
286	286	}
…	…
290	290	for (auto is_vol : { " ", "volatile" }) {
291	291	for (auto cvq : qualifiersSingle) {
292		cout << "forall(~~dtype DT~~) void ?{}( " << cvq << type << " * " << is_vol << " &, zero_t);" << endl;
	292	cout << "forall(DT &) void ?{}( " << cvq << type << " * " << is_vol << " &, zero_t);" << endl;
293	293	}
294	294	}
…	…
317	317	for (auto op : pointerOperators) {
318	318	auto forall = [&op]() {
319		cout << "forall(~~dtype DT~~" << op.sized << ") ";
	319	cout << "forall(DT &" << op.sized << ") ";
320	320	};
321	321	for (auto type : { "DT"/, "void"/ } ) {
…	…
408	408	for (auto is_vol : { " ", "volatile" }) {
409	409	for (auto cvq : qualifiersPair) {
410		cout << "forall(~~dtype DT~~) " << cvq.first << "void * ?=?( " << cvq.first << "void * " << is_vol << " &, " << cvq.second << "DT *);" << endl;
	410	cout << "forall(DT &) " << cvq.first << "void * ?=?( " << cvq.first << "void * " << is_vol << " &, " << cvq.second << "DT *);" << endl;
411	411	}
412	412	for (auto cvq : qualifiersSingle) {
413		cout << "forall(~~dtype DT~~) " << cvq << " DT * ?=?( " << cvq << " DT * " << is_vol << " &, zero_t);" << endl;
	413	cout << "forall(DT &) " << cvq << " DT * ?=?( " << cvq << " DT * " << is_vol << " &, zero_t);" << endl;
414	414	}
415	415	}

libcfa/prelude/prelude.old.cf

-                      r5869cea
+                      r7b91c0e
 // ------------------------------------------------------------
 trait sized(dtype T) {};
+trait sized(T &) {};
 // ------------------------------------------------------------
 …
 long double _Complex    ?--( long double _Complex & ),          ?--( volatile long double _Complex & );
 forall( dtype T | sized(T) ) T *                         ?++(                T *& );
 forall( dtype T | sized(T) ) const T *           ?++( const          T *& );
 forall( dtype T | sized(T) ) volatile T *                ?++(       volatile T *& );
 forall( dtype T | sized(T) ) const volatile T *  ?++( const volatile T *& );
 forall( dtype T | sized(T) ) T *                         ?--(                T *& );
 forall( dtype T | sized(T) ) const T *           ?--( const          T *& );
 forall( dtype T | sized(T) ) volatile T *                ?--(       volatile T *& );
 forall( dtype T | sized(T) ) const volatile T *  ?--( const volatile T *& );
 forall( dtype T | sized(T) ) T &                 ?[?](                T *,          ptrdiff_t );
 forall( dtype T | sized(T) ) const T &   ?[?]( const          T *,          ptrdiff_t );
 forall( dtype T | sized(T) ) volatile T &        ?[?](       volatile T *,          ptrdiff_t );
 forall( dtype T | sized(T) ) const volatile T & ?[?]( const volatile T *,           ptrdiff_t );
 forall( dtype T | sized(T) ) T &                 ?[?](          ptrdiff_t,                T * );
 forall( dtype T | sized(T) ) const T &   ?[?](          ptrdiff_t, const          T * );
 forall( dtype T | sized(T) ) volatile T &        ?[?](          ptrdiff_t,       volatile T * );
 forall( dtype T | sized(T) ) const volatile T & ?[?](           ptrdiff_t, const volatile T * );
+forall( T & | sized(T) ) T *                     ?++(                T *& );
+forall( T & | sized(T) ) const T *               ?++( const          T *& );
+forall( T & | sized(T) ) volatile T *            ?++(       volatile T *& );
+forall( T & | sized(T) ) const volatile T *      ?++( const volatile T *& );
+forall( T & | sized(T) ) T *                     ?--(                T *& );
+forall( T & | sized(T) ) const T *               ?--( const          T *& );
+forall( T & | sized(T) ) volatile T *            ?--(       volatile T *& );
+forall( T & | sized(T) ) const volatile T *      ?--( const volatile T *& );
+forall( T & | sized(T) ) T &             ?[?](                T *,          ptrdiff_t );
+forall( T & | sized(T) ) const T &       ?[?]( const          T *,          ptrdiff_t );
+forall( T & | sized(T) ) volatile T &    ?[?](       volatile T *,          ptrdiff_t );
+forall( T & | sized(T) ) const volatile T & ?[?]( const volatile T *,       ptrdiff_t );
+forall( T & | sized(T) ) T &             ?[?](          ptrdiff_t,                T * );
+forall( T & | sized(T) ) const T &       ?[?](          ptrdiff_t, const          T * );
+forall( T & | sized(T) ) volatile T &    ?[?](          ptrdiff_t,       volatile T * );
+forall( T & | sized(T) ) const volatile T & ?[?](               ptrdiff_t, const volatile T * );
 // ------------------------------------------------------------
 …
 long double _Complex    ++?( long double _Complex & ),          --?( long double _Complex & );
 forall( dtype T | sized(T) ) T *                         ++?(                T *& );
 forall( dtype T | sized(T) ) const T *           ++?( const          T *& );
 forall( dtype T | sized(T) ) volatile T *                ++?(       volatile T *& );
 forall( dtype T | sized(T) ) const volatile T *  ++?( const volatile T *& );
 forall( dtype T | sized(T) ) T *                         --?(                T *& );
 forall( dtype T | sized(T) ) const T *           --?( const          T *& );
 forall( dtype T | sized(T) ) volatile T *                --?(       volatile T *& );
 forall( dtype T | sized(T) ) const volatile T *  --?( const volatile T *& );
 forall( dtype T | sized(T) ) T &                 *?(                 T * );
 forall( dtype T | sized(T) ) const T &           *?( const           T * );
 forall( dtype T | sized(T) ) volatile T &        *?(       volatile  T * );
 forall( dtype T | sized(T) ) const volatile T & *?( const volatile  T * );
+forall( T & | sized(T) ) T *                     ++?(                T *& );
+forall( T & | sized(T) ) const T *               ++?( const          T *& );
+forall( T & | sized(T) ) volatile T *            ++?(       volatile T *& );
+forall( T & | sized(T) ) const volatile T *      ++?( const volatile T *& );
+forall( T & | sized(T) ) T *                     --?(                T *& );
+forall( T & | sized(T) ) const T *               --?( const          T *& );
+forall( T & | sized(T) ) volatile T *            --?(       volatile T *& );
+forall( T & | sized(T) ) const volatile T *      --?( const volatile T *& );
+forall( T & | sized(T) ) T &             *?(                 T * );
+forall( T & | sized(T) ) const T &               *?( const           T * );
+forall( T & | sized(T) ) volatile T &    *?(       volatile  T * );
+forall( T & | sized(T) ) const volatile T & *?( const volatile  T * );
 forall( ftype FT ) FT &          *?( FT * );
 …
                 !?( float _Complex ),           !?( double _Complex ),          !?( long double _Complex );
 forall( dtype DT ) int !?(                DT * );
 forall( dtype DT ) int !?( const          DT * );
 forall( dtype DT ) int !?(       volatile DT * );
 forall( dtype DT ) int !?( const volatile DT * );
+forall( DT & ) int !?(                DT * );
+forall( DT & ) int !?( const          DT * );
+forall( DT & ) int !?(       volatile DT * );
+forall( DT & ) int !?( const volatile DT * );
 forall( ftype FT ) int !?( FT * );
 …
 long double _Complex    ?+?( long double _Complex, long double _Complex ),      ?-?( long double _Complex, long double _Complex );
 forall( dtype T | sized(T) ) T *                ?+?(                T *,          ptrdiff_t );
 forall( dtype T | sized(T) ) T *                ?+?(          ptrdiff_t,                T * );
 forall( dtype T | sized(T) ) const T *          ?+?( const          T *,          ptrdiff_t );
 forall( dtype T | sized(T) ) const T *          ?+?(          ptrdiff_t, const          T * );
 forall( dtype T | sized(T) ) volatile T *       ?+?(       volatile T *,          ptrdiff_t );
 forall( dtype T | sized(T) ) volatile T *       ?+?(          ptrdiff_t,       volatile T * );
 forall( dtype T | sized(T) ) const volatile T * ?+?( const volatile T *,          ptrdiff_t );
 forall( dtype T | sized(T) ) const volatile T * ?+?(          ptrdiff_t, const volatile T * );
 forall( dtype T | sized(T) ) T *                ?-?(                T *,          ptrdiff_t );
 forall( dtype T | sized(T) ) const T *          ?-?( const          T *,          ptrdiff_t );
 forall( dtype T | sized(T) ) volatile T *       ?-?(       volatile T *,          ptrdiff_t );
 forall( dtype T | sized(T) ) const volatile T * ?-?( const volatile T *,          ptrdiff_t );
 forall( dtype T | sized(T) ) ptrdiff_t          ?-?( const volatile T *, const volatile T * );
+forall( T & | sized(T) ) T *            ?+?(                T *,          ptrdiff_t );
+forall( T & | sized(T) ) T *            ?+?(          ptrdiff_t,                T * );
+forall( T & | sized(T) ) const T *              ?+?( const          T *,          ptrdiff_t );
+forall( T & | sized(T) ) const T *              ?+?(          ptrdiff_t, const          T * );
+forall( T & | sized(T) ) volatile T *   ?+?(       volatile T *,          ptrdiff_t );
+forall( T & | sized(T) ) volatile T *   ?+?(          ptrdiff_t,       volatile T * );
+forall( T & | sized(T) ) const volatile T *     ?+?( const volatile T *,          ptrdiff_t );
+forall( T & | sized(T) ) const volatile T *     ?+?(          ptrdiff_t, const volatile T * );
+forall( T & | sized(T) ) T *            ?-?(                T *,          ptrdiff_t );
+forall( T & | sized(T) ) const T *              ?-?( const          T *,          ptrdiff_t );
+forall( T & | sized(T) ) volatile T *   ?-?(       volatile T *,          ptrdiff_t );
+forall( T & | sized(T) ) const volatile T *     ?-?( const volatile T *,          ptrdiff_t );
+forall( T & | sized(T) ) ptrdiff_t              ?-?( const volatile T *, const volatile T * );
 // ------------------------------------------------------------
 …
            ?>?( long double, long double ),                             ?>=?( long double, long double );
 forall( dtype DT ) signed int ?<?(                 DT *,                DT * );
 forall( dtype DT ) signed int ?<?(  const          DT *, const          DT * );
 forall( dtype DT ) signed int ?<?(        volatile DT *,       volatile DT * );
 forall( dtype DT ) signed int ?<?(  const volatile DT *, const volatile DT * );
 forall( dtype DT ) signed int ?>?(                 DT *,                DT * );
 forall( dtype DT ) signed int ?>?(  const          DT *, const          DT * );
 forall( dtype DT ) signed int ?>?(        volatile DT *,       volatile DT * );
 forall( dtype DT ) signed int ?>?(  const volatile DT *, const volatile DT * );
 forall( dtype DT ) signed int ?<=?(                 DT *,                DT * );
 forall( dtype DT ) signed int ?<=?(  const          DT *, const          DT * );
 forall( dtype DT ) signed int ?<=?(        volatile DT *,       volatile DT * );
 forall( dtype DT ) signed int ?<=?( const volatile DT *, const volatile DT * );
 forall( dtype DT ) signed int ?>=?(                 DT *,                DT * );
 forall( dtype DT ) signed int ?>=?(  const          DT *, const          DT * );
 forall( dtype DT ) signed int ?>=?(        volatile DT *,       volatile DT * );
 forall( dtype DT ) signed int ?>=?( const volatile DT *, const volatile DT * );
+forall( DT & ) signed int ?<?(                 DT *,                DT * );
+forall( DT & ) signed int ?<?(  const          DT *, const          DT * );
+forall( DT & ) signed int ?<?(        volatile DT *,       volatile DT * );
+forall( DT & ) signed int ?<?(  const volatile DT *, const volatile DT * );
+forall( DT & ) signed int ?>?(                 DT *,                DT * );
+forall( DT & ) signed int ?>?(  const          DT *, const          DT * );
+forall( DT & ) signed int ?>?(        volatile DT *,       volatile DT * );
+forall( DT & ) signed int ?>?(  const volatile DT *, const volatile DT * );
+forall( DT & ) signed int ?<=?(                 DT *,                DT * );
+forall( DT & ) signed int ?<=?(  const          DT *, const          DT * );
+forall( DT & ) signed int ?<=?(        volatile DT *,       volatile DT * );
+forall( DT & ) signed int ?<=?( const volatile DT *, const volatile DT * );
+forall( DT & ) signed int ?>=?(                 DT *,                DT * );
+forall( DT & ) signed int ?>=?(  const          DT *, const          DT * );
+forall( DT & ) signed int ?>=?(        volatile DT *,       volatile DT * );
+forall( DT & ) signed int ?>=?( const volatile DT *, const volatile DT * );
 // ------------------------------------------------------------
 …
 signed int ?==?( one_t, one_t ),                                                        ?!=?( one_t, one_t );
 forall( dtype DT ) signed int ?==?(                DT *,                DT * );
 forall( dtype DT ) signed int ?==?( const          DT *, const          DT * );
 forall( dtype DT ) signed int ?==?(       volatile DT *,       volatile DT * );
 forall( dtype DT ) signed int ?==?( const volatile DT *, const volatile DT * );
+forall( DT & ) signed int ?==?(            DT *,                DT * );
+forall( DT & ) signed int ?==?( const      DT *, const          DT * );
+forall( DT & ) signed int ?==?(       volatile DT *,       volatile DT * );
+forall( DT & ) signed int ?==?( const volatile DT *, const volatile DT * );
 forall( ftype FT ) signed int ?==?( FT *, FT * );
 forall( dtype DT ) signed int ?!=?(                DT *,                DT * );
 forall( dtype DT ) signed int ?!=?( const          DT *, const          DT * );
 forall( dtype DT ) signed int ?!=?(       volatile DT *,       volatile DT * );
 forall( dtype DT ) signed int ?!=?( const volatile DT *, const volatile DT * );
+forall( DT & ) signed int ?!=?(            DT *,                DT * );
+forall( DT & ) signed int ?!=?( const      DT *, const          DT * );
+forall( DT & ) signed int ?!=?(       volatile DT *,       volatile DT * );
+forall( DT & ) signed int ?!=?( const volatile DT *, const volatile DT * );
 forall( ftype FT ) signed int ?!=?( FT *, FT * );
 …
 forall( ftype FT ) FT *                 ?=?( FT *&, FT * );
 forall( ftype FT ) FT *                 ?=?( FT * volatile &, FT * );
 forall( dtype DT ) DT *                 ?=?(                 DT *          &,                   DT * );
 forall( dtype DT ) DT *                 ?=?(                 DT * volatile &,                   DT * );
 forall( dtype DT ) const DT *           ?=?( const           DT *          &,                   DT * );
 forall( dtype DT ) const DT *           ?=?( const           DT * volatile &,                   DT * );
 forall( dtype DT ) const DT *           ?=?( const           DT *          &, const             DT * );
 forall( dtype DT ) const DT *           ?=?( const           DT * volatile &, const             DT * );
 forall( dtype DT ) volatile DT *        ?=?(       volatile  DT *          &,                   DT * );
 forall( dtype DT ) volatile DT *        ?=?(       volatile  DT * volatile &,                   DT * );
 forall( dtype DT ) volatile DT *        ?=?(       volatile  DT *          &,       volatile    DT * );
 forall( dtype DT ) volatile DT *        ?=?(       volatile  DT * volatile &,       volatile    DT * );
 forall( dtype DT ) const volatile DT *  ?=?( const volatile  DT *          &,                   DT * );
 forall( dtype DT ) const volatile DT *  ?=?( const volatile  DT * volatile &,                   DT * );
 forall( dtype DT ) const volatile DT *  ?=?( const volatile  DT *          &, const             DT * );
 forall( dtype DT ) const volatile DT *  ?=?( const volatile  DT * volatile &, const             DT * );
 forall( dtype DT ) const volatile DT *  ?=?( const volatile  DT *          &,       volatile    DT * );
 forall( dtype DT ) const volatile DT *  ?=?( const volatile  DT * volatile &,       volatile    DT * );
 forall( dtype DT ) const volatile DT *  ?=?( const volatile  DT *          &, const volatile    DT * );
 forall( dtype DT ) const volatile DT *  ?=?( const volatile  DT * volatile &, const volatile    DT * );
 forall( dtype DT ) void *                ?=?(                void *          &,                 DT * );
 forall( dtype DT ) void *                ?=?(                void * volatile &,                 DT * );
 forall( dtype DT ) const void *          ?=?( const          void *          &,                 DT * );
 forall( dtype DT ) const void *          ?=?( const          void * volatile &,                 DT * );
 forall( dtype DT ) const void *          ?=?( const          void *          &, const           DT * );
 forall( dtype DT ) const void *          ?=?( const          void * volatile &, const           DT * );
 forall( dtype DT ) volatile void *       ?=?(       volatile void *          &,                 DT * );
 forall( dtype DT ) volatile void *       ?=?(       volatile void * volatile &,                 DT * );
 forall( dtype DT ) volatile void *       ?=?(       volatile void *          &,       volatile  DT * );
 forall( dtype DT ) volatile void *       ?=?(       volatile void * volatile &,       volatile  DT * );
 forall( dtype DT ) const volatile void * ?=?( const volatile void *          &,                 DT * );
 forall( dtype DT ) const volatile void * ?=?( const volatile void * volatile &,                 DT * );
 forall( dtype DT ) const volatile void * ?=?( const volatile void *          &, const           DT * );
 forall( dtype DT ) const volatile void * ?=?( const volatile void * volatile &, const           DT * );
 forall( dtype DT ) const volatile void * ?=?( const volatile void *          &,       volatile  DT * );
 forall( dtype DT ) const volatile void * ?=?( const volatile void * volatile &,       volatile  DT * );
 forall( dtype DT ) const volatile void * ?=?( const volatile void *          &, const volatile  DT * );
 forall( dtype DT ) const volatile void * ?=?( const volatile void * volatile &, const volatile  DT * );
+forall( ftyep FT ) FT *                 ?=?( FT * volatile &, FT * );
+forall( DT & ) DT *                     ?=?(                 DT *          &,                   DT * );
+forall( DT & ) DT *                     ?=?(                 DT * volatile &,                   DT * );
+forall( DT & ) const DT *               ?=?( const           DT *          &,                   DT * );
+forall( DT & ) const DT *               ?=?( const           DT * volatile &,                   DT * );
+forall( DT & ) const DT *               ?=?( const           DT *          &, const             DT * );
+forall( DT & ) const DT *               ?=?( const           DT * volatile &, const             DT * );
+forall( DT & ) volatile DT *    ?=?(       volatile  DT *          &,                   DT * );
+forall( DT & ) volatile DT *    ?=?(       volatile  DT * volatile &,                   DT * );
+forall( DT & ) volatile DT *    ?=?(       volatile  DT *          &,       volatile    DT * );
+forall( DT & ) volatile DT *    ?=?(       volatile  DT * volatile &,       volatile    DT * );
+forall( DT & ) const volatile DT *      ?=?( const volatile  DT *          &,                   DT * );
+forall( DT & ) const volatile DT *  ?=?( const volatile  DT * volatile &,                       DT * );
+forall( DT & ) const volatile DT *  ?=?( const volatile  DT *      &, const             DT * );
+forall( DT & ) const volatile DT *  ?=?( const volatile  DT * volatile &, const         DT * );
+forall( DT & ) const volatile DT *  ?=?( const volatile  DT *      &,       volatile    DT * );
+forall( DT & ) const volatile DT *  ?=?( const volatile  DT * volatile &,           volatile    DT * );
+forall( DT & ) const volatile DT *  ?=?( const volatile  DT *      &, const volatile    DT * );
+forall( DT & ) const volatile DT *  ?=?( const volatile  DT * volatile &, const volatile        DT * );
+forall( DT & ) void *            ?=?(                void *          &,                 DT * );
+forall( DT & ) void *            ?=?(                void * volatile &,                 DT * );
+forall( DT & ) const void *              ?=?( const          void *          &,                 DT * );
+forall( DT & ) const void *              ?=?( const          void * volatile &,                 DT * );
+forall( DT & ) const void *              ?=?( const          void *          &, const           DT * );
+forall( DT & ) const void *              ?=?( const          void * volatile &, const           DT * );
+forall( DT & ) volatile void *   ?=?(       volatile void *          &,                 DT * );
+forall( DT & ) volatile void *   ?=?(       volatile void * volatile &,                 DT * );
+forall( DT & ) volatile void *   ?=?(       volatile void *          &,       volatile  DT * );
+forall( DT & ) volatile void *   ?=?(       volatile void * volatile &,       volatile  DT * );
+forall( DT & ) const volatile void * ?=?( const volatile void *      &,                 DT * );
+forall( DT & ) const volatile void * ?=?( const volatile void * volatile &,                     DT * );
+forall( DT & ) const volatile void * ?=?( const volatile void *      &, const           DT * );
+forall( DT & ) const volatile void * ?=?( const volatile void * volatile &, const               DT * );
+forall( DT & ) const volatile void * ?=?( const volatile void *      &,       volatile  DT * );
+forall( DT & ) const volatile void * ?=?( const volatile void * volatile &,           volatile  DT * );
+forall( DT & ) const volatile void * ?=?( const volatile void *      &, const volatile  DT * );
+forall( DT & ) const volatile void * ?=?( const volatile void * volatile &, const volatile      DT * );
 //forall( dtype DT ) DT *                       ?=?(                DT *          &, zero_t );
 //forall( dtype DT ) DT *                       ?=?(                DT * volatile &, zero_t );
 forall( dtype DT ) const DT *           ?=?( const          DT *          &, zero_t );
 forall( dtype DT ) const DT *           ?=?( const          DT * volatile &, zero_t );
+forall( DT & ) const DT *               ?=?( const          DT *          &, zero_t );
+forall( DT & ) const DT *               ?=?( const          DT * volatile &, zero_t );
 //forall( dtype DT ) volatile DT *      ?=?( volatile       DT *          &, zero_t );
 //forall( dtype DT ) volatile DT *      ?=?( volatile       DT * volatile &, zero_t );
 forall( dtype DT ) const volatile DT *  ?=?( const volatile DT *          &, zero_t );
 forall( dtype DT ) const volatile DT *  ?=?( const volatile DT * volatile &, zero_t );
+forall( DT & ) const volatile DT *      ?=?( const volatile DT *          &, zero_t );
+forall( DT & ) const volatile DT *      ?=?( const volatile DT * volatile &, zero_t );
 forall( ftype FT ) FT *                 ?=?( FT *          &, zero_t );
 forall( ftype FT ) FT *                 ?=?( FT * volatile &, zero_t );
 forall( dtype T | sized(T) ) T *                ?+=?(                T *          &, ptrdiff_t );
 forall( dtype T | sized(T) ) T *                ?+=?(                T * volatile &, ptrdiff_t );
 forall( dtype T | sized(T) ) const T *          ?+=?( const          T *          &, ptrdiff_t );
 forall( dtype T | sized(T) ) const T *          ?+=?( const          T * volatile &, ptrdiff_t );
 forall( dtype T | sized(T) ) volatile T *       ?+=?(       volatile T *          &, ptrdiff_t );
 forall( dtype T | sized(T) ) volatile T *       ?+=?(       volatile T * volatile &, ptrdiff_t );
 forall( dtype T | sized(T) ) const volatile T * ?+=?( const volatile T *          &, ptrdiff_t );
 forall( dtype T | sized(T) ) const volatile T * ?+=?( const volatile T * volatile &, ptrdiff_t );
 forall( dtype T | sized(T) ) T *                ?-=?(                T *          &, ptrdiff_t );
 forall( dtype T | sized(T) ) T *                ?-=?(                T * volatile &, ptrdiff_t );
 forall( dtype T | sized(T) ) const T *          ?-=?( const          T *          &, ptrdiff_t );
 forall( dtype T | sized(T) ) const T *          ?-=?( const          T * volatile &, ptrdiff_t );
 forall( dtype T | sized(T) ) volatile T *       ?-=?(       volatile T *          &, ptrdiff_t );
 forall( dtype T | sized(T) ) volatile T *       ?-=?(       volatile T * volatile &, ptrdiff_t );
 forall( dtype T | sized(T) ) const volatile T * ?-=?( const volatile T *          &, ptrdiff_t );
 forall( dtype T | sized(T) ) const volatile T * ?-=?( const volatile T * volatile &, ptrdiff_t );
+forall( T & | sized(T) ) T *            ?+=?(                T *          &, ptrdiff_t );
+forall( T & | sized(T) ) T *            ?+=?(                T * volatile &, ptrdiff_t );
+forall( T & | sized(T) ) const T *              ?+=?( const          T *          &, ptrdiff_t );
+forall( T & | sized(T) ) const T *              ?+=?( const          T * volatile &, ptrdiff_t );
+forall( T & | sized(T) ) volatile T *   ?+=?(       volatile T *          &, ptrdiff_t );
+forall( T & | sized(T) ) volatile T *   ?+=?(       volatile T * volatile &, ptrdiff_t );
+forall( T & | sized(T) ) const volatile T *     ?+=?( const volatile T *          &, ptrdiff_t );
+forall( T & | sized(T) ) const volatile T *     ?+=?( const volatile T * volatile &, ptrdiff_t );
+forall( T & | sized(T) ) T *            ?-=?(                T *          &, ptrdiff_t );
+forall( T & | sized(T) ) T *            ?-=?(                T * volatile &, ptrdiff_t );
+forall( T & | sized(T) ) const T *              ?-=?( const          T *          &, ptrdiff_t );
+forall( T & | sized(T) ) const T *              ?-=?( const          T * volatile &, ptrdiff_t );
+forall( T & | sized(T) ) volatile T *   ?-=?(       volatile T *          &, ptrdiff_t );
+forall( T & | sized(T) ) volatile T *   ?-=?(       volatile T * volatile &, ptrdiff_t );
+forall( T & | sized(T) ) const volatile T *     ?-=?( const volatile T *          &, ptrdiff_t );
+forall( T & | sized(T) ) const volatile T *     ?-=?( const volatile T * volatile &, ptrdiff_t );
 _Bool                   ?=?( _Bool &, _Bool ),                                  ?=?( volatile _Bool &, _Bool );
 …
 forall( ftype FT ) void ?{}( FT * volatile &, FT * );
 forall( dtype DT ) void ?{}(                 DT *          &,                   DT * );
 forall( dtype DT ) void ?{}( const           DT *          &,                   DT * );
 forall( dtype DT ) void ?{}( const           DT *          &, const             DT * );
 forall( dtype DT ) void ?{}(       volatile  DT *          &,                   DT * );
 forall( dtype DT ) void ?{}(       volatile  DT *          &,       volatile    DT * );
 forall( dtype DT ) void ?{}( const volatile  DT *          &,                   DT * );
 forall( dtype DT ) void ?{}( const volatile  DT *          &, const             DT * );
 forall( dtype DT ) void ?{}( const volatile  DT *          &,       volatile    DT * );
 forall( dtype DT ) void ?{}( const volatile  DT *          &, const volatile    DT * );
 forall( dtype DT ) void ?{}(                 void *          &,                 DT * );
 forall( dtype DT ) void ?{}( const           void *          &,                 DT * );
 forall( dtype DT ) void ?{}( const           void *          &, const           DT * );
 forall( dtype DT ) void ?{}(        volatile void *          &,                 DT * );
 forall( dtype DT ) void ?{}(        volatile void *          &,       volatile  DT * );
 forall( dtype DT ) void ?{}( const volatile void *           &,                 DT * );
 forall( dtype DT ) void ?{}( const volatile void *           &, const           DT * );
 forall( dtype DT ) void ?{}( const volatile void *           &,       volatile  DT * );
 forall( dtype DT ) void ?{}( const volatile void *           &, const volatile  DT * );
+forall( DT & ) void ?{}(                     DT *          &,                   DT * );
+forall( DT & ) void ?{}( const       DT *          &,                   DT * );
+forall( DT & ) void ?{}( const       DT *          &, const             DT * );
+forall( DT & ) void ?{}(           volatile  DT *          &,                   DT * );
+forall( DT & ) void ?{}(           volatile  DT *          &,       volatile    DT * );
+forall( DT & ) void ?{}( const volatile  DT *      &,                   DT * );
+forall( DT & ) void ?{}( const volatile  DT *      &, const             DT * );
+forall( DT & ) void ?{}( const volatile  DT *      &,       volatile    DT * );
+forall( DT & ) void ?{}( const volatile  DT *      &, const volatile    DT * );
+forall( DT & ) void ?{}(                     void *          &,                 DT * );
+forall( DT & ) void ?{}( const       void *          &,                 DT * );
+forall( DT & ) void ?{}( const       void *          &, const           DT * );
+forall( DT & ) void ?{}(            volatile void *          &,                 DT * );
+forall( DT & ) void ?{}(            volatile void *          &,       volatile  DT * );
+forall( DT & ) void ?{}( const volatile void *       &,                 DT * );
+forall( DT & ) void ?{}( const volatile void *       &, const           DT * );
+forall( DT & ) void ?{}( const volatile void *       &,       volatile  DT * );
+forall( DT & ) void ?{}( const volatile void *       &, const volatile  DT * );
 //forall( dtype DT ) void ?{}(              DT *          &, zero_t );
 //forall( dtype DT ) void ?{}(              DT * volatile &, zero_t );
 forall( dtype DT ) void ?{}( const          DT *          &, zero_t );
+forall( DT & ) void ?{}( const      DT *          &, zero_t );
 //forall( dtype DT ) void ?{}( volatile     DT *          &, zero_t );
 //forall( dtype DT ) void ?{}( volatile     DT * volatile &, zero_t );
 forall( dtype DT ) void ?{}( const volatile DT *          &, zero_t );
+forall( DT & ) void ?{}( const volatile DT *      &, zero_t );
 forall( ftype FT ) void ?{}( FT *          &, zero_t );
 …
 forall( ftype FT ) void ?{}( FT *          & );
 forall( dtype DT ) void ?{}(                 DT *          &);
 forall( dtype DT ) void ?{}( const           DT *          &);
 forall( dtype DT ) void ?{}(       volatile  DT *          &);
 forall( dtype DT ) void ?{}( const volatile  DT *          &);
+forall( DT & ) void     ?{}(                 DT *          &);
+forall( DT & ) void     ?{}( const           DT *          &);
+forall( DT & ) void     ?{}(       volatile  DT *          &);
+forall( DT & ) void ?{}( const volatile  DT *      &);
 void    ?{}(                void *          &);
 …
 forall( ftype FT ) void ^?{}( FT *         & );
 forall( dtype DT ) void ^?{}(                DT *          &);
 forall( dtype DT ) void ^?{}( const          DT *          &);
 forall( dtype DT ) void ^?{}(      volatile  DT *          &);
 forall( dtype DT ) void ^?{}( const volatile  DT *         &);
+forall( DT & ) void     ^?{}(                DT *          &);
+forall( DT & ) void     ^?{}( const          DT *          &);
+forall( DT & ) void     ^?{}(      volatile  DT *          &);
+forall( DT & ) void ^?{}( const volatile  DT *     &);
 void ^?{}(                  void *          &);

libcfa/prelude/sync-builtins.cf

-                      r5869cea
+                      r7b91c0e
 _Bool __sync_bool_compare_and_swap(volatile unsigned __int128 *, unsigned __int128, unsigned __int128,...);
 #endif
 forall(dtype T) _Bool __sync_bool_compare_and_swap(T * volatile *, T *, T*, ...);
+forall(T &) _Bool __sync_bool_compare_and_swap(T * volatile *, T *, T*, ...);
 char __sync_val_compare_and_swap(volatile char *, char, char,...);
 …
 unsigned __int128 __sync_val_compare_and_swap(volatile unsigned __int128 *, unsigned __int128, unsigned __int128,...);
 #endif
 forall(dtype T) T * __sync_val_compare_and_swap(T * volatile *, T *, T*,...);
+forall(T &) T * __sync_val_compare_and_swap(T * volatile *, T *, T*,...);
 char __sync_lock_test_and_set(volatile char *, char,...);
 …
 void __atomic_exchange(volatile unsigned __int128 *, volatile unsigned __int128 *, volatile unsigned __int128 *, int);
 #endif
 forall(dtype T) T * __atomic_exchange_n(T * volatile *, T *, int);
 forall(dtype T) void __atomic_exchange(T * volatile *, T * volatile *, T * volatile *, int);
+forall(T &) T * __atomic_exchange_n(T * volatile *, T *, int);
+forall(T &) void __atomic_exchange(T * volatile *, T * volatile *, T * volatile *, int);
 _Bool __atomic_load_n(const volatile _Bool *, int);
 …
 void __atomic_load(const volatile unsigned __int128 *, volatile unsigned __int128 *, int);
 #endif
 forall(dtype T) T * __atomic_load_n(T * const volatile *, int);
 forall(dtype T) void __atomic_load(T * const volatile *, T **, int);
+forall(T &) T * __atomic_load_n(T * const volatile *, int);
+forall(T &) void __atomic_load(T * const volatile *, T **, int);
 _Bool __atomic_compare_exchange_n(volatile char *, char *, char, _Bool, int, int);
 …
 _Bool __atomic_compare_exchange   (volatile unsigned __int128 *, unsigned __int128 *, unsigned __int128 *, _Bool, int, int);
 #endif
 forall(dtype T) _Bool __atomic_compare_exchange_n (T * volatile *, T **, T*, _Bool, int, int);
 forall(dtype T) _Bool __atomic_compare_exchange   (T * volatile *, T **, T**, _Bool, int, int);
+forall(T &) _Bool __atomic_compare_exchange_n (T * volatile *, T **, T*, _Bool, int, int);
+forall(T &) _Bool __atomic_compare_exchange   (T * volatile *, T **, T**, _Bool, int, int);
 void __atomic_store_n(volatile _Bool *, _Bool, int);
 …
 void __atomic_store(volatile unsigned __int128 *, unsigned __int128 *, int);
 #endif
 forall(dtype T) void __atomic_store_n(T * volatile *, T *, int);
 forall(dtype T) void __atomic_store(T * volatile *, T **, int);
+forall(T &) void __atomic_store_n(T * volatile *, T *, int);
+forall(T &) void __atomic_store(T * volatile *, T **, int);
 char __atomic_add_fetch  (volatile char *, char, int);

libcfa/src/bitmanip.hfa

-                      r5869cea
+                      r7b91c0e
         unsigned long long int floor2( unsigned long long int n, unsigned long long int align ) { verify( is_pow2( align ) ); return n & -align; }
         // forall( otype T | { T ?&?( T, T ); T -?( T ); } )
+        // forall( T | { T ?&?( T, T ); T -?( T ); } )
         // T floor2( T n, T align ) { verify( is_pow2( align ) ); return n & -align; }
 …
         unsigned long long int ceiling2( unsigned long long int n, unsigned long long int align ) { verify( is_pow2( align ) ); return -floor2( -n, align ); }
         // forall( otype T | { T floor2( T, T ); T -?( T ); } )
+        // forall( T | { T floor2( T, T ); T -?( T ); } )
         // T ceiling2( T n, T align ) { verify( is_pow2( align ) ); return -floor2( -n, align ); }
 } // distribution

libcfa/src/bits/algorithm.hfa

-                      r5869cea
+                      r7b91c0e
 #ifdef SAFE_SORT
 forall( otype T | {  int ?<?( T, T ); int ?>?( T, T ); } ) static inline void __libcfa_small_sort2( T * arr );
 forall( otype T | {  int ?<?( T, T ); int ?>?( T, T ); } ) static inline void __libcfa_small_sort3( T * arr );
 forall( otype T | {  int ?<?( T, T ); int ?>?( T, T ); } ) static inline void __libcfa_small_sort4( T * arr );
 forall( otype T | {  int ?<?( T, T ); int ?>?( T, T ); } ) static inline void __libcfa_small_sort5( T * arr );
 forall( otype T | {  int ?<?( T, T ); int ?>?( T, T ); } ) static inline void __libcfa_small_sort6( T * arr );
 forall( otype T | {  int ?<?( T, T ); int ?>?( T, T ); } ) static inline void __libcfa_small_sortN( T * arr, size_t dim );
+forall( T | {  int ?<?( T, T ); int ?>?( T, T ); } ) static inline void __libcfa_small_sort2( T * arr );
+forall( T | {  int ?<?( T, T ); int ?>?( T, T ); } ) static inline void __libcfa_small_sort3( T * arr );
+forall( T | {  int ?<?( T, T ); int ?>?( T, T ); } ) static inline void __libcfa_small_sort4( T * arr );
+forall( T | {  int ?<?( T, T ); int ?>?( T, T ); } ) static inline void __libcfa_small_sort5( T * arr );
+forall( T | {  int ?<?( T, T ); int ?>?( T, T ); } ) static inline void __libcfa_small_sort6( T * arr );
+forall( T | {  int ?<?( T, T ); int ?>?( T, T ); } ) static inline void __libcfa_small_sortN( T * arr, size_t dim );
 forall( otype T | {  int ?<?( T, T ); int ?>?( T, T ); } )
+forall( T | {  int ?<?( T, T ); int ?>?( T, T ); } )
 static inline void __libcfa_small_sort( T * arr, size_t dim ) {
         switch( dim ) {
 …
 #define SWAP(x,y) { T a = min(arr[x], arr[y]); T b = max(arr[x], arr[y]); arr[x] = a; arr[y] = b;}
 forall( otype T | {  int ?<?( T, T ); int ?>?( T, T ); } )
+forall( T | {  int ?<?( T, T ); int ?>?( T, T ); } )
 static inline void __libcfa_small_sort2( T * arr ) {
         SWAP(0, 1);
+}
 forall( otype T | {  int ?<?( T, T ); int ?>?( T, T ); } )
+forall( T | {  int ?<?( T, T ); int ?>?( T, T ); } )
 static inline void __libcfa_small_sort3( T * arr ) {
         SWAP(1, 2);
 …
+}
 forall( otype T | {  int ?<?( T, T ); int ?>?( T, T ); } )
+forall( T | {  int ?<?( T, T ); int ?>?( T, T ); } )
 static inline void __libcfa_small_sort4( T * arr ) {
         SWAP(0, 1);
 …
+}
 forall( otype T | {  int ?<?( T, T ); int ?>?( T, T ); } )
+forall( T | {  int ?<?( T, T ); int ?>?( T, T ); } )
 static inline void __libcfa_small_sort5( T * arr ) {
         SWAP(0, 1);
 …
+}
 forall( otype T | {  int ?<?( T, T ); int ?>?( T, T ); } )
+forall( T | {  int ?<?( T, T ); int ?>?( T, T ); } )
 static inline void __libcfa_small_sort6( T * arr ) {
         SWAP(1, 2);
 …
+}
 forall( otype T | {  int ?<?( T, T ); int ?>?( T, T ); } )
+forall( T | {  int ?<?( T, T ); int ?>?( T, T ); } )
 static inline void __libcfa_small_sortN( T * arr, size_t dim ) {
         int i, j;
 …
 static inline void __libcfa_small_sortN( void* * arr, size_t dim );
 forall( dtype T )
+forall( T & )
 static inline void __libcfa_small_sort( T* * arr, size_t dim ) {
         switch( dim ) {

libcfa/src/bits/collection.hfa

-                      r5869cea
+                      r7b91c0e
         // // wrappers to make Collection have T
         // forall( dtype T ) {
+        // forall( T & ) {
         //      T *& Next( T * n ) {
         //              return (T *)Next( (Colable *)n );
 …
 } // distribution
 forall( dtype T | { T *& Next ( T * ); } ) {
+static inline forall( T & | { T *& Next ( T * ); } ) {
         bool listed( T * n ) {
                 return Next( n ) != 0p;
 …
         } // post: elts = null
         forall( dtype T ) {
+        forall( T & ) {
                 T * Curr( ColIter & ci ) with( ci ) {
                         return (T *)curr;

libcfa/src/bits/containers.hfa

-                      r5869cea
+                      r7b91c0e
 #ifdef __cforall
         forall(dtype T)
+        forall(T &)
 #else
         #define T void
 …
 #ifdef __cforall
         // forall(otype T | sized(T))
+        // forall(T | sized(T))
         // static inline void ?{}(__small_array(T) & this) {}
         forall(dtype T | sized(T))
+        forall(T & | sized(T))
         static inline T & ?[?]( __small_array(T) & this, __lock_size_t idx ) {
                 return ((typeof(this.data))this.data)[idx];
+        }
         forall(dtype T | sized(T))
+        forall(T & | sized(T))
         static inline T & ?[?]( const __small_array(T) & this, __lock_size_t idx ) {
                 return ((typeof(this.data))this.data)[idx];
+        }
         forall(dtype T)
+        forall(T &)
         static inline T * begin( const __small_array(T) & this ) {
                 return ((typeof(this.data))this.data);
+        }
         forall(dtype T | sized(T))
+        forall(T & | sized(T))
         static inline T * end( const __small_array(T) & this ) {
                 return ((typeof(this.data))this.data) + this.size;
 …
 #ifdef __cforall
         trait is_node(dtype T) {
+        trait is_node(T &) {
                 T *& get_next( T & );
         };
 …
 //-----------------------------------------------------------------------------
 #ifdef __cforall
         forall(dtype TYPE)
+        forall(TYPE &)
         #define T TYPE
 #else
 …
 #ifdef __cforall
         forall(dtype T)
+        forall(T &)
         static inline void ?{}( __stack(T) & this ) {
                 (this.top){ 0p };
+        }
         static inline forall( dtype T | is_node(T) ) {
+        static inline forall( T & | is_node(T) ) {
                 void push( __stack(T) & this, T * val ) {
                         verify( !get_next( *val ) );
 …
 //-----------------------------------------------------------------------------
 #ifdef __cforall
         forall(dtype TYPE)
+        forall(TYPE &)
         #define T TYPE
 #else
 …
 #ifdef __cforall
         static inline forall( dtype T | is_node(T) ) {
+        static inline forall( T & | is_node(T) ) {
                 void ?{}( __queue(T) & this ) with( this ) {
                         (this.head){ 1p };
 …
 //-----------------------------------------------------------------------------
 #ifdef __cforall
         forall(dtype TYPE)
+        forall(TYPE &)
         #define T TYPE
         #define __getter_t * [T * & next, T * & prev] ( T & )
 …
 #ifdef __cforall
         forall(dtype T )
+        forall(T & )
         static inline [void] ?{}( __dllist(T) & this, * [T * & next, T * & prev] ( T & ) __get ) {
                 (this.head){ 0p };
 …
         #define next 0
         #define prev 1
         static inline forall(dtype T) {
+        static inline forall(T &) {
                 void push_front( __dllist(T) & this, T & node ) with( this ) {
                         verify(__get);

libcfa/src/bits/defs.hfa

-                      r5869cea
+                      r7b91c0e
 // file "LICENCE" distributed with Cforall.
 //
+// defs.hfa --
+// defs.hfa -- Commen macros, functions and typedefs
+// Most files depend on them and they are always useful to have.
+//
+//  *** Must not contain code specific to libcfathread ***
 //
 // Author           : Thierry Delisle
 …
         #endif
+}
+// pause to prevent excess processor bus usage
+#if defined( __i386 ) || defined( __x86_64 )
+        #define Pause() __asm__ __volatile__ ( "pause" : : : )
+#elif defined( __ARM_ARCH )
+        #define Pause() __asm__ __volatile__ ( "YIELD" : : : )
+#else
+        #error unsupported architecture
+#endif

libcfa/src/bits/locks.hfa

-                      r5869cea
+                      r7b91c0e
 // file "LICENCE" distributed with Cforall.
 //
+// bits/locks.hfa -- Fast internal locks.
+// bits/locks.hfa -- Basic spinlocks that are reused in the system.
+// Used for locks that aren't specific to cforall threads and can be used anywhere
+//
+//  *** Must not contain code specific to libcfathread ***
 //
 // Author           : Thierry Delisle
 …
 #include "bits/defs.hfa"
 #include <assert.h>
-#ifdef __cforall
-        extern "C" {
-                #include <pthread.h>
+        }
-#endif
-// pause to prevent excess processor bus usage
-#if defined( __i386 ) || defined( __x86_64 )
-        #define Pause() __asm__ __volatile__ ( "pause" : : : )
-#elif defined( __ARM_ARCH )
-        #define Pause() __asm__ __volatile__ ( "YIELD" : : : )
-#else
-        #error unsupported architecture
-#endif
 struct __spinlock_t {
 …
                 enable_interrupts_noPoll();
+        }
-        #ifdef __CFA_WITH_VERIFY__
-                extern bool __cfaabi_dbg_in_kernel();
-        #endif
-        extern "C" {
-                char * strerror(int);
+        }
-        #define CHECKED(x) { int err = x; if( err != 0 ) abort("KERNEL ERROR: Operation \"" #x "\" return error %d - %s\n", err, strerror(err)); }
-        struct __bin_sem_t {
-                pthread_mutex_t         lock;
-                pthread_cond_t          cond;
-                int                     val;
-        };
-        static inline void ?{}(__bin_sem_t & this) with( this ) {
-                // Create the mutex with error checking
-                pthread_mutexattr_t mattr;
-                pthread_mutexattr_init( &mattr );
-                pthread_mutexattr_settype( &mattr, PTHREAD_MUTEX_ERRORCHECK_NP);
-                pthread_mutex_init(&lock, &mattr);
-                pthread_cond_init (&cond, (const pthread_condattr_t *)0p);  // workaround trac#208: cast should not be required
-                val = 0;
+        }
-        static inline void ^?{}(__bin_sem_t & this) with( this ) {
-                CHECKED( pthread_mutex_destroy(&lock) );
-                CHECKED( pthread_cond_destroy (&cond) );
+        }
-        static inline void wait(__bin_sem_t & this) with( this ) {
-                verify(__cfaabi_dbg_in_kernel());
-                CHECKED( pthread_mutex_lock(&lock) );
-                        while(val < 1) {
-                                pthread_cond_wait(&cond, &lock);
+                        }
-                        val -= 1;
-                CHECKED( pthread_mutex_unlock(&lock) );
+        }
-        static inline bool post(__bin_sem_t & this) with( this ) {
-                bool needs_signal = false;
-                CHECKED( pthread_mutex_lock(&lock) );
-                        if(val < 1) {
-                                val += 1;
-                                pthread_cond_signal(&cond);
-                                needs_signal = true;
+                        }
-                CHECKED( pthread_mutex_unlock(&lock) );
-                return needs_signal;
+        }
-        #undef CHECKED
-        struct $thread;
-        extern void park( void );
-        extern void unpark( struct $thread * this );
-        static inline struct $thread * active_thread ();
-        // Semaphore which only supports a single thread
-        struct single_sem {
-                struct $thread * volatile ptr;
-        };
-        static inline {
-                void  ?{}(single_sem & this) {
-                        this.ptr = 0p;
+                }
-                void ^?{}(single_sem &) {}
-                bool wait(single_sem & this) {
-                        for() {
-                                struct $thread * expected = this.ptr;
-                                if(expected == 1p) {
-                                        if(__atomic_compare_exchange_n(&this.ptr, &expected, 0p, false, __ATOMIC_SEQ_CST, __ATOMIC_SEQ_CST)) {
-                                                return false;
+                                        }
+                                }
-                                else {
-                                        /* paranoid */ verify( expected == 0p );
-                                        if(__atomic_compare_exchange_n(&this.ptr, &expected, active_thread(), false, __ATOMIC_SEQ_CST, __ATOMIC_SEQ_CST)) {
-                                                park();
-                                                return true;
+                                        }
+                                }
+                        }
+                }
-                bool post(single_sem & this) {
-                        for() {
-                                struct $thread * expected = this.ptr;
-                                if(expected == 1p) return false;
-                                if(expected == 0p) {
-                                        if(__atomic_compare_exchange_n(&this.ptr, &expected, 1p, false, __ATOMIC_SEQ_CST, __ATOMIC_SEQ_CST)) {
-                                                return false;
+                                        }
+                                }
-                                else {
-                                        if(__atomic_compare_exchange_n(&this.ptr, &expected, 0p, false, __ATOMIC_SEQ_CST, __ATOMIC_SEQ_CST)) {
-                                                unpark( expected );
-                                                return true;
+                                        }
+                                }
+                        }
+                }
+        }
-        // Synchronozation primitive which only supports a single thread and one post
-        // Similar to a binary semaphore with a 'one shot' semantic
-        // is expected to be discarded after each party call their side
-        struct oneshot {
-                // Internal state :
-                //     0p     : is initial state (wait will block)
-                //     1p     : fulfilled (wait won't block)
-                // any thread : a thread is currently waiting
-                struct $thread * volatile ptr;
-        };
-        static inline {
-                void  ?{}(oneshot & this) {
-                        this.ptr = 0p;
+                }
-                void ^?{}(oneshot &) {}
-                // Wait for the post, return immidiately if it already happened.
-                // return true if the thread was parked
-                bool wait(oneshot & this) {
-                        for() {
-                                struct $thread * expected = this.ptr;
-                                if(expected == 1p) return false;
-                                /* paranoid */ verify( expected == 0p );
-                                if(__atomic_compare_exchange_n(&this.ptr, &expected, active_thread(), false, __ATOMIC_SEQ_CST, __ATOMIC_SEQ_CST)) {
-                                        park();
-                                        /* paranoid */ verify( this.ptr == 1p );
-                                        return true;
+                                }
+                        }
+                }
-                // Mark as fulfilled, wake thread if needed
-                // return true if a thread was unparked
-                bool post(oneshot & this) {
-                        struct $thread * got = __atomic_exchange_n( &this.ptr, 1p, __ATOMIC_SEQ_CST);
-                        if( got == 0p ) return false;
-                        unpark( got );
-                        return true;
+                }
+        }
-        // base types for future to build upon
-        // It is based on the 'oneshot' type to allow multiple futures
-        // to block on the same instance, permitting users to block a single
-        // thread on "any of" [a given set of] futures.
-        // does not support multiple threads waiting on the same future
-        struct future_t {
-                // Internal state :
-                //     0p      : is initial state (wait will block)
-                //     1p      : fulfilled (wait won't block)
-                //     2p      : in progress ()
-                //     3p      : abandoned, server should delete
-                // any oneshot : a context has been setup to wait, a thread could wait on it
-                struct oneshot * volatile ptr;
-        };
-        static inline {
-                void  ?{}(future_t & this) {
-                        this.ptr = 0p;
+                }
-                void ^?{}(future_t &) {}
-                void reset(future_t & this) {
-                        // needs to be in 0p or 1p
-                        __atomic_exchange_n( &this.ptr, 0p, __ATOMIC_SEQ_CST);
+                }
-                // check if the future is available
-                bool available( future_t & this ) {
-                        return this.ptr == 1p;
+                }
-                // Prepare the future to be waited on
-                // intented to be use by wait, wait_any, waitfor, etc. rather than used directly
-                bool setup( future_t & this, oneshot & wait_ctx ) {
-                        /* paranoid */ verify( wait_ctx.ptr == 0p );
-                        // The future needs to set the wait context
-                        for() {
-                                struct oneshot * expected = this.ptr;
-                                // Is the future already fulfilled?
-                                if(expected == 1p) return false; // Yes, just return false (didn't block)
-                                // The future is not fulfilled, try to setup the wait context
-                                /* paranoid */ verify( expected == 0p );
-                                if(__atomic_compare_exchange_n(&this.ptr, &expected, &wait_ctx, false, __ATOMIC_SEQ_CST, __ATOMIC_SEQ_CST)) {
-                                        return true;
+                                }
+                        }
+                }
-                // Stop waiting on a future
-                // When multiple futures are waited for together in "any of" pattern
-                // futures that weren't fulfilled before the thread woke up
-                // should retract the wait ctx
-                // intented to be use by wait, wait_any, waitfor, etc. rather than used directly
-                void retract( future_t & this, oneshot & wait_ctx ) {
-                        // Remove the wait context
-                        struct oneshot * got = __atomic_exchange_n( &this.ptr, 0p, __ATOMIC_SEQ_CST);
-                        // got == 0p: future was never actually setup, just return
-                        if( got == 0p ) return;
-                        // got == wait_ctx: since fulfil does an atomic_swap,
-                        // if we got back the original then no one else saw context
-                        // It is safe to delete (which could happen after the return)
-                        if( got == &wait_ctx ) return;
-                        // got == 1p: the future is ready and the context was fully consumed
-                        // the server won't use the pointer again
-                        // It is safe to delete (which could happen after the return)
-                        if( got == 1p ) return;
-                        // got == 2p: the future is ready but the context hasn't fully been consumed
-                        // spin until it is safe to move on
-                        if( got == 2p ) {
-                                while( this.ptr != 1p ) Pause();
-                                return;
+                        }
-                        // got == any thing else, something wen't wrong here, abort
-                        abort("Future in unexpected state");
+                }
-                // Mark the future as abandoned, meaning it will be deleted by the server
-                bool abandon( future_t & this ) {
-                        /* paranoid */ verify( this.ptr != 3p );
-                        // Mark the future as abandonned
-                        struct oneshot * got = __atomic_exchange_n( &this.ptr, 3p, __ATOMIC_SEQ_CST);
-                        // If the future isn't already fulfilled, let the server delete it
-                        if( got == 0p ) return false;
-                        // got == 2p: the future is ready but the context hasn't fully been consumed
-                        // spin until it is safe to move on
-                        if( got == 2p ) {
-                                while( this.ptr != 1p ) Pause();
-                                got = 1p;
+                        }
-                        // The future is completed delete it now
-                        /* paranoid */ verify( this.ptr != 1p );
-                        free( &this );
-                        return true;
+                }
-                // from the server side, mark the future as fulfilled
-                // delete it if needed
-                bool fulfil( future_t & this ) {
-                        for() {
-                                struct oneshot * expected = this.ptr;
-                                // was this abandoned?
-                                #if defined(__GNUC__) && __GNUC__ >= 7
-                                        #pragma GCC diagnostic push
-                                        #pragma GCC diagnostic ignored "-Wfree-nonheap-object"
-                                #endif
-                                        if( expected == 3p ) { free( &this ); return false; }
-                                #if defined(__GNUC__) && __GNUC__ >= 7
-                                        #pragma GCC diagnostic pop
-                                #endif
-                                /* paranoid */ verify( expected != 1p ); // Future is already fulfilled, should not happen
-                                /* paranoid */ verify( expected != 2p ); // Future is bein fulfilled by someone else, this is even less supported then the previous case.
-                                // If there is a wait context, we need to consume it and mark it as consumed after
-                                // If there is no context then we can skip the in progress phase
-                                struct oneshot * want = expected == 0p ? 1p : 2p;
-                                if(__atomic_compare_exchange_n(&this.ptr, &expected, want, false, __ATOMIC_SEQ_CST, __ATOMIC_SEQ_CST)) {
-                                        if( expected == 0p ) { /* paranoid */ verify( this.ptr == 1p); return false; }
-                                        bool ret = post( *expected );
-                                        __atomic_store_n( &this.ptr, 1p, __ATOMIC_SEQ_CST);
-                                        return ret;
+                                }
+                        }
+                }
-                // Wait for the future to be fulfilled
-                bool wait( future_t & this ) {
-                        oneshot temp;
-                        if( !setup(this, temp) ) return false;
-                        // Wait context is setup, just wait on it
-                        bool ret = wait( temp );
-                        // Wait for the future to tru
-                        while( this.ptr == 2p ) Pause();
-                        // Make sure the state makes sense
-                        // Should be fulfilled, could be in progress but it's out of date if so
-                        // since if that is the case, the oneshot was fulfilled (unparking this thread)
-                        // and the oneshot should not be needed any more
-                        __attribute__((unused)) struct oneshot * was = this.ptr;
-                        /* paranoid */ verifyf( was == 1p, "Expected this.ptr to be 1p, was %p\n", was );
-                        // Mark the future as fulfilled, to be consistent
-                        // with potential calls to avail
-                        // this.ptr = 1p;
-                        return ret;
+                }
+        }
 #endif

libcfa/src/bits/queue.hfa

-                      r5869cea
+                      r7b91c0e
 // instead of being null.
 forall( dtype T | { T *& Next ( T * ); } ) {
+forall( T & | { T *& Next ( T * ); } ) {
         struct Queue {
                 inline Collection;                                                              // Plan 9 inheritance
 …
 } // distribution
 forall( dtype T | { T *& Next ( T * ); } ) {
+forall( T & | { T *& Next ( T * ); } ) {
         struct QueueIter {
                 inline ColIter;                                                                 // Plan 9 inheritance

libcfa/src/bits/sequence.hfa

-                      r5869cea
+                      r7b91c0e
         // // wrappers to make Collection have T
         // forall( dtype T ) {
+        // forall( T & ) {
         //      T *& Back( T * n ) {
         //              return (T *)Back( (Seqable *)n );
 …
 // and the back field of the last node points at the first node (circular).
 forall( dtype T | { T *& Back ( T * ); T *& Next ( T * ); } ) {
+forall( T & | { T *& Back ( T * ); T *& Next ( T * ); } ) {
         struct Sequence {
                 inline Collection;                                                              // Plan 9 inheritance
 …
 } // distribution
 forall( dtype T | { T *& Back ( T * ); T *& Next ( T * ); } ) {
+forall( T & | { T *& Back ( T * ); T *& Next ( T * ); } ) {
         // SeqIter(T) is used to iterate over a Sequence(T) in head-to-tail order.
         struct SeqIter {

libcfa/src/bits/stack.hfa

-                      r5869cea
+                      r7b91c0e
 // instead of being null.
 forall( dtype T | { T *& Next ( T * ); } ) {
+forall( T & | { T *& Next ( T * ); } ) {
         struct Stack {
                 inline Collection;                                                              // Plan 9 inheritance
 …
 // order returned by drop().
 forall( dtype T | { T *& Next ( T * ); } ) {
+forall( T & | { T *& Next ( T * ); } ) {
         struct StackIter {
                 inline ColIter;                                                                 // Plan 9 inheritance

libcfa/src/common.cfa

r5869cea	r7b91c0e
23	23	[ long int, long int ] div( long int num, long int denom ) { ldiv_t qr = ldiv( num, denom ); return [ qr.quot, qr.rem ]; }
24	24	[ long long int, long long int ] div( long long int num, long long int denom ) { lldiv_t qr = lldiv( num, denom ); return [ qr.quot, qr.rem ]; }
25		forall( ~~otype~~ T \| { T ?/?( T, T ); T ?%?( T, T ); } )
	25	forall( T \| { T ?/?( T, T ); T ?%?( T, T ); } )
26	26	[ T, T ] div( T num, T denom ) { return [ num / denom, num % denom ]; }
27	27

libcfa/src/common.hfa

-                      r5869cea
+                      r7b91c0e
 [ long int, long int ] div( long int num, long int denom );
 [ long long int, long long int ] div( long long int num, long long int denom );
 forall( otype T | { T ?/?( T, T ); T ?%?( T, T ); } )
+forall( T | { T ?/?( T, T ); T ?%?( T, T ); } )
 [ T, T ] div( T num, T demon );
 …
 } // distribution
 forall( otype T | { void ?{}( T &, zero_t ); int ?<?( T, T ); T -?( T ); } )
+forall( T | { void ?{}( T &, zero_t ); int ?<?( T, T ); T -?( T ); } )
 T abs( T );
 …
         intptr_t min( intptr_t t1, intptr_t t2 ) { return t1 < t2 ? t1 : t2; } // optimization
         uintptr_t min( uintptr_t t1, uintptr_t t2 ) { return t1 < t2 ? t1 : t2; } // optimization
         forall( otype T | { int ?<?( T, T ); } )
+        forall( T | { int ?<?( T, T ); } )
         T min( T t1, T t2 ) { return t1 < t2 ? t1 : t2; }
 …
         intptr_t max( intptr_t t1, intptr_t t2 ) { return t1 > t2 ? t1 : t2; } // optimization
         uintptr_t max( uintptr_t t1, uintptr_t t2 ) { return t1 > t2 ? t1 : t2; } // optimization
         forall( otype T | { int ?>?( T, T ); } )
+        forall( T | { int ?>?( T, T ); } )
         T max( T t1, T t2 ) { return t1 > t2 ? t1 : t2; }
         forall( otype T | { T min( T, T ); T max( T, T ); } )
+        forall( T | { T min( T, T ); T max( T, T ); } )
         T clamp( T value, T min_val, T max_val ) { return max( min_val, min( value, max_val ) ); }
         forall( otype T )
+        forall( T )
         void swap( T & v1, T & v2 ) { T temp = v1; v1 = v2; v2 = temp; }
 } // distribution

libcfa/src/concurrency/coroutine.cfa

-                      r5869cea
+                      r7b91c0e
 //-----------------------------------------------------------------------------
 FORALL_DATA_INSTANCE(CoroutineCancelled, (dtype coroutine_t), (coroutine_t))
 forall(dtype T)
+FORALL_DATA_INSTANCE(CoroutineCancelled, (coroutine_t &), (coroutine_t))
+forall(T &)
 void mark_exception(CoroutineCancelled(T) *) {}
 forall(dtype T)
+forall(T &)
 void copy(CoroutineCancelled(T) * dst, CoroutineCancelled(T) * src) {
         dst->virtual_table = src->virtual_table;
 …
+}
 forall(dtype T)
+forall(T &)
 const char * msg(CoroutineCancelled(T) *) {
         return "CoroutineCancelled(...)";
 …
 // This code should not be inlined. It is the error path on resume.
 forall(dtype T | is_coroutine(T))
+forall(T & | is_coroutine(T))
 void __cfaehm_cancelled_coroutine( T & cor, $coroutine * desc ) {
         verify( desc->cancellation );
 …
 // Part of the Public API
 // Not inline since only ever called once per coroutine
 forall(dtype T | is_coroutine(T))
+forall(T & | is_coroutine(T))
 void prime(T& cor) {
         $coroutine* this = get_coroutine(cor);

libcfa/src/concurrency/coroutine.hfa

-                      r5869cea
+                      r7b91c0e
 //-----------------------------------------------------------------------------
 // Exception thrown from resume when a coroutine stack is cancelled.
 FORALL_DATA_EXCEPTION(CoroutineCancelled, (dtype coroutine_t), (coroutine_t)) (
+FORALL_DATA_EXCEPTION(CoroutineCancelled, (coroutine_t &), (coroutine_t)) (
         coroutine_t * the_coroutine;
         exception_t * the_exception;
 );
 forall(dtype T)
+forall(T &)
 void copy(CoroutineCancelled(T) * dst, CoroutineCancelled(T) * src);
 forall(dtype T)
+forall(T &)
 const char * msg(CoroutineCancelled(T) *);
 …
 // Anything that implements this trait can be resumed.
 // Anything that is resumed is a coroutine.
 trait is_coroutine(dtype T | IS_RESUMPTION_EXCEPTION(CoroutineCancelled, (T))) {
+trait is_coroutine(T & | IS_RESUMPTION_EXCEPTION(CoroutineCancelled, (T))) {
         void main(T & this);
         $coroutine * get_coroutine(T & this);
 …
 //-----------------------------------------------------------------------------
 // Public coroutine API
 forall(dtype T | is_coroutine(T))
+forall(T & | is_coroutine(T))
 void prime(T & cor);
 …
         void __cfactx_invoke_coroutine(void (*main)(void *), void * this);
         forall(dtype T)
+        forall(T &)
         void __cfactx_start(void (*main)(T &), struct $coroutine * cor, T & this, void (*invoke)(void (*main)(void *), void *));
 …
+}
 forall(dtype T | is_coroutine(T))
+forall(T & | is_coroutine(T))
 void __cfaehm_cancelled_coroutine( T & cor, $coroutine * desc );
 // Resume implementation inlined for performance
 forall(dtype T | is_coroutine(T))
+forall(T & | is_coroutine(T))
 static inline T & resume(T & cor) {
         // optimization : read TLS once and reuse it

libcfa/src/concurrency/future.hfa

-                      r5869cea
+                      r7b91c0e
 #include "monitor.hfa"
 forall( otype T ) {
+forall( T ) {
         struct future {
                 inline future_t;
 …
+}
 forall( otype T ) {
+forall( T ) {
         monitor multi_future {
                 inline future_t;

libcfa/src/concurrency/io/types.hfa

-                      r5869cea
+                      r7b91c0e
 // file "LICENCE" distributed with Cforall.
 //
+// io/types.hfa --
+// io/types.hfa -- PRIVATE
+// Types used by the I/O subsystem
 //
 // Author           : Thierry Delisle
 …
 #include "bits/locks.hfa"
+#include "kernel/fwd.hfa"
 #if defined(CFA_HAVE_LINUX_IO_URING_H)

libcfa/src/concurrency/kernel.cfa

-                      r5869cea
+                      r7b91c0e
+        }
         V( this->terminated );
+        post( this->terminated );
         if(this == mainProcessor) {
 …
 // Unexpected Terminating logic
 //=============================================================================================
+extern "C" {
+        extern void __cfaabi_real_abort(void);
+}
+static volatile bool kernel_abort_called = false;
+void * kernel_abort(void) __attribute__ ((__nothrow__)) {
+        // abort cannot be recursively entered by the same or different processors because all signal handlers return when
+        // the globalAbort flag is true.
+        bool first = !__atomic_test_and_set( &kernel_abort_called, __ATOMIC_SEQ_CST);
+        // first task to abort ?
+        if ( !first ) {
+                // We aren't the first to abort.
+                // I give up, just let C handle it
+                __cfaabi_real_abort();
+        }
+        // disable interrupts, it no longer makes sense to try to interrupt this processor
+        disable_interrupts();
+        return __cfaabi_tls.this_thread;
+}
+void kernel_abort_msg( void * kernel_data, char * abort_text, int abort_text_size ) {
+        $thread * thrd = ( $thread * ) kernel_data;
+void __kernel_abort_msg( char * abort_text, int abort_text_size ) {
+        $thread * thrd = __cfaabi_tls.this_thread;
         if(thrd) {
 …
+}
 int kernel_abort_lastframe( void ) __attribute__ ((__nothrow__)) {
         return get_coroutine(kernelTLS().this_thread) == get_coroutine(mainThread) ? 4 : 2;
+int __kernel_abort_lastframe( void ) __attribute__ ((__nothrow__)) {
+        return get_coroutine(__cfaabi_tls.this_thread) == get_coroutine(mainThread) ? 4 : 2;
+}
 …
 // Kernel Utilities
 //=============================================================================================
-//-----------------------------------------------------------------------------
-// Locks
-void  ?{}( semaphore & this, int count = 1 ) {
-        (this.lock){};
-        this.count = count;
-        (this.waiting){};
+}
-void ^?{}(semaphore & this) {}
-bool P(semaphore & this) with( this ){
-        lock( lock __cfaabi_dbg_ctx2 );
-        count -= 1;
-        if ( count < 0 ) {
-                // queue current task
-                append( waiting, active_thread() );
-                // atomically release spin lock and block
-                unlock( lock );
-                park();
-                return true;
+        }
-        else {
-            unlock( lock );
-            return false;
+        }
+}
-bool V(semaphore & this) with( this ) {
-        $thread * thrd = 0p;
-        lock( lock __cfaabi_dbg_ctx2 );
-        count += 1;
-        if ( count <= 0 ) {
-                // remove task at head of waiting list
-                thrd = pop_head( waiting );
+        }
-        unlock( lock );
-        // make new owner
-        unpark( thrd );
-        return thrd != 0p;
+}
-bool V(semaphore & this, unsigned diff) with( this ) {
-        $thread * thrd = 0p;
-        lock( lock __cfaabi_dbg_ctx2 );
-        int release = max(-count, (int)diff);
-        count += diff;
-        for(release) {
-                unpark( pop_head( waiting ) );
+        }
-        unlock( lock );
-        return thrd != 0p;
+}
 //-----------------------------------------------------------------------------
 // Debug

libcfa/src/concurrency/kernel.hfa

-                      r5869cea
+                      r7b91c0e
 // file "LICENCE" distributed with Cforall.
 //
 // kernel --
+// kernel -- Header containing the core of the kernel API
 //
 // Author           : Thierry Delisle
 …
 extern "C" {
         #include <bits/pthreadtypes.h>
+        #include <pthread.h>
         #include <linux/types.h>
+}
 //-----------------------------------------------------------------------------
+// Locks
+struct semaphore {
+        __spinlock_t lock;
+        int count;
+        __queue_t($thread) waiting;
+};
+void  ?{}(semaphore & this, int count = 1);
+void ^?{}(semaphore & this);
+bool   P (semaphore & this);
+bool   V (semaphore & this);
+bool   V (semaphore & this, unsigned count);
+// Underlying Locks
+#ifdef __CFA_WITH_VERIFY__
+        extern bool __cfaabi_dbg_in_kernel();
+#endif
+extern "C" {
+        char * strerror(int);
+}
+#define CHECKED(x) { int err = x; if( err != 0 ) abort("KERNEL ERROR: Operation \"" #x "\" return error %d - %s\n", err, strerror(err)); }
+struct __bin_sem_t {
+        pthread_mutex_t         lock;
+        pthread_cond_t          cond;
+        int                     val;
+};
+static inline void ?{}(__bin_sem_t & this) with( this ) {
+        // Create the mutex with error checking
+        pthread_mutexattr_t mattr;
+        pthread_mutexattr_init( &mattr );
+        pthread_mutexattr_settype( &mattr, PTHREAD_MUTEX_ERRORCHECK_NP);
+        pthread_mutex_init(&lock, &mattr);
+        pthread_cond_init (&cond, (const pthread_condattr_t *)0p);  // workaround trac#208: cast should not be required
+        val = 0;
+}
+static inline void ^?{}(__bin_sem_t & this) with( this ) {
+        CHECKED( pthread_mutex_destroy(&lock) );
+        CHECKED( pthread_cond_destroy (&cond) );
+}
+static inline void wait(__bin_sem_t & this) with( this ) {
+        verify(__cfaabi_dbg_in_kernel());
+        CHECKED( pthread_mutex_lock(&lock) );
+                while(val < 1) {
+                        pthread_cond_wait(&cond, &lock);
+                }
+                val -= 1;
+        CHECKED( pthread_mutex_unlock(&lock) );
+}
+static inline bool post(__bin_sem_t & this) with( this ) {
+        bool needs_signal = false;
+        CHECKED( pthread_mutex_lock(&lock) );
+                if(val < 1) {
+                        val += 1;
+                        pthread_cond_signal(&cond);
+                        needs_signal = true;
+                }
+        CHECKED( pthread_mutex_unlock(&lock) );
+        return needs_signal;
+}
+#undef CHECKED
 …
         // Termination synchronisation (user semaphore)
         semaphore terminated;
+        oneshot terminated;
         // pthread Stack

libcfa/src/concurrency/kernel/fwd.hfa

-                      r5869cea
+                      r7b91c0e
 // file "LICENCE" distributed with Cforall.
 //
+// kernel/fwd.hfa --
+// kernel/fwd.hfa -- PUBLIC
+// Fundamental code needed to implement threading M.E.S. algorithms.
 //
 // Author           : Thierry Delisle
 …
                 extern uint64_t thread_rand();
+                // Semaphore which only supports a single thread
+                struct single_sem {
+                        struct $thread * volatile ptr;
+                };
+                static inline {
+                        void  ?{}(single_sem & this) {
+                                this.ptr = 0p;
+                        }
+                        void ^?{}(single_sem &) {}
+                        bool wait(single_sem & this) {
+                                for() {
+                                        struct $thread * expected = this.ptr;
+                                        if(expected == 1p) {
+                                                if(__atomic_compare_exchange_n(&this.ptr, &expected, 0p, false, __ATOMIC_SEQ_CST, __ATOMIC_SEQ_CST)) {
+                                                        return false;
+                                                }
+                                        }
+                                        else {
+                                                /* paranoid */ verify( expected == 0p );
+                                                if(__atomic_compare_exchange_n(&this.ptr, &expected, active_thread(), false, __ATOMIC_SEQ_CST, __ATOMIC_SEQ_CST)) {
+                                                        park();
+                                                        return true;
+                                                }
+                                        }
+                                }
+                        }
+                        bool post(single_sem & this) {
+                                for() {
+                                        struct $thread * expected = this.ptr;
+                                        if(expected == 1p) return false;
+                                        if(expected == 0p) {
+                                                if(__atomic_compare_exchange_n(&this.ptr, &expected, 1p, false, __ATOMIC_SEQ_CST, __ATOMIC_SEQ_CST)) {
+                                                        return false;
+                                                }
+                                        }
+                                        else {
+                                                if(__atomic_compare_exchange_n(&this.ptr, &expected, 0p, false, __ATOMIC_SEQ_CST, __ATOMIC_SEQ_CST)) {
+                                                        unpark( expected );
+                                                        return true;
+                                                }
+                                        }
+                                }
+                        }
+                }
+                // Synchronozation primitive which only supports a single thread and one post
+                // Similar to a binary semaphore with a 'one shot' semantic
+                // is expected to be discarded after each party call their side
+                struct oneshot {
+                        // Internal state :
+                        //     0p     : is initial state (wait will block)
+                        //     1p     : fulfilled (wait won't block)
+                        // any thread : a thread is currently waiting
+                        struct $thread * volatile ptr;
+                };
+                static inline {
+                        void  ?{}(oneshot & this) {
+                                this.ptr = 0p;
+                        }
+                        void ^?{}(oneshot &) {}
+                        // Wait for the post, return immidiately if it already happened.
+                        // return true if the thread was parked
+                        bool wait(oneshot & this) {
+                                for() {
+                                        struct $thread * expected = this.ptr;
+                                        if(expected == 1p) return false;
+                                        /* paranoid */ verify( expected == 0p );
+                                        if(__atomic_compare_exchange_n(&this.ptr, &expected, active_thread(), false, __ATOMIC_SEQ_CST, __ATOMIC_SEQ_CST)) {
+                                                park();
+                                                /* paranoid */ verify( this.ptr == 1p );
+                                                return true;
+                                        }
+                                }
+                        }
+                        // Mark as fulfilled, wake thread if needed
+                        // return true if a thread was unparked
+                        bool post(oneshot & this) {
+                                struct $thread * got = __atomic_exchange_n( &this.ptr, 1p, __ATOMIC_SEQ_CST);
+                                if( got == 0p ) return false;
+                                unpark( got );
+                                return true;
+                        }
+                }
+                // base types for future to build upon
+                // It is based on the 'oneshot' type to allow multiple futures
+                // to block on the same instance, permitting users to block a single
+                // thread on "any of" [a given set of] futures.
+                // does not support multiple threads waiting on the same future
+                struct future_t {
+                        // Internal state :
+                        //     0p      : is initial state (wait will block)
+                        //     1p      : fulfilled (wait won't block)
+                        //     2p      : in progress ()
+                        //     3p      : abandoned, server should delete
+                        // any oneshot : a context has been setup to wait, a thread could wait on it
+                        struct oneshot * volatile ptr;
+                };
+                static inline {
+                        void  ?{}(future_t & this) {
+                                this.ptr = 0p;
+                        }
+                        void ^?{}(future_t &) {}
+                        void reset(future_t & this) {
+                                // needs to be in 0p or 1p
+                                __atomic_exchange_n( &this.ptr, 0p, __ATOMIC_SEQ_CST);
+                        }
+                        // check if the future is available
+                        bool available( future_t & this ) {
+                                return this.ptr == 1p;
+                        }
+                        // Prepare the future to be waited on
+                        // intented to be use by wait, wait_any, waitfor, etc. rather than used directly
+                        bool setup( future_t & this, oneshot & wait_ctx ) {
+                                /* paranoid */ verify( wait_ctx.ptr == 0p );
+                                // The future needs to set the wait context
+                                for() {
+                                        struct oneshot * expected = this.ptr;
+                                        // Is the future already fulfilled?
+                                        if(expected == 1p) return false; // Yes, just return false (didn't block)
+                                        // The future is not fulfilled, try to setup the wait context
+                                        /* paranoid */ verify( expected == 0p );
+                                        if(__atomic_compare_exchange_n(&this.ptr, &expected, &wait_ctx, false, __ATOMIC_SEQ_CST, __ATOMIC_SEQ_CST)) {
+                                                return true;
+                                        }
+                                }
+                        }
+                        // Stop waiting on a future
+                        // When multiple futures are waited for together in "any of" pattern
+                        // futures that weren't fulfilled before the thread woke up
+                        // should retract the wait ctx
+                        // intented to be use by wait, wait_any, waitfor, etc. rather than used directly
+                        void retract( future_t & this, oneshot & wait_ctx ) {
+                                // Remove the wait context
+                                struct oneshot * got = __atomic_exchange_n( &this.ptr, 0p, __ATOMIC_SEQ_CST);
+                                // got == 0p: future was never actually setup, just return
+                                if( got == 0p ) return;
+                                // got == wait_ctx: since fulfil does an atomic_swap,
+                                // if we got back the original then no one else saw context
+                                // It is safe to delete (which could happen after the return)
+                                if( got == &wait_ctx ) return;
+                                // got == 1p: the future is ready and the context was fully consumed
+                                // the server won't use the pointer again
+                                // It is safe to delete (which could happen after the return)
+                                if( got == 1p ) return;
+                                // got == 2p: the future is ready but the context hasn't fully been consumed
+                                // spin until it is safe to move on
+                                if( got == 2p ) {
+                                        while( this.ptr != 1p ) Pause();
+                                        return;
+                                }
+                                // got == any thing else, something wen't wrong here, abort
+                                abort("Future in unexpected state");
+                        }
+                        // Mark the future as abandoned, meaning it will be deleted by the server
+                        bool abandon( future_t & this ) {
+                                /* paranoid */ verify( this.ptr != 3p );
+                                // Mark the future as abandonned
+                                struct oneshot * got = __atomic_exchange_n( &this.ptr, 3p, __ATOMIC_SEQ_CST);
+                                // If the future isn't already fulfilled, let the server delete it
+                                if( got == 0p ) return false;
+                                // got == 2p: the future is ready but the context hasn't fully been consumed
+                                // spin until it is safe to move on
+                                if( got == 2p ) {
+                                        while( this.ptr != 1p ) Pause();
+                                        got = 1p;
+                                }
+                                // The future is completed delete it now
+                                /* paranoid */ verify( this.ptr != 1p );
+                                free( &this );
+                                return true;
+                        }
+                        // from the server side, mark the future as fulfilled
+                        // delete it if needed
+                        bool fulfil( future_t & this ) {
+                                for() {
+                                        struct oneshot * expected = this.ptr;
+                                        // was this abandoned?
+                                        #if defined(__GNUC__) && __GNUC__ >= 7
+                                                #pragma GCC diagnostic push
+                                                #pragma GCC diagnostic ignored "-Wfree-nonheap-object"
+                                        #endif
+                                                if( expected == 3p ) { free( &this ); return false; }
+                                        #if defined(__GNUC__) && __GNUC__ >= 7
+                                                #pragma GCC diagnostic pop
+                                        #endif
+                                        /* paranoid */ verify( expected != 1p ); // Future is already fulfilled, should not happen
+                                        /* paranoid */ verify( expected != 2p ); // Future is bein fulfilled by someone else, this is even less supported then the previous case.
+                                        // If there is a wait context, we need to consume it and mark it as consumed after
+                                        // If there is no context then we can skip the in progress phase
+                                        struct oneshot * want = expected == 0p ? 1p : 2p;
+                                        if(__atomic_compare_exchange_n(&this.ptr, &expected, want, false, __ATOMIC_SEQ_CST, __ATOMIC_SEQ_CST)) {
+                                                if( expected == 0p ) { /* paranoid */ verify( this.ptr == 1p); return false; }
+                                                bool ret = post( *expected );
+                                                __atomic_store_n( &this.ptr, 1p, __ATOMIC_SEQ_CST);
+                                                return ret;
+                                        }
+                                }
+                        }
+                        // Wait for the future to be fulfilled
+                        bool wait( future_t & this ) {
+                                oneshot temp;
+                                if( !setup(this, temp) ) return false;
+                                // Wait context is setup, just wait on it
+                                bool ret = wait( temp );
+                                // Wait for the future to tru
+                                while( this.ptr == 2p ) Pause();
+                                // Make sure the state makes sense
+                                // Should be fulfilled, could be in progress but it's out of date if so
+                                // since if that is the case, the oneshot was fulfilled (unparking this thread)
+                                // and the oneshot should not be needed any more
+                                __attribute__((unused)) struct oneshot * was = this.ptr;
+                                /* paranoid */ verifyf( was == 1p, "Expected this.ptr to be 1p, was %p\n", was );
+                                // Mark the future as fulfilled, to be consistent
+                                // with potential calls to avail
+                                // this.ptr = 1p;
+                                return ret;
+                        }
+                }
                 //-----------------------------------------------------------------------
                 // Statics call at the end of each thread to register statistics

libcfa/src/concurrency/kernel/startup.cfa

-                      r5869cea
+                      r7b91c0e
         void ?{}(processor & this) with( this ) {
                 ( this.idle ){};
                 ( this.terminated ){ 0 };
+                ( this.terminated ){};
                 ( this.runner ){};
                 init( this, "Main Processor", *mainCluster );
 …
 void ?{}(processor & this, const char name[], cluster & _cltr) {
         ( this.idle ){};
         ( this.terminated ){ 0 };
+        ( this.terminated ){};
         ( this.runner ){};
 …
                 __wake_proc( &this );
                 P( terminated );
+                wait( terminated );
                 /* paranoid */ verify( active_processor() != &this);
+        }

libcfa/src/concurrency/locks.cfa

-                      r5869cea
+                      r7b91c0e
 //-----------------------------------------------------------------------------
 // info_thread
 forall(dtype L | is_blocking_lock(L)) {
+forall(L & | is_blocking_lock(L)) {
         struct info_thread {
                 // used to put info_thread on a dl queue (aka sequence)
 …
 //-----------------------------------------------------------------------------
 // alarm node wrapper
 forall(dtype L | is_blocking_lock(L)) {
+forall(L & | is_blocking_lock(L)) {
         struct alarm_node_wrap {
                 alarm_node_t alarm_node;
 …
 //-----------------------------------------------------------------------------
 // condition variable
 forall(dtype L | is_blocking_lock(L)) {
+forall(L & | is_blocking_lock(L)) {
         void ?{}( condition_variable(L) & this ){
 …
         bool wait( condition_variable(L) & this, L & l, uintptr_t info, Time time         ) with(this) { WAIT_TIME( info, &l , time ) }
+}
+//-----------------------------------------------------------------------------
+// Semaphore
+void  ?{}( semaphore & this, int count = 1 ) {
+        (this.lock){};
+        this.count = count;
+        (this.waiting){};
+}
+void ^?{}(semaphore & this) {}
+bool P(semaphore & this) with( this ){
+        lock( lock __cfaabi_dbg_ctx2 );
+        count -= 1;
+        if ( count < 0 ) {
+                // queue current task
+                append( waiting, active_thread() );
+                // atomically release spin lock and block
+                unlock( lock );
+                park();
+                return true;
+        }
+        else {
+            unlock( lock );
+            return false;
+        }
+}
+bool V(semaphore & this) with( this ) {
+        $thread * thrd = 0p;
+        lock( lock __cfaabi_dbg_ctx2 );
+        count += 1;
+        if ( count <= 0 ) {
+                // remove task at head of waiting list
+                thrd = pop_head( waiting );
+        }
+        unlock( lock );
+        // make new owner
+        unpark( thrd );
+        return thrd != 0p;
+}
+bool V(semaphore & this, unsigned diff) with( this ) {
+        $thread * thrd = 0p;
+        lock( lock __cfaabi_dbg_ctx2 );
+        int release = max(-count, (int)diff);
+        count += diff;
+        for(release) {
+                unpark( pop_head( waiting ) );
+        }
+        unlock( lock );
+        return thrd != 0p;
+}

libcfa/src/concurrency/locks.hfa

-                      r5869cea
+                      r7b91c0e
 //-----------------------------------------------------------------------------
 // is_blocking_lock
 trait is_blocking_lock(dtype L | sized(L)) {
+trait is_blocking_lock(L & | sized(L)) {
         // For synchronization locks to use when acquiring
         void on_notify( L &, struct $thread * );
 …
 // the info thread is a wrapper around a thread used
 // to store extra data for use in the condition variable
 forall(dtype L | is_blocking_lock(L)) {
+forall(L & | is_blocking_lock(L)) {
         struct info_thread;
 …
 //-----------------------------------------------------------------------------
 // Synchronization Locks
 forall(dtype L | is_blocking_lock(L)) {
+forall(L & | is_blocking_lock(L)) {
         struct condition_variable {
                 // Spin lock used for mutual exclusion
 …
         bool wait( condition_variable(L) & this, L & l, uintptr_t info, Time time );
+}
+//-----------------------------------------------------------------------------
+// Semaphore
+struct semaphore {
+        __spinlock_t lock;
+        int count;
+        __queue_t($thread) waiting;
+};
+void  ?{}(semaphore & this, int count = 1);
+void ^?{}(semaphore & this);
+bool   P (semaphore & this);
+bool   V (semaphore & this);
+bool   V (semaphore & this, unsigned count);

libcfa/src/concurrency/monitor.cfa

-                      r5869cea
+                      r7b91c0e
 static inline [$thread *, int] search_entry_queue( const __waitfor_mask_t &, $monitor * monitors [], __lock_size_t count );
 forall(dtype T | sized( T ))
+forall(T & | sized( T ))
 static inline __lock_size_t insert_unique( T * array [], __lock_size_t & size, T * val );
 static inline __lock_size_t count_max    ( const __waitfor_mask_t & mask );
 …
+}
 forall(dtype T | sized( T ))
+forall(T & | sized( T ))
 static inline __lock_size_t insert_unique( T * array [], __lock_size_t & size, T * val ) {
         if( !val ) return size;

libcfa/src/concurrency/monitor.hfa

-                      r5869cea
+                      r7b91c0e
 #include "stdlib.hfa"
 trait is_monitor(dtype T) {
+trait is_monitor(T &) {
         $monitor * get_monitor( T & );
         void ^?{}( T & mutex );
 …
 void ^?{}( monitor_dtor_guard_t & this );
 static inline forall( dtype T | sized(T) | { void ^?{}( T & mutex ); } )
+static inline forall( T & | sized(T) | { void ^?{}( T & mutex ); } )
 void delete( T * th ) {
         ^(*th){};

libcfa/src/concurrency/mutex.cfa

-                      r5869cea
+                      r7b91c0e
+}
 forall(dtype L | is_lock(L))
+forall(L & | is_lock(L))
 void wait(condition_variable & this, L & l) {
         lock( this.lock __cfaabi_dbg_ctx2 );
 …
 //-----------------------------------------------------------------------------
 // Scopes
 forall(dtype L | is_lock(L))
+forall(L & | is_lock(L))
 void lock_all  ( L * locks[], size_t count) {
         // Sort locks based on addresses
 …
+}
 forall(dtype L | is_lock(L))
+forall(L & | is_lock(L))
 void unlock_all( L * locks[], size_t count) {
         // Lock all

libcfa/src/concurrency/mutex.hfa

-                      r5869cea
+                      r7b91c0e
 void unlock(recursive_mutex_lock & this) __attribute__((deprecated("use concurrency/locks.hfa instead")));
 trait is_lock(dtype L | sized(L)) {
+trait is_lock(L & | sized(L)) {
         void lock  (L &);
         void unlock(L &);
 …
 void wait(condition_variable & this) __attribute__((deprecated("use concurrency/locks.hfa instead")));
 forall(dtype L | is_lock(L))
+forall(L & | is_lock(L))
 void wait(condition_variable & this, L & l) __attribute__((deprecated("use concurrency/locks.hfa instead")));
 //-----------------------------------------------------------------------------
 // Scopes
 forall(dtype L | is_lock(L)) {
+forall(L & | is_lock(L)) {
         #if !defined( __TUPLE_ARRAYS_EXIST__ )
         void lock  ( L * locks [], size_t count);

libcfa/src/concurrency/preemption.cfa

-                      r5869cea
+                      r7b91c0e
+}
+// Prevent preemption since we are about to start terminating things
+void __kernel_abort_lock(void) {
+        signal_block( SIGUSR1 );
+}
 // Raii ctor/dtor for the preemption_scope
 // Used by thread to control when they want to receive preemption signals

libcfa/src/concurrency/thread.cfa

-                      r5869cea
+                      r7b91c0e
+}
 FORALL_DATA_INSTANCE(ThreadCancelled, (dtype thread_t), (thread_t))
+FORALL_DATA_INSTANCE(ThreadCancelled, (thread_t &), (thread_t))
 forall(dtype T)
+forall(T &)
 void copy(ThreadCancelled(T) * dst, ThreadCancelled(T) * src) {
         dst->virtual_table = src->virtual_table;
 …
+}
 forall(dtype T)
+forall(T &)
 const char * msg(ThreadCancelled(T) *) {
         return "ThreadCancelled";
+}
 forall(dtype T)
+forall(T &)
 static void default_thread_cancel_handler(ThreadCancelled(T) & ) {
         abort( "Unhandled thread cancellation.\n" );
+}
 forall(dtype T | is_thread(T) | IS_EXCEPTION(ThreadCancelled, (T)))
+forall(T & | is_thread(T) | IS_EXCEPTION(ThreadCancelled, (T)))
 void ?{}( thread_dtor_guard_t & this,
                 T & thrd, void(*defaultResumptionHandler)(ThreadCancelled(T) &)) {
         $monitor * m = get_monitor(thrd);
+                T & thrd, void(*cancelHandler)(ThreadCancelled(T) &)) {
+        $monitor * m = get_monitor(thrd);
         $thread * desc = get_thread(thrd);
         // Setup the monitor guard
         void (*dtor)(T& mutex this) = ^?{};
         bool join = defaultResumptionHandler != (void(*)(ThreadCancelled(T)&))0;
+        bool join = cancelHandler != (void(*)(ThreadCancelled(T)&))0;
         (this.mg){&m, (void(*)())dtor, join};
 …
+        }
         desc->state = Cancelled;
+        if (!join) {
+                defaultResumptionHandler = default_thread_cancel_handler;
+        }
+        void(*defaultResumptionHandler)(ThreadCancelled(T) &) =
+                join ? cancelHandler : default_thread_cancel_handler;
         ThreadCancelled(T) except;
 …
 //-----------------------------------------------------------------------------
 // Starting and stopping threads
 forall( dtype T | is_thread(T) )
+forall( T & | is_thread(T) )
 void __thrd_start( T & this, void (*main_p)(T &) ) {
         $thread * this_thrd = get_thread(this);
 …
 //-----------------------------------------------------------------------------
 // Support for threads that don't ues the thread keyword
 forall( dtype T | sized(T) | is_thread(T) | { void ?{}(T&); } )
+forall( T & | sized(T) | is_thread(T) | { void ?{}(T&); } )
 void ?{}( scoped(T)& this ) with( this ) {
         handle{};
 …
+}
 forall( dtype T, ttype P | sized(T) | is_thread(T) | { void ?{}(T&, P); } )
+forall( T &, P... | sized(T) | is_thread(T) | { void ?{}(T&, P); } )
 void ?{}( scoped(T)& this, P params ) with( this ) {
         handle{ params };
 …
+}
 forall( dtype T | sized(T) | is_thread(T) )
+forall( T & | sized(T) | is_thread(T) )
 void ^?{}( scoped(T)& this ) with( this ) {
         ^handle{};
 …
 //-----------------------------------------------------------------------------
 forall(dtype T | is_thread(T) | IS_RESUMPTION_EXCEPTION(ThreadCancelled, (T)))
+forall(T & | is_thread(T) | IS_RESUMPTION_EXCEPTION(ThreadCancelled, (T)))
 T & join( T & this ) {
         thread_dtor_guard_t guard = { this, defaultResumptionHandler };

libcfa/src/concurrency/thread.hfa

-                      r5869cea
+                      r7b91c0e
 //-----------------------------------------------------------------------------
 // thread trait
 trait is_thread(dtype T) {
+trait is_thread(T &) {
         void ^?{}(T& mutex this);
         void main(T& this);
 …
 };
 FORALL_DATA_EXCEPTION(ThreadCancelled, (dtype thread_t), (thread_t)) (
+FORALL_DATA_EXCEPTION(ThreadCancelled, (thread_t &), (thread_t)) (
         thread_t * the_thread;
         exception_t * the_exception;
 );
 forall(dtype T)
+forall(T &)
 void copy(ThreadCancelled(T) * dst, ThreadCancelled(T) * src);
 forall(dtype T)
+forall(T &)
 const char * msg(ThreadCancelled(T) *);
 …
 // Inline getters for threads/coroutines/monitors
 forall( dtype T | is_thread(T) )
+forall( T & | is_thread(T) )
 static inline $coroutine* get_coroutine(T & this) __attribute__((const)) { return &get_thread(this)->self_cor; }
 forall( dtype T | is_thread(T) )
+forall( T & | is_thread(T) )
 static inline $monitor  * get_monitor  (T & this) __attribute__((const)) { return &get_thread(this)->self_mon; }
 …
 extern struct cluster * mainCluster;
 forall( dtype T | is_thread(T) )
+forall( T & | is_thread(T) )
 void __thrd_start( T & this, void (*)(T &) );
 …
 };
 forall( dtype T | is_thread(T) | IS_EXCEPTION(ThreadCancelled, (T)) )
+forall( T & | is_thread(T) | IS_EXCEPTION(ThreadCancelled, (T)) )
 void ?{}( thread_dtor_guard_t & this, T & thrd, void(*)(ThreadCancelled(T) &) );
 void ^?{}( thread_dtor_guard_t & this );
 …
 // thread runner
 // Structure that actually start and stop threads
 forall( dtype T | sized(T) | is_thread(T) )
+forall( T & | sized(T) | is_thread(T) )
 struct scoped {
         T handle;
 };
 forall( dtype T | sized(T) | is_thread(T) | { void ?{}(T&); } )
+forall( T & | sized(T) | is_thread(T) | { void ?{}(T&); } )
 void ?{}( scoped(T)& this );
 forall( dtype T, ttype P | sized(T) | is_thread(T) | { void ?{}(T&, P); } )
+forall( T &, P... | sized(T) | is_thread(T) | { void ?{}(T&, P); } )
 void ?{}( scoped(T)& this, P params );
 forall( dtype T | sized(T) | is_thread(T) )
+forall( T & | sized(T) | is_thread(T) )
 void ^?{}( scoped(T)& this );
 …
 void unpark( $thread * this );
 forall( dtype T | is_thread(T) )
+forall( T & | is_thread(T) )
 static inline void unpark( T & this ) { if(!&this) return; unpark( get_thread( this ) );}
 …
 //----------
 // join
 forall( dtype T | is_thread(T) | IS_RESUMPTION_EXCEPTION(ThreadCancelled, (T)) )
+forall( T & | is_thread(T) | IS_RESUMPTION_EXCEPTION(ThreadCancelled, (T)) )
 T & join( T & this );

libcfa/src/containers/list.hfa

-                      r5869cea
+                      r7b91c0e
 #define __DLISTED_MGD_JUSTIMPL(STRUCT)
 forall( dtype tE ) {
+forall( tE & ) {
         struct $mgd_link {
                 tE *elem;
 …
                 (this.is_terminator){ 1 };
+        }
         forall ( otype tInit | { void ?{}( $mgd_link(tE) &, tInit); } )
+        forall ( tInit | { void ?{}( $mgd_link(tE) &, tInit); } )
         static inline void ?=?( $mgd_link(tE) &this, tInit i ) {
                 ^?{}( this );
 …
   __DLISTED_MGD_COMMON(STRUCT, STRUCT, $links)
 trait $dlistable(dtype Tnode, dtype Telem) {
+trait $dlistable(Tnode &, Telem &) {
         $mgd_link(Telem) & $prev_link(Tnode &);
         $mgd_link(Telem) & $next_link(Tnode &);
 …
 };
 forall (dtype Tnode, dtype Telem | $dlistable(Tnode, Telem)) {
+forall (Tnode &, Telem & | $dlistable(Tnode, Telem)) {
         // implemented as a sentinel item in an underlying cicrular list

libcfa/src/containers/maybe.cfa

-                      r5869cea
+                      r7b91c0e
 forall(otype T)
+forall(T)
 void ?{}(maybe(T) & this) {
         this.has_value = false;
+}
 forall(otype T)
+forall(T)
 void ?{}(maybe(T) & this, T value) {
         this.has_value = true;
 …
+}
 forall(otype T)
+forall(T)
 void ?{}(maybe(T) & this, maybe(T) other) {
         this.has_value = other.has_value;
 …
+}
 forall(otype T)
+forall(T)
 maybe(T) ?=?(maybe(T) & this, maybe(T) that) {
         if (this.has_value && that.has_value) {
 …
+}
 forall(otype T)
+forall(T)
 void ^?{}(maybe(T) & this) {
         if (this.has_value) {
 …
+}
 forall(otype T)
+forall(T)
 bool ?!=?(maybe(T) this, zero_t) {
         return this.has_value;
+}
 forall(otype T)
+forall(T)
 maybe(T) maybe_value(T value) {
         return (maybe(T)){value};
+}
 forall(otype T)
+forall(T)
 maybe(T) maybe_none() {
         return (maybe(T)){};
+}
 forall(otype T)
+forall(T)
 bool has_value(maybe(T) * this) {
         return this->has_value;
+}
 forall(otype T)
+forall(T)
 T get(maybe(T) * this) {
         assertf(this->has_value, "attempt to get from maybe without value");
 …
+}
 forall(otype T)
+forall(T)
 void set(maybe(T) * this, T value) {
         if (this->has_value) {
 …
+}
 forall(otype T)
+forall(T)
 void set_none(maybe(T) * this) {
         if (this->has_value) {

libcfa/src/containers/maybe.hfa

-                      r5869cea
+                      r7b91c0e
 // DO NOT USE DIRECTLY!
 forall(otype T)
+forall(T)
 struct maybe {
     bool has_value;
 …
 forall(otype T)
+forall(T)
 void ?{}(maybe(T) & this);
 forall(otype T)
+forall(T)
 void ?{}(maybe(T) & this, T value);
 forall(otype T)
+forall(T)
 void ?{}(maybe(T) & this, maybe(T) other);
 forall(otype T)
+forall(T)
 void ^?{}(maybe(T) & this);
 forall(otype T)
+forall(T)
 maybe(T) ?=?(maybe(T) & this, maybe(T) other);
 forall(otype T)
+forall(T)
 bool ?!=?(maybe(T) this, zero_t);
 /* Waiting for bug#11 to be fixed.
 forall(otype T)
+forall(T)
 maybe(T) maybe_value(T value);
 forall(otype T)
+forall(T)
 maybe(T) maybe_none();
 */
 forall(otype T)
+forall(T)
 bool has_value(maybe(T) * this);
 forall(otype T)
+forall(T)
 T get(maybe(T) * this);
 forall(otype T)
+forall(T)
 void set(maybe(T) * this, T value);
 forall(otype T)
+forall(T)
 void set_none(maybe(T) * this);

libcfa/src/containers/pair.cfa

-                      r5869cea
+                      r7b91c0e
 #include <containers/pair.hfa>
 forall(otype R, otype S
+forall(R, S
         | { int ?==?(R, R); int ?<?(R, R); int ?<?(S, S); })
 int ?<?(pair(R, S) p, pair(R, S) q) {
 …
+}
 forall(otype R, otype S
+forall(R, S
         | { int ?==?(R, R); int ?<?(R, R); int ?<=?(S, S); })
 int ?<=?(pair(R, S) p, pair(R, S) q) {
 …
+}
 forall(otype R, otype S | { int ?==?(R, R); int ?==?(S, S); })
+forall(R, S | { int ?==?(R, R); int ?==?(S, S); })
 int ?==?(pair(R, S) p, pair(R, S) q) {
         return p.first == q.first && p.second == q.second;
+}
 forall(otype R, otype S | { int ?!=?(R, R); int ?!=?(S, S); })
+forall(R, S | { int ?!=?(R, R); int ?!=?(S, S); })
 int ?!=?(pair(R, S) p, pair(R, S) q) {
         return p.first != q.first || p.second != q.second;
+}
 forall(otype R, otype S
+forall(R, S
         | { int ?==?(R, R); int ?>?(R, R); int ?>?(S, S); })
 int ?>?(pair(R, S) p, pair(R, S) q) {
 …
+}
 forall(otype R, otype S
+forall(R, S
         | { int ?==?(R, R); int ?>?(R, R); int ?>=?(S, S); })
 int ?>=?(pair(R, S) p, pair(R, S) q) {

libcfa/src/containers/pair.hfa

-                      r5869cea
+                      r7b91c0e
 #pragma once
 forall(otype R, otype S) struct pair {
+forall(R, S) struct pair {
         R first;
         S second;
 };
 forall(otype R, otype S
+forall(R, S
         | { int ?==?(R, R); int ?<?(R, R); int ?<?(S, S); })
 int ?<?(pair(R, S) p, pair(R, S) q);
 forall(otype R, otype S
+forall(R, S
         | { int ?==?(R, R); int ?<?(R, R); int ?<=?(S, S); })
 int ?<=?(pair(R, S) p, pair(R, S) q);
 forall(otype R, otype S | { int ?==?(R, R); int ?==?(S, S); })
+forall(R, S | { int ?==?(R, R); int ?==?(S, S); })
 int ?==?(pair(R, S) p, pair(R, S) q);
 forall(otype R, otype S | { int ?!=?(R, R); int ?!=?(S, S); })
+forall(R, S | { int ?!=?(R, R); int ?!=?(S, S); })
 int ?!=?(pair(R, S) p, pair(R, S) q);
 forall(otype R, otype S
+forall(R, S
         | { int ?==?(R, R); int ?>?(R, R); int ?>?(S, S); })
 int ?>?(pair(R, S) p, pair(R, S) q);
 forall(otype R, otype S
+forall(R, S
         | { int ?==?(R, R); int ?>?(R, R); int ?>=?(S, S); })
 int ?>=?(pair(R, S) p, pair(R, S) q);

libcfa/src/containers/result.cfa

-                      r5869cea
+                      r7b91c0e
 forall(otype T, otype E)
+forall(T, E)
 void ?{}(result(T, E) & this) {
         this.has_value = false;
 …
+}
 forall(otype T, otype E)
+forall(T, E)
 void ?{}(result(T, E) & this, one_t, T value) {
         this.has_value = true;
 …
+}
 forall(otype T, otype E)
+forall(T, E)
 void ?{}(result(T, E) & this, zero_t, E error) {
         this.has_value = false;
 …
+}
 forall(otype T, otype E)
+forall(T, E)
 void ?{}(result(T, E) & this, result(T, E) other) {
         this.has_value = other.has_value;
 …
+}
 forall(otype T, otype E)
+forall(T, E)
 result(T, E) ?=?(result(T, E) & this, result(T, E) that) {
         if (this.has_value && that.has_value) {
 …
+}
 forall(otype T, otype E)
+forall(T, E)
 void ^?{}(result(T, E) & this) {
         if (this.has_value) {
 …
+}
 forall(otype T, otype E)
+forall(T, E)
 bool ?!=?(result(T, E) this, zero_t) {
         return this.has_value;
+}
 forall(otype T, otype E)
+forall(T, E)
 result(T, E) result_value(T value) {
         return (result(T, E)){1, value};
+}
 forall(otype T, otype E)
+forall(T, E)
 result(T, E) result_error(E error) {
         return (result(T, E)){0, error};
+}
 forall(otype T, otype E)
+forall(T, E)
 bool has_value(result(T, E) * this) {
         return this->has_value;
+}
 forall(otype T, otype E)
+forall(T, E)
 T get(result(T, E) * this) {
         assertf(this->has_value, "attempt to get from result without value");
 …
+}
 forall(otype T, otype E)
+forall(T, E)
 E get_error(result(T, E) * this) {
         assertf(!this->has_value, "attempt to get from result without error");
 …
+}
 forall(otype T, otype E)
+forall(T, E)
 void set(result(T, E) * this, T value) {
         if (this->has_value) {
 …
+}
 forall(otype T, otype E)
+forall(T, E)
 void set_error(result(T, E) * this, E error) {
         if (this->has_value) {

libcfa/src/containers/result.hfa

-                      r5869cea
+                      r7b91c0e
 // DO NOT USE DIRECTLY!
 forall(otype T, otype E)
+forall(T, E)
 union inner_result{
         T value;
 …
 };
 forall(otype T, otype E)
+forall(T, E)
 struct result {
         bool has_value;
 …
 forall(otype T, otype E)
+forall(T, E)
 void ?{}(result(T, E) & this);
 forall(otype T, otype E)
+forall(T, E)
 void ?{}(result(T, E) & this, one_t, T value);
 forall(otype T, otype E)
+forall(T, E)
 void ?{}(result(T, E) & this, zero_t, E error);
 forall(otype T, otype E)
+forall(T, E)
 void ?{}(result(T, E) & this, result(T, E) other);
 forall(otype T, otype E)
+forall(T, E)
 void ^?{}(result(T, E) & this);
 forall(otype T, otype E)
+forall(T, E)
 result(T, E) ?=?(result(T, E) & this, result(T, E) other);
 forall(otype T, otype E)
+forall(T, E)
 bool ?!=?(result(T, E) this, zero_t);
 /* Wating for bug#11 to be fixed.
 forall(otype T, otype E)
+forall(T, E)
 result(T, E) result_value(T value);
 forall(otype T, otype E)
+forall(T, E)
 result(T, E) result_error(E error);
 */
 forall(otype T, otype E)
+forall(T, E)
 bool has_value(result(T, E) * this);
 forall(otype T, otype E)
+forall(T, E)
 T get(result(T, E) * this);
 forall(otype T, otype E)
+forall(T, E)
 E get_error(result(T, E) * this);
 forall(otype T, otype E)
+forall(T, E)
 void set(result(T, E) * this, T value);
 forall(otype T, otype E)
+forall(T, E)
 void set_error(result(T, E) * this, E error);

libcfa/src/containers/stackLockFree.hfa

-                      r5869cea
+                      r7b91c0e
 // Created On       : Wed May 13 20:58:58 2020
 // Last Modified By : Peter A. Buhr
 // Last Modified On : Sun Jun 14 13:25:09 2020
 // Update Count     : 64
+// Last Modified On : Wed Jan 20 20:40:03 2021
+// Update Count     : 67
 //
 …
 #include <stdint.h>
 forall( dtype T )
+forall( T & )
 union Link {
         struct {                                                                                        // 32/64-bit x 2
 …
 }; // Link
 forall( otype T | sized(T) | { Link(T) * ?`next( T * ); } ) {
+forall( T | sized(T) | { Link(T) * ?`next( T * ); } ) {
         struct StackLF {
                 Link(T) stack;
 …
                 void push( StackLF(T) & this, T & n ) with(this) {
                         *( &n )`next = stack;                                   // atomic assignment unnecessary, or use CAA
+                        *( &n )`next = stack;                                           // atomic assignment unnecessary, or use CAA
                         for () {                                                                        // busy wait
                           if ( __atomic_compare_exchange_n( &stack.atom, &( &n )`next->atom, (Link(T))@{ {&n, ( &n )`next->count + 1} }.atom, false, __ATOMIC_SEQ_CST, __ATOMIC_SEQ_CST ) ) break; // attempt to update top node
 …
+                                }
                                 if( next == 0p ) return false;
                                 link = (next)`next;
+                                link = ( next )`next;
+                        }
+                }

libcfa/src/containers/vector.cfa

r5869cea	r7b91c0e
18	18	#include <stdlib.hfa>

Context Navigation

Legend: