Changeset dafbde8

doc/bibliography/pl.bib

-              r92bfda0
+              rdafbde8
     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/fangren_yu_COOP_F20/Report.tex

-              r92bfda0
+              rdafbde8
 \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}

driver/cfa.cc

-              r92bfda0
+              rdafbde8
 // 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

-              r92bfda0
+              rdafbde8
 // 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

r92bfda0	rdafbde8
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

-              r92bfda0
+              rdafbde8
 // ------------------------------------------------------------
 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

-              r92bfda0
+              rdafbde8
 _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

-              r92bfda0
+              rdafbde8
         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

-              r92bfda0
+              rdafbde8
 #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

-              r92bfda0
+              rdafbde8
         // // 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 * ); } ) {
+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

-              r92bfda0
+              rdafbde8
 #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/queue.hfa

-              r92bfda0
+              rdafbde8
 // 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

-              r92bfda0
+              rdafbde8
         // // 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

-              r92bfda0
+              rdafbde8
 // 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

r92bfda0	rdafbde8
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

-              r92bfda0
+              rdafbde8
 [ 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

-              r92bfda0
+              rdafbde8
 //-----------------------------------------------------------------------------
 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

-              r92bfda0
+              rdafbde8
 //-----------------------------------------------------------------------------
 // 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

-              r92bfda0
+              rdafbde8
 #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/setup.cfa

-              r92bfda0
+              rdafbde8
         static struct {
+                pthread_t       thrd;    // pthread handle to io poller thread
+                void *          stack;   // pthread stack for io poller thread
+                int             epollfd; // file descriptor to the epoll instance
+                volatile bool   run;     // Whether or not to continue
+                volatile size_t epoch;   // Epoch used for memory reclamation
+                      pthread_t  thrd;    // pthread handle to io poller thread
+                      void *     stack;   // pthread stack for io poller thread
+                      int        epollfd; // file descriptor to the epoll instance
+                volatile     bool run;     // Whether or not to continue
+                volatile     bool stopped; // Whether the poller has finished running
+                volatile uint64_t epoch;   // Epoch used for memory reclamation
         } iopoll;
 …
                 __cfadbg_print_safe(io_core, "Kernel : Starting io poller thread\n" );
+                iopoll.run = true;
+                iopoll.stack = __create_pthread( &iopoll.thrd, iopoll_loop, 0p );
+                iopoll.epoch = 0;
+                iopoll.stack   = __create_pthread( &iopoll.thrd, iopoll_loop, 0p );
+                iopoll.run     = true;
+                iopoll.stopped = false;
+                iopoll.epoch   = 0;
+        }
 …
+                        }
+                }
+                __atomic_store_n(&iopoll.stopped, true, __ATOMIC_SEQ_CST);
                 __cfadbg_print_safe(io_core, "Kernel : IO poller thread stopping\n" );
 …
                 // Wait for the next epoch
                 while(curr == __atomic_load_n(&iopoll.epoch, __ATOMIC_RELAXED)) yield();
+                while(curr == iopoll.epoch && !iopoll.stopped) Pause();
+        }

libcfa/src/concurrency/locks.cfa

-              r92bfda0
+              rdafbde8
 //-----------------------------------------------------------------------------
 // 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 ){

libcfa/src/concurrency/locks.hfa

-              r92bfda0
+              rdafbde8
 //-----------------------------------------------------------------------------
 // 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

libcfa/src/concurrency/monitor.cfa

-              r92bfda0
+              rdafbde8
 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

-              r92bfda0
+              rdafbde8
 #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

-              r92bfda0
+              rdafbde8
+}
 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

-              r92bfda0
+              rdafbde8
 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/thread.cfa

-              r92bfda0
+              rdafbde8
+}
 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

-              r92bfda0
+              rdafbde8
 //-----------------------------------------------------------------------------
 // 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

-              r92bfda0
+              rdafbde8
 #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

-              r92bfda0
+              rdafbde8
 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

-              r92bfda0
+              rdafbde8
 // 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

-              r92bfda0
+              rdafbde8
 #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

-              r92bfda0
+              rdafbde8
 #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

-              r92bfda0
+              rdafbde8
 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

-              r92bfda0
+              rdafbde8
 // 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

-              r92bfda0
+              rdafbde8
 #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;

libcfa/src/containers/vector.cfa

r92bfda0	rdafbde8
18	18	#include <stdlib.hfa>
19	19
20		forall(~~otype T, otype~~ allocator_t \| allocator_c(T, allocator_t))
	20	forall(T, allocator_t \| allocator_c(T, allocator_t))
21	21	void copy_internal(vector(T, allocator_t)* this, vector(T, allocator_t)* other);
22	22
23	23	//------------------------------------------------------------------------------
24	24	//Initialization
25		forall(~~otype T, otype~~ allocator_t \| allocator_c(T, allocator_t))
	25	forall(T, allocator_t \| allocator_c(T, allocator_t))
26	26	void ?{}(vector(T, allocator_t)& this)
27	27	{
…	…
30	30	}
31	31
32		forall(~~otype T, otype~~ allocator_t \| allocator_c(T, allocator_t))
	32	forall(T, allocator_t \| allocator_c(T, allocator_t))
33	33	void ?{}(vector(T, allocator_t)& this, vector(T, allocator_t) rhs)
34	34	{
…	…
37	37	}
38	38
39		// forall(~~otype T, otype~~ allocator_t \| allocator_c(T, allocator_t))
	39	// forall(T, allocator_t \| allocator_c(T, allocator_t))
40	40	// vector(T, allocator_t) ?=?(vector(T, allocator_t)* this, vector(T, allocator_t) rhs)
41	41	// {
…	…
45	45	// }
46	46
47		forall(~~otype T, otype~~ allocator_t \| allocator_c(T, allocator_t))
	47	forall(T, allocator_t \| allocator_c(T, allocator_t))
48	48	void ^?{}(vector(T, allocator_t)& this)
49	49	{
…	…
54	54	//------------------------------------------------------------------------------
55	55	//Modifiers
56		forall(~~otype T, otype~~ allocator_t \| allocator_c(T, allocator_t))
	56	forall(T, allocator_t \| allocator_c(T, allocator_t))
57	57	void push_back(vector(T, allocator_t)* this, T value)
58	58	{
…	…
62	62	}
63	63
64		forall(~~otype T, otype~~ allocator_t \| allocator_c(T, allocator_t))
	64	forall(T, allocator_t \| allocator_c(T, allocator_t))
65	65	void pop_back(vector(T, allocator_t)* this)
66	66	{
…	…
69	69	}
70	70
71		forall(~~otype T, otype~~ allocator_t \| allocator_c(T, allocator_t))
	71	forall(T, allocator_t \| allocator_c(T, allocator_t))
72	72	void clear(vector(T, allocator_t)* this)
73	73	{
…	…
82	82	//Internal Helpers
83	83
84		forall(~~otype T, otype~~ allocator_t \| allocator_c(T, allocator_t))
	84	forall(T, allocator_t \| allocator_c(T, allocator_t))
85	85	void copy_internal(vector(T, allocator_t)* this, vector(T, allocator_t)* other)
86	86	{
…	…
93	93	//------------------------------------------------------------------------------
94	94	//Allocator
95		forall(~~otype~~ T)
	95	forall(T)
96	96	void ?{}(heap_allocator(T)& this)
97	97	{
…	…
100	100	}
101	101
102		forall(~~otype~~ T)
	102	forall(T)
103	103	void ?{}(heap_allocator(T)& this, heap_allocator(T) rhs)
104	104	{
…	…
107	107	}
108	108
109		forall(~~otype~~ T)
	109	forall(T)
110	110	heap_allocator(T) ?=?(heap_allocator(T)& this, heap_allocator(T) rhs)
111	111	{
…	…
115	115	}
116	116
117		forall(~~otype~~ T)
	117	forall(T)
118	118	void ^?{}(heap_allocator(T)& this)
119	119	{
…	…
121	121	}
122	122
123		forall(~~otype~~ T)
	123	forall(T)
124	124	inline void realloc_storage(heap_allocator(T)* this, size_t size)
125	125	{

libcfa/src/containers/vector.hfa

-              r92bfda0
+              rdafbde8
 //------------------------------------------------------------------------------
 //Allocator
 forall(otype T)
+forall(T)
 struct heap_allocator
+{
 …
 };
 forall(otype T)
+forall(T)
 void ?{}(heap_allocator(T)& this);
 forall(otype T)
+forall(T)
 void ?{}(heap_allocator(T)& this, heap_allocator(T) rhs);
 forall(otype T)
+forall(T)
 heap_allocator(T) ?=?(heap_allocator(T)& this, heap_allocator(T) rhs);
 forall(otype T)
+forall(T)
 void ^?{}(heap_allocator(T)& this);
 forall(otype T)
+forall(T)
 void realloc_storage(heap_allocator(T)* this, size_t size);
 forall(otype T)
+forall(T)
 static inline T* data(heap_allocator(T)* this)
+{
 …
 //------------------------------------------------------------------------------
 //Declaration
 trait allocator_c(otype T, otype allocator_t)
+trait allocator_c(T, allocator_t)
+{
         void realloc_storage(allocator_t*, size_t);
 …
 };
 forall(otype T, otype allocator_t = heap_allocator(T) | allocator_c(T, allocator_t))
+forall(T, allocator_t = heap_allocator(T) | allocator_c(T, allocator_t))
 struct vector;
 //------------------------------------------------------------------------------
 //Initialization
 forall(otype T, otype allocator_t | allocator_c(T, allocator_t))
+forall(T, allocator_t | allocator_c(T, allocator_t))
 void ?{}(vector(T, allocator_t)& this);
 forall(otype T, otype allocator_t | allocator_c(T, allocator_t))
+forall(T, allocator_t | allocator_c(T, allocator_t))
 void ?{}(vector(T, allocator_t)& this, vector(T, allocator_t) rhs);
 forall(otype T, otype allocator_t | allocator_c(T, allocator_t))
+forall(T, allocator_t | allocator_c(T, allocator_t))
 vector(T, allocator_t) ?=?(vector(T, allocator_t)& this, vector(T, allocator_t) rhs);
 forall(otype T, otype allocator_t | allocator_c(T, allocator_t))
+forall(T, allocator_t | allocator_c(T, allocator_t))
 void ^?{}(vector(T, allocator_t)& this);
 forall(otype T, otype allocator_t = heap_allocator(T) | allocator_c(T, allocator_t))
+forall(T, allocator_t = heap_allocator(T) | allocator_c(T, allocator_t))
 struct vector
+{
 …
 //------------------------------------------------------------------------------
 //Capacity
 forall(otype T, otype allocator_t | allocator_c(T, allocator_t))
+forall(T, allocator_t | allocator_c(T, allocator_t))
 static inline bool empty(vector(T, allocator_t)* this)
+{
 …
+}
 forall(otype T, otype allocator_t | allocator_c(T, allocator_t))
+forall(T, allocator_t | allocator_c(T, allocator_t))
 static inline size_t size(vector(T, allocator_t)* this)
+{
 …
+}
 forall(otype T, otype allocator_t | allocator_c(T, allocator_t))
+forall(T, allocator_t | allocator_c(T, allocator_t))
 static inline void reserve(vector(T, allocator_t)* this, size_t size)
+{
 …
 //------------------------------------------------------------------------------
 //Element access
 forall(otype T, otype allocator_t | allocator_c(T, allocator_t))
+forall(T, allocator_t | allocator_c(T, allocator_t))
 static inline T at(vector(T, allocator_t)* this, size_t index)
+{
 …
+}
 forall(otype T, otype allocator_t | allocator_c(T, allocator_t))
+forall(T, allocator_t | allocator_c(T, allocator_t))
 static inline T ?[?](vector(T, allocator_t)* this, size_t index)
+{
 …
+}
 forall(otype T, otype allocator_t | allocator_c(T, allocator_t))
+forall(T, allocator_t | allocator_c(T, allocator_t))
 static inline T front(vector(T, allocator_t)* this)
+{
 …
+}
 forall(otype T, otype allocator_t | allocator_c(T, allocator_t))
+forall(T, allocator_t | allocator_c(T, allocator_t))
 static inline T back(vector(T, allocator_t)* this)
+{
 …
 //------------------------------------------------------------------------------
 //Modifiers
 forall(otype T, otype allocator_t | allocator_c(T, allocator_t))
+forall(T, allocator_t | allocator_c(T, allocator_t))
 void push_back(vector(T, allocator_t)* this, T value);
 forall(otype T, otype allocator_t | allocator_c(T, allocator_t))
+forall(T, allocator_t | allocator_c(T, allocator_t))
 void pop_back(vector(T, allocator_t)* this);
 forall(otype T, otype allocator_t | allocator_c(T, allocator_t))
+forall(T, allocator_t | allocator_c(T, allocator_t))
 void clear(vector(T, allocator_t)* this);
 //------------------------------------------------------------------------------
 //Iterators
 forall(otype T, otype allocator_t | allocator_c(T, allocator_t))
+forall(T, allocator_t | allocator_c(T, allocator_t))
 static inline T* begin(vector(T, allocator_t)* this)
+{
 …
+}
 // forall(otype T, otype allocator_t | allocator_c(T, allocator_t))
+// forall(T, allocator_t | allocator_c(T, allocator_t))
 // static inline const T* cbegin(const vector(T, allocator_t)* this)
 // {
 …
 // }
 forall(otype T, otype allocator_t | allocator_c(T, allocator_t))
+forall(T, allocator_t | allocator_c(T, allocator_t))
 static inline T* end(vector(T, allocator_t)* this)
+{
 …
+}
 // forall(otype T, otype allocator_t | allocator_c(T, allocator_t))
+// forall(T, allocator_t | allocator_c(T, allocator_t))
 // static inline const T* cend(const vector(T, allocator_t)* this)
 // {

libcfa/src/exception.h

-              r92bfda0
+              rdafbde8
 // implemented in the .c file either so they all have to be inline.
 trait is_exception(dtype exceptT, dtype virtualT) {
+trait is_exception(exceptT &, virtualT &) {
         /* The first field must be a pointer to a virtual table.
          * That virtual table must be a decendent of the base exception virtual table.
 …
 };
 trait is_termination_exception(dtype exceptT, dtype virtualT | is_exception(exceptT, virtualT)) {
+trait is_termination_exception(exceptT &, virtualT & | is_exception(exceptT, virtualT)) {
         void defaultTerminationHandler(exceptT &);
 };
 trait is_resumption_exception(dtype exceptT, dtype virtualT | is_exception(exceptT, virtualT)) {
+trait is_resumption_exception(exceptT &, virtualT & | is_exception(exceptT, virtualT)) {
         void defaultResumptionHandler(exceptT &);
 };
 forall(dtype exceptT, dtype virtualT | is_termination_exception(exceptT, virtualT))
+forall(exceptT &, virtualT & | is_termination_exception(exceptT, virtualT))
 static inline void $throw(exceptT & except) {
         __cfaehm_throw_terminate(
 …
+}
 forall(dtype exceptT, dtype virtualT | is_resumption_exception(exceptT, virtualT))
+forall(exceptT &, virtualT & | is_resumption_exception(exceptT, virtualT))
 static inline void $throwResume(exceptT & except) {
         __cfaehm_throw_resume(
 …
+}
 forall(dtype exceptT, dtype virtualT | is_exception(exceptT, virtualT))
+forall(exceptT &, virtualT & | is_exception(exceptT, virtualT))
 static inline void cancel_stack(exceptT & except) __attribute__((noreturn)) {
         __cfaehm_cancel_stack( (exception_t *)&except );
+}
 forall(dtype exceptT, dtype virtualT | is_exception(exceptT, virtualT))
+forall(exceptT &, virtualT & | is_exception(exceptT, virtualT))
 static inline void defaultTerminationHandler(exceptT & except) {
         return cancel_stack( except );
+}
 forall(dtype exceptT, dtype virtualT | is_exception(exceptT, virtualT))
+forall(exceptT &, virtualT & | is_exception(exceptT, virtualT))
 static inline void defaultResumptionHandler(exceptT & except) {
         throw except;

libcfa/src/executor.cfa

r92bfda0	rdafbde8
7	7	#include <containers/list.hfa>
8	8
9		forall( ~~dtype T~~ \| $dlistable(T, T) ) {
	9	forall( T & \| $dlistable(T, T) ) {
10	10	monitor Buffer { // unbounded buffer
11	11	dlist( T, T ) queue; // unbounded list of work requests

libcfa/src/gmp.hfa

-              r92bfda0
+              rdafbde8
         // I/O
         forall( dtype istype | istream( istype ) )
+        forall( istype & | istream( istype ) )
                 istype & ?|?( istype & is, Int & mp ) {
                 gmp_scanf( "%Zd", &mp );
 …
         } // ?|?
         forall( dtype ostype | ostream( ostype ) ) {
+        forall( ostype & | ostream( ostype ) ) {
                 ostype & ?|?( ostype & os, Int mp ) {
                         if ( $sepPrt( os ) ) fmt( os, "%s", $sepGetCur( os ) );

libcfa/src/iostream.cfa

-              r92bfda0
+              rdafbde8
 forall( dtype ostype | ostream( ostype ) ) {
+forall( ostype & | ostream( ostype ) ) {
         ostype & ?|?( ostype & os, bool b ) {
                 if ( $sepPrt( os ) ) fmt( os, "%s", $sepGetCur( os ) );
 …
 // tuples
 forall( dtype ostype, otype T, ttype Params | writeable( T, ostype ) | { ostype & ?|?( ostype &, Params ); } ) {
+forall( ostype &, T, Params... | writeable( T, ostype ) | { ostype & ?|?( ostype &, Params ); } ) {
         ostype & ?|?( ostype & os, T arg, Params rest ) {
                 (ostype &)(os | arg);                                                   // print first argument
 …
 // writes the range [begin, end) to the given stream
 forall( dtype ostype, otype elt_type | writeable( elt_type, ostype ), otype iterator_type | iterator( iterator_type, elt_type ) ) {
+forall( ostype &, elt_type | writeable( elt_type, ostype ), iterator_type | iterator( iterator_type, elt_type ) ) {
         void write( iterator_type begin, iterator_type end, ostype & os ) {
                 void print( elt_type i ) { os | i; }
 …
 // Default prefix for non-decimal prints is 0b, 0, 0x.
 #define IntegralFMTImpl( T, IFMTNP, IFMTP ) \
 forall( dtype ostype | ostream( ostype ) ) { \
+forall( ostype & | ostream( ostype ) ) { \
         ostype & ?|?( ostype & os, _Ostream_Manip(T) f ) { \
                 if ( $sepPrt( os ) ) fmt( os, "%s", $sepGetCur( os ) ); \
 …
 // Default prefix for non-decimal prints is 0b, 0, 0x.
 #define IntegralFMTImpl128( T, SIGNED, CODE, IFMTNP, IFMTP ) \
 forall( dtype ostype | ostream( ostype ) ) \
+forall( ostype & | ostream( ostype ) ) \
 static void base10_128( ostype & os, _Ostream_Manip(T) f ) { \
         if ( f.val > UINT64_MAX ) { \
 …
         } /* if */ \
 } /* base10_128 */ \
 forall( dtype ostype | ostream( ostype ) ) { \
+forall( ostype & | ostream( ostype ) ) { \
         ostype & ?|?( ostype & os, _Ostream_Manip(T) f ) { \
                 if ( $sepPrt( os ) ) fmt( os, "%s", $sepGetCur( os ) ); \
 …
 #if defined( __SIZEOF_INT128__ )
 // Default prefix for non-decimal prints is 0b, 0, 0x.
 forall( dtype ostype | ostream( ostype ) )
+forall( ostype & | ostream( ostype ) )
 static inline void base_128( ostype & os, unsigned int128 val, unsigned int128 power, _Ostream_Manip(uint64_t) & f, unsigned int maxdig, unsigned int bits, unsigned int cnt = 0 ) {
         int wd = 1;                                                                                     // f.wd is never 0 because 0 implies left-pad
 …
 #define IntegralFMTImpl128( T ) \
 forall( dtype ostype | ostream( ostype ) ) { \
+forall( ostype & | ostream( ostype ) ) { \
         ostype & ?|?( ostype & os, _Ostream_Manip(T) f ) { \
                 _Ostream_Manip(uint64_t) fmt; \
 …
 #define FloatingPointFMTImpl( T, DFMTNP, DFMTP ) \
 forall( dtype ostype | ostream( ostype ) ) { \
+forall( ostype & | ostream( ostype ) ) { \
         ostype & ?|?( ostype & os, _Ostream_Manip(T) f ) { \
                 if ( $sepPrt( os ) ) fmt( os, "%s", $sepGetCur( os ) ); \
 …
 // *********************************** character ***********************************
 forall( dtype ostype | ostream( ostype ) ) {
+forall( ostype & | ostream( ostype ) ) {
         ostype & ?|?( ostype & os, _Ostream_Manip(char) f ) {
                 if ( f.base != 'c' ) {                                                  // bespoke binary/octal/hex format
 …
 // *********************************** C string ***********************************
 forall( dtype ostype | ostream( ostype ) ) {
+forall( ostype & | ostream( ostype ) ) {
         ostype & ?|?( ostype & os, _Ostream_Manip(const char *) f ) {
                 if ( ! f.val ) return os;                                               // null pointer ?
 …
 forall( dtype istype | istream( istype ) ) {
+forall( istype & | istream( istype ) ) {
         istype & ?|?( istype & is, bool & b ) {
                 char val[6];
 …
 // *********************************** manipulators ***********************************
 forall( dtype istype | istream( istype ) )
+forall( istype & | istream( istype ) )
 istype & ?|?( istype & is, _Istream_Cstr f ) {
         // skip xxx
 …
 } // ?|?
 forall( dtype istype | istream( istype ) )
+forall( istype & | istream( istype ) )
 istype & ?|?( istype & is, _Istream_Char f ) {
         fmt( is, "%*c" );                                                                       // argument variable unused
 …
 #define InputFMTImpl( T, CODE ) \
 forall( dtype istype | istream( istype ) ) \
+forall( istype & | istream( istype ) ) \
 istype & ?|?( istype & is, _Istream_Manip(T) f ) { \
         enum { size = 16 }; \
 …
 InputFMTImpl( long double, "Lf" )
 forall( dtype istype | istream( istype ) )
+forall( istype & | istream( istype ) )
 istype & ?|?( istype & is, _Istream_Manip(float _Complex) fc ) {
         float re, im;
 …
 } // ?|?
 forall( dtype istype | istream( istype ) )
+forall( istype & | istream( istype ) )
 istype & ?|?( istype & is, _Istream_Manip(double _Complex) dc ) {
         double re, im;
 …
 } // ?|?
 forall( dtype istype | istream( istype ) )
+forall( istype & | istream( istype ) )
 istype & ?|?( istype & is, _Istream_Manip(long double _Complex) ldc ) {
         long double re, im;

libcfa/src/iostream.hfa

-              r92bfda0
+              rdafbde8
 trait ostream( dtype ostype ) {
+trait ostream( ostype & ) {
         // private
         bool $sepPrt( ostype & );                                                       // get separator state (on/off)
 …
 }; // ostream
 // trait writeable( otype T ) {
 //      forall( dtype ostype | ostream( ostype ) ) ostype & ?|?( ostype &, T );
+// trait writeable( T ) {
+//      forall( ostype & | ostream( ostype ) ) ostype & ?|?( ostype &, T );
 // }; // writeable
 trait writeable( otype T, dtype ostype | ostream( ostype ) ) {
+trait writeable( T, ostype & | ostream( ostype ) ) {
         ostype & ?|?( ostype &, T );
 }; // writeable
 …
 // implement writable for intrinsic types
 forall( dtype ostype | ostream( ostype ) ) {
+forall( ostype & | ostream( ostype ) ) {
         ostype & ?|?( ostype &, bool );
         void ?|?( ostype &, bool );
 …
 // tuples
 forall( dtype ostype, otype T, ttype Params | writeable( T, ostype ) | { ostype & ?|?( ostype &, Params ); } ) {
+forall( ostype &, T, Params... | writeable( T, ostype ) | { ostype & ?|?( ostype &, Params ); } ) {
         ostype & ?|?( ostype & os, T arg, Params rest );
         void ?|?( ostype & os, T arg, Params rest );
 …
 // writes the range [begin, end) to the given stream
 forall( dtype ostype, otype elt_type | writeable( elt_type, ostype ), otype iterator_type | iterator( iterator_type, elt_type ) ) {
+forall( ostype &, elt_type | writeable( elt_type, ostype ), iterator_type | iterator( iterator_type, elt_type ) ) {
         void write( iterator_type begin, iterator_type end, ostype & os );
         void write_reverse( iterator_type begin, iterator_type end, ostype & os );
 …
 // *********************************** manipulators ***********************************
 forall( otype T )
+forall( T )
 struct _Ostream_Manip {
         T val;                                                                                          // polymorphic base-type
 …
         _Ostream_Manip(T) & sign( _Ostream_Manip(T) & fmt ) { fmt.flags.sign = true; return fmt; } \
 } /* distribution */ \
 forall( dtype ostype | ostream( ostype ) ) { \
+forall( ostype & | ostream( ostype ) ) { \
         ostype & ?|?( ostype & os, _Ostream_Manip(T) f ); \
         void ?|?( ostype & os, _Ostream_Manip(T) f ); \
 …
         _Ostream_Manip(T) & nodp( _Ostream_Manip(T) & fmt ) { fmt.flags.nobsdp = true; return fmt; } \
 } /* distribution */ \
 forall( dtype ostype | ostream( ostype ) ) { \
+forall( ostype & | ostream( ostype ) ) { \
         ostype & ?|?( ostype & os, _Ostream_Manip(T) f ); \
         void ?|?( ostype & os, _Ostream_Manip(T) f ); \
 …
         _Ostream_Manip(char) & nobase( _Ostream_Manip(char) & fmt ) { fmt.flags.nobsdp = true; return fmt; }
 } // distribution
 forall( dtype ostype | ostream( ostype ) ) {
+forall( ostype & | ostream( ostype ) ) {
         ostype & ?|?( ostype & os, _Ostream_Manip(char) f );
         void ?|?( ostype & os, _Ostream_Manip(char) f );
 …
         _Ostream_Manip(const char *) & nobase( _Ostream_Manip(const char *) & fmt ) { fmt.flags.nobsdp = true; return fmt; }
 } // distribution
 forall( dtype ostype | ostream( ostype ) ) {
+forall( ostype & | ostream( ostype ) ) {
         ostype & ?|?( ostype & os, _Ostream_Manip(const char *) f );
         void ?|?( ostype & os, _Ostream_Manip(const char *) f );
 …
 trait istream( dtype istype ) {
+trait istream( istype & ) {
         void nlOn( istype & );                                                          // read newline
         void nlOff( istype & );                                                         // scan newline
 …
 }; // istream
 trait readable( otype T ) {
         forall( dtype istype | istream( istype ) ) istype & ?|?( istype &, T );
+trait readable( T ) {
+        forall( istype & | istream( istype ) ) istype & ?|?( istype &, T );
 }; // readable
 forall( dtype istype | istream( istype ) ) {
+forall( istype & | istream( istype ) ) {
         istype & ?|?( istype &, bool & );
 …
         _Istream_Cstr & wdi( unsigned int w, _Istream_Cstr & fmt ) { fmt.wd = w; return fmt; }
 } // distribution
 forall( dtype istype | istream( istype ) ) istype & ?|?( istype & is, _Istream_Cstr f );
+forall( istype & | istream( istype ) ) istype & ?|?( istype & is, _Istream_Cstr f );
 struct _Istream_Char {
 …
         _Istream_Char & ignore( _Istream_Char & fmt ) { fmt.ignore = true; return fmt; }
 } // distribution
 forall( dtype istype | istream( istype ) ) istype & ?|?( istype & is, _Istream_Char f );
 forall( dtype T | sized( T ) )
+forall( istype & | istream( istype ) ) istype & ?|?( istype & is, _Istream_Char f );
+forall( T & | sized( T ) )
 struct _Istream_Manip {
         T & val;                                                                                        // polymorphic base-type
 …
         _Istream_Manip(T) & wdi( unsigned int w, _Istream_Manip(T) & fmt ) { fmt.wd = w; return fmt; } \
 } /* distribution */ \
 forall( dtype istype | istream( istype ) ) { \
+forall( istype & | istream( istype ) ) { \
         istype & ?|?( istype & is, _Istream_Manip(T) f ); \
 } // ?|?
 …
 #include <time_t.hfa>                                                                   // Duration (constructors) / Time (constructors)
 forall( dtype ostype | ostream( ostype ) ) {
+forall( ostype & | ostream( ostype ) ) {
         ostype & ?|?( ostype & os, Duration dur );
         void ?|?( ostype & os, Duration dur );

libcfa/src/iterator.cfa

-              r92bfda0
+              rdafbde8
 #include "iterator.hfa"
 forall( otype iterator_type, otype elt_type | iterator( iterator_type, elt_type ) )
+forall( iterator_type, elt_type | iterator( iterator_type, elt_type ) )
 void for_each( iterator_type begin, iterator_type end, void (* func)( elt_type ) ) {
         for ( iterator_type i = begin; i != end; ++i ) {
 …
 } // for_each
 forall( otype iterator_type, otype elt_type | iterator( iterator_type, elt_type ) )
+forall( iterator_type, elt_type | iterator( iterator_type, elt_type ) )
 void for_each_reverse( iterator_type begin, iterator_type end, void (* func)( elt_type ) ) {
         for ( iterator_type i = end; i != begin; ) {

libcfa/src/iterator.hfa

-              r92bfda0
+              rdafbde8
 // An iterator can be used to traverse a data structure.
 trait iterator( otype iterator_type, otype elt_type ) {
+trait iterator( iterator_type, elt_type ) {
         // point to the next element
 //      iterator_type ?++( iterator_type & );
 …
 };
 trait iterator_for( otype iterator_type, otype collection_type, otype elt_type | iterator( iterator_type, elt_type ) ) {
+trait iterator_for( iterator_type, collection_type, elt_type | iterator( iterator_type, elt_type ) ) {
 //      [ iterator_type begin, iterator_type end ] get_iterators( collection_type );
         iterator_type begin( collection_type );
 …
 };
 forall( otype iterator_type, otype elt_type | iterator( iterator_type, elt_type ) )
+forall( iterator_type, elt_type | iterator( iterator_type, elt_type ) )
 void for_each( iterator_type begin, iterator_type end, void (* func)( elt_type ) );
 forall( otype iterator_type, otype elt_type | iterator( iterator_type, elt_type ) )
+forall( iterator_type, elt_type | iterator( iterator_type, elt_type ) )
 void for_each_reverse( iterator_type begin, iterator_type end, void (* func)( elt_type ) );

libcfa/src/math.hfa

-              r92bfda0
+              rdafbde8
         unsigned long long int floor( unsigned long long int n, unsigned long long int align ) { return n / align * align; }
         // forall( otype T | { T ?/?( T, T ); T ?*?( T, T ); } )
+        // forall( T | { T ?/?( T, T ); T ?*?( T, T ); } )
         // T floor( T n, T align ) { return n / align * align; }
 …
         unsigned long long int ceiling_div( unsigned long long int n, unsigned long long int align ) { return (n + (align - 1)) / align; }
         // forall( otype T | { T ?+?( T, T ); T ?-?( T, T ); T ?%?( T, T ); } )
+        // forall( T | { T ?+?( T, T ); T ?-?( T, T ); T ?%?( T, T ); } )
         // T ceiling_div( T n, T align ) { verify( is_pow2( align ) );return (n + (align - 1)) / align; }
 …
         unsigned long long int ceiling( unsigned long long int n, unsigned long long int align ) { return floor( n + (n % align != 0 ? align - 1 : 0), align ); }
         // forall( otype T | { void ?{}( T &, one_t ); T ?+?( T, T ); T ?-?( T, T ); T ?/?( T, T ); } )
+        // forall( T | { void ?{}( T &, one_t ); T ?+?( T, T ); T ?-?( T, T ); T ?/?( T, T ); } )
         // T ceiling( T n, T align ) { return return floor( n + (n % align != 0 ? align - 1 : 0), align ); *}
 …
 static inline {
         forall( otype T | { void ?{}( T &, one_t ); T ?+?( T, T ); T ?-?( T, T );T ?*?( T, T ); } )
+        forall( T | { void ?{}( T &, one_t ); T ?+?( T, T ); T ?-?( T, T );T ?*?( T, T ); } )
         T lerp( T x, T y, T a ) { return x * ((T){1} - a) + y * a; }
         forall( otype T | { void ?{}( T &, zero_t ); void ?{}( T &, one_t ); int ?<?( T, T ); } )
+        forall( T | { void ?{}( T &, zero_t ); void ?{}( T &, one_t ); int ?<?( T, T ); } )
         T step( T edge, T x ) { return x < edge ? (T){0} : (T){1}; }
         forall( otype T | { void ?{}( T &, int ); T clamp( T, T, T ); T ?-?( T, T ); T ?*?( T, T ); T ?/?( T, T ); } )
+        forall( T | { void ?{}( T &, int ); T clamp( T, T, T ); T ?-?( T, T ); T ?*?( T, T ); T ?/?( T, T ); } )
         T smoothstep( T edge0, T edge1, T x ) { T t = clamp( (x - edge0) / (edge1 - edge0), (T){0}, (T){1} ); return t * t * ((T){3} - (T){2} * t); }
 } // distribution

libcfa/src/memory.cfa

-              r92bfda0
+              rdafbde8
 // Internal data object.
 forall(dtype T | sized(T), ttype Args | { void ?{}(T &, Args); })
+forall(T & | sized(T), Args... | { void ?{}(T &, Args); })
 void ?{}(counter_data(T) & this, Args args) {
         (this.counter){1};
 …
+}
 forall(dtype T | sized(T) | { void ^?{}(T &); })
+forall(T & | sized(T) | { void ^?{}(T &); })
 void ^?{}(counter_data(T) & this) {
         assert(0 == this.counter);
 …
 // This is one of many pointers keeping this alive.
 forall(dtype T | sized(T))
+forall(T & | sized(T))
 void ?{}(counter_ptr(T) & this) {
         this.data = 0p;
+}
 forall(dtype T | sized(T))
+forall(T & | sized(T))
 void ?{}(counter_ptr(T) & this, zero_t) {
         this.data = 0p;
+}
 forall(dtype T | sized(T) | { void ^?{}(T &); })
+forall(T & | sized(T) | { void ^?{}(T &); })
 static void internal_decrement(counter_ptr(T) & this) {
         if (this.data && 0 == --this.data->counter) {
 …
+}
 forall(dtype T | sized(T))
+forall(T & | sized(T))
 static void internal_copy(counter_ptr(T) & this, counter_ptr(T) & that) {
         this.data = that.data;
 …
+}
 forall(dtype T | sized(T) | { void ^?{}(T &); })
+forall(T & | sized(T) | { void ^?{}(T &); })
 void ?{}(counter_ptr(T) & this, counter_ptr(T) that) {
         // `that` is a copy but it should have neither a constructor
 …
+}
 forall(dtype T | sized(T), ttype Args | { void ?{}(T&, Args); })
+forall(T & | sized(T), Args... | { void ?{}(T&, Args); })
 void ?{}(counter_ptr(T) & this, Args args) {
         this.data = (counter_data(T)*)new(args);
+}
 forall(dtype T | sized(T) | { void ^?{}(T &); })
+forall(T & | sized(T) | { void ^?{}(T &); })
 void ^?{}(counter_ptr(T) & this) {
         internal_decrement(this);
+}
 forall(dtype T | sized(T))
+forall(T & | sized(T))
 T & *?(counter_ptr(T) & this) {
         return *((this.data) ? &this.data->object : 0p);
+}
 forall(dtype T | sized(T) | { void ^?{}(T &); })
+forall(T & | sized(T) | { void ^?{}(T &); })
 void ?=?(counter_ptr(T) & this, counter_ptr(T) that) {
         if (this.data != that.data) {
 …
+}
 forall(dtype T | sized(T) | { void ^?{}(T &); })
+forall(T & | sized(T) | { void ^?{}(T &); })
 void ?=?(counter_ptr(T) & this, zero_t) {
         internal_decrement(this);
 …
+}
 forall(dtype T | sized(T))
+forall(T & | sized(T))
 int ?==?(counter_ptr(T) const & this, counter_ptr(T) const & that) {
         return this.data == that.data;
+}
 forall(dtype T | sized(T))
+forall(T & | sized(T))
 int ?!=?(counter_ptr(T) const & this, counter_ptr(T) const & that) {
         return !?==?(this, that);
+}
 forall(dtype T | sized(T))
+forall(T & | sized(T))
 int ?==?(counter_ptr(T) const & this, zero_t) {
         return this.data == 0;
+}
 forall(dtype T | sized(T))
+forall(T & | sized(T))
 int ?!=?(counter_ptr(T) const & this, zero_t) {
         return !?==?(this, (zero_t)0);
 …
 // This is the only pointer that keeps this alive.
 forall(dtype T)
+forall(T &)
 void ?{}(unique_ptr(T) & this) {
         this.data = 0p;
+}
 forall(dtype T)
+forall(T &)
 void ?{}(unique_ptr(T) & this, zero_t) {
         this.data = 0p;
+}
 forall(dtype T | sized(T), ttype Args | { void ?{}(T &, Args); })
+forall(T & | sized(T), Args... | { void ?{}(T &, Args); })
 void ?{}(unique_ptr(T) & this, Args args) {
         this.data = (T *)new(args);
+}
 forall(dtype T | { void ^?{}(T &); })
+forall(T & | { void ^?{}(T &); })
 void ^?{}(unique_ptr(T) & this) {
         delete(this.data);
+}
 forall(dtype T)
+forall(T &)
 T & *?(unique_ptr(T) & this) {
         return *this.data;
+}
 forall(dtype T | { void ^?{}(T &); })
+forall(T & | { void ^?{}(T &); })
 void ?=?(unique_ptr(T) & this, zero_t) {
         delete(this.data);
 …
+}
 forall(dtype T | { void ^?{}(T &); })
+forall(T & | { void ^?{}(T &); })
 void move(unique_ptr(T) & this, unique_ptr(T) & that) {
         delete(this.data);
 …
+}
 forall(dtype T)
+forall(T &)
 int ?==?(unique_ptr(T) const & this, unique_ptr(T) const & that) {
         return this.data == that.data;
+}
 forall(dtype T)
+forall(T &)
 int ?!=?(unique_ptr(T) const & this, unique_ptr(T) const & that) {
         return !?==?(this, that);
+}
 forall(dtype T)
+forall(T &)
 int ?==?(unique_ptr(T) const & this, zero_t) {
         return this.data == 0;
+}
 forall(dtype T)
+forall(T &)
 int ?!=?(unique_ptr(T) const & this, zero_t) {
         return !?==?(this, (zero_t)0);

libcfa/src/memory.hfa

-              r92bfda0
+              rdafbde8
 // Internal data object.
 forall(dtype T | sized(T)) {
+forall(T & | sized(T)) {
         struct counter_data {
                 unsigned int counter;
 …
         };
         forall(ttype Args | { void ?{}(T &, Args); })
+        forall(Args... | { void ?{}(T &, Args); })
         void ?{}(counter_data(T) & this, Args args);
 …
 // This is one of many pointers keeping this alive.
 forall(dtype T | sized(T)) {
+forall(T & | sized(T)) {
         struct counter_ptr {
                 counter_data(T) * data;
 …
         forall( | { void ^?{}(T &); })
         void ?{}(counter_ptr(T) & this, counter_ptr(T) that);
         forall(ttype Args | { void ?{}(T&, Args); })
+        forall(Args... | { void ?{}(T&, Args); })
         void ?{}(counter_ptr(T) & this, Args args);
 …
 // This is the only pointer that keeps this alive.
 forall(dtype T) {
+forall(T &) {
         struct unique_ptr {
                 T * data;
 …
         void ?{}(unique_ptr(T) & this, zero_t);
         void ?{}(unique_ptr(T) & this, unique_ptr(T) that) = void;
         forall( | sized(T), ttype Args | { void ?{}(T &, Args); })
+        forall( | sized(T), Args... | { void ?{}(T &, Args); })
         void ?{}(unique_ptr(T) & this, Args args);

libcfa/src/parseargs.hfa

-              r92bfda0
+              rdafbde8
 static inline void ?{}( cfa_option & this ) {}
 forall(dtype T | { bool parse(const char *, T & ); })
+forall(T & | { bool parse(const char *, T & ); })
 static inline void ?{}( cfa_option & this, char short_name, const char * long_name, const char * help, T & variable ) {
       this.val        = 0;
 …
+}
 forall(dtype T)
+forall(T &)
 static inline void ?{}( cfa_option & this, char short_name, const char * long_name, const char * help, T & variable, bool (*parse)(const char *, T & )) {
       this.val        = 0;

libcfa/src/rational.cfa

-              r92bfda0
+              rdafbde8
 #include "stdlib.hfa"
 forall( otype RationalImpl | arithmetic( RationalImpl ) ) {
+forall( RationalImpl | arithmetic( RationalImpl ) ) {
         // helper routines
 …
         // I/O
         forall( dtype istype | istream( istype ) | { istype & ?|?( istype &, RationalImpl & ); } )
+        forall( istype & | istream( istype ) | { istype & ?|?( istype &, RationalImpl & ); } )
         istype & ?|?( istype & is, Rational(RationalImpl) & r ) {
                 is | r.numerator | r.denominator;
 …
         } // ?|?
         forall( dtype ostype | ostream( ostype ) | { ostype & ?|?( ostype &, RationalImpl ); } ) {
+        forall( ostype & | ostream( ostype ) | { ostype & ?|?( ostype &, RationalImpl ); } ) {
                 ostype & ?|?( ostype & os, Rational(RationalImpl) r ) {
                         return os | r.numerator | '/' | r.denominator;
 …
 } // distribution
 forall( otype RationalImpl | arithmetic( RationalImpl ) | { RationalImpl ?\?( RationalImpl, unsigned long ); } )
+forall( RationalImpl | arithmetic( RationalImpl ) | { RationalImpl ?\?( RationalImpl, unsigned long ); } )
 Rational(RationalImpl) ?\?( Rational(RationalImpl) x, long int y ) {
         if ( y < 0 ) {
 …
 // conversion
 forall( otype RationalImpl | arithmetic( RationalImpl ) | { double convert( RationalImpl ); } )
+forall( RationalImpl | arithmetic( RationalImpl ) | { double convert( RationalImpl ); } )
 double widen( Rational(RationalImpl) r ) {
         return convert( r.numerator ) / convert( r.denominator );
 } // widen
 forall( otype RationalImpl | arithmetic( RationalImpl ) | { double convert( RationalImpl ); RationalImpl convert( double ); } )
+forall( RationalImpl | arithmetic( RationalImpl ) | { double convert( RationalImpl ); RationalImpl convert( double ); } )
 Rational(RationalImpl) narrow( double f, RationalImpl md ) {
         // http://www.ics.uci.edu/~eppstein/numth/frap.c

libcfa/src/rational.hfa

-              r92bfda0
+              rdafbde8
 #include "iostream.hfa"
 trait scalar( otype T ) {
+trait scalar( T ) {
 };
 trait arithmetic( otype T | scalar( T ) ) {
+trait arithmetic( T | scalar( T ) ) {
         int !?( T );
         int ?==?( T, T );
 …
 // implementation
 forall( otype RationalImpl | arithmetic( RationalImpl ) ) {
+forall( RationalImpl | arithmetic( RationalImpl ) ) {
         struct Rational {
                 RationalImpl numerator, denominator;                    // invariant: denominator > 0
 …
         // I/O
         forall( dtype istype | istream( istype ) | { istype & ?|?( istype &, RationalImpl & ); } )
+        forall( istype & | istream( istype ) | { istype & ?|?( istype &, RationalImpl & ); } )
         istype & ?|?( istype &, Rational(RationalImpl) & );
         forall( dtype ostype | ostream( ostype ) | { ostype & ?|?( ostype &, RationalImpl ); } ) {
+        forall( ostype & | ostream( ostype ) | { ostype & ?|?( ostype &, RationalImpl ); } ) {
                 ostype & ?|?( ostype &, Rational(RationalImpl) );
                 void ?|?( ostype &, Rational(RationalImpl) );
 …
 } // distribution
 forall( otype RationalImpl | arithmetic( RationalImpl ) |{RationalImpl ?\?( RationalImpl, unsigned long );} )
+forall( RationalImpl | arithmetic( RationalImpl ) |{RationalImpl ?\?( RationalImpl, unsigned long );} )
 Rational(RationalImpl) ?\?( Rational(RationalImpl) x, long int y );
 // conversion
 forall( otype RationalImpl | arithmetic( RationalImpl ) | { double convert( RationalImpl ); } )
+forall( RationalImpl | arithmetic( RationalImpl ) | { double convert( RationalImpl ); } )
 double widen( Rational(RationalImpl) r );
 forall( otype RationalImpl | arithmetic( RationalImpl ) | { double convert( RationalImpl );  RationalImpl convert( double );} )
+forall( RationalImpl | arithmetic( RationalImpl ) | { double convert( RationalImpl );  RationalImpl convert( double );} )
 Rational(RationalImpl) narrow( double f, RationalImpl md );

libcfa/src/stdlib.cfa

-              r92bfda0
+              rdafbde8
 // Cforall allocation/deallocation and constructor/destructor, array types
 forall( dtype T | sized(T), ttype TT | { void ?{}( T &, TT ); } )
+forall( T & | sized(T), TT... | { void ?{}( T &, TT ); } )
 T * anew( size_t dim, TT p ) {
         T * arr = alloc( dim );
 …
 } // anew
 forall( dtype T | sized(T) | { void ^?{}( T & ); } )
+forall( T & | sized(T) | { void ^?{}( T & ); } )
 void adelete( T arr[] ) {
         if ( arr ) {                                                                            // ignore null
 …
 } // adelete
 forall( dtype T | sized(T) | { void ^?{}( T & ); }, ttype TT | { void adelete( TT ); } )
+forall( T & | sized(T) | { void ^?{}( T & ); }, TT... | { void adelete( TT ); } )
 void adelete( T arr[], TT rest ) {
         if ( arr ) {                                                                            // ignore null
 …
 //---------------------------------------
 forall( otype E | { int ?<?( E, E ); } ) {
+forall( E | { int ?<?( E, E ); } ) {
         E * bsearch( E key, const E * vals, size_t dim ) {
                 int cmp( const void * t1, const void * t2 ) {
 …
 forall( otype K, otype E | { int ?<?( K, K ); K getKey( const E & ); } ) {
+forall( K, E | { int ?<?( K, K ); K getKey( const E & ); } ) {
         E * bsearch( K key, const E * vals, size_t dim ) {
                 int cmp( const void * t1, const void * t2 ) {

libcfa/src/stdlib.hfa

-              r92bfda0
+              rdafbde8
 // Created On       : Thu Jan 28 17:12:35 2016
 // Last Modified By : Peter A. Buhr
 // Last Modified On : Sat Dec 12 13:52:34 2020
 // Update Count     : 536
+// Last Modified On : Sat Jan 16 09:07:10 2021
+// Update Count     : 568
 //
 …
         else return (T *)alignment( _Alignof(T), dim, sizeof(T) )
 static inline forall( dtype T | sized(T) ) {
+static inline forall( T & | sized(T) ) {
         // CFA safe equivalents, i.e., implicit size specification
 …
 . Replace the current forall-block that contains defintions of S_fill and S_realloc with following:
                 forall( dtype T | sized(T) ) {
+                forall( T & | sized(T) ) {
                         union  U_fill           { char c; T * a; T t; };
                         struct S_fill           { char tag; U_fill(T) fill; };
 …
 typedef struct S_resize                 { inline void *;  }     T_resize;
 forall( dtype T ) {
+forall( T & ) {
         struct S_fill           { char tag; char c; size_t size; T * at; char t[50]; };
         struct S_realloc        { inline T *; };
 …
 static inline T_resize  ?`resize  ( void * a )  { return (T_resize){a}; }
 static inline forall( dtype T | sized(T) ) {
+static inline forall( T & | sized(T) ) {
         S_fill(T) ?`fill ( T t ) {
                 S_fill(T) ret = { 't' };
                 size_t size = sizeof(T);
+                if(size > sizeof(ret.t)) { printf("ERROR: const object of size greater than 50 bytes given for dynamic memory fill\n"); exit(1); }
+                if ( size > sizeof(ret.t) ) {
+                        abort( "ERROR: const object of size greater than 50 bytes given for dynamic memory fill\n" );
+                } // if
                 memcpy( &ret.t, &t, size );
                 return ret;
 …
         S_realloc(T)    ?`realloc ( T * a )                             { return (S_realloc(T)){a}; }
         T * $alloc_internal( void * Resize, T * Realloc, size_t Align, size_t Dim, S_fill(T) Fill) {
+        T * $alloc_internal( void * Resize, T * Realloc, size_t Align, size_t Dim, S_fill(T) Fill ) {
                 T * ptr = NULL;
                 size_t size = sizeof(T);
 …
                         ptr = (T*) (void *) resize( (void *)Resize, Align, Dim * size );
                 } else if ( Realloc ) {
                         if (Fill.tag != '0') copy_end = min(malloc_size( Realloc ), Dim * size);
                         ptr = (T*) (void *) realloc( (void *)Realloc, Align, Dim * size );
+                        if ( Fill.tag != '0' ) copy_end = min(malloc_size( Realloc ), Dim * size );
+                        ptr = (T *) (void *) realloc( (void *)Realloc, Align, Dim * size );
                 } else {
                         ptr = (T*) (void *) memalign( Align, Dim * size );
+                }
                 if(Fill.tag == 'c') {
+                        ptr = (T *) (void *) memalign( Align, Dim * size );
+                }
+                if ( Fill.tag == 'c' ) {
                         memset( (char *)ptr + copy_end, (int)Fill.c, Dim * size - copy_end );
                 } else if(Fill.tag == 't') {
+                } else if ( Fill.tag == 't' ) {
                         for ( int i = copy_end; i < Dim * size; i += size ) {
+#pragma GCC diagnostic push
+#pragma GCC diagnostic ignored "-Warray-bounds"
+#pragma GCC diagnostic push
+#pragma GCC diagnostic ignored "-Wstringop-overflow="
+                                memcpy( (char *)ptr + i, &Fill.t, size );
+#pragma GCC diagnostic pop
+#pragma GCC diagnostic pop
+                                #pragma GCC diagnostic push
+                                #pragma GCC diagnostic ignored "-Wmaybe-uninitialized"
+                                memcpy( (char *)ptr + i, &Fill.t, sizeof(Fill.t) );
+                                #pragma GCC diagnostic pop
+                        }
                 } else if(Fill.tag == 'a') {
+                } else if ( Fill.tag == 'a' ) {
                         memcpy( (char *)ptr + copy_end, Fill.at, min(Dim * size - copy_end, Fill.size) );
+                } else if(Fill.tag == 'T') {
+                        for ( int i = copy_end; i < Dim * size; i += size ) {
+                                memcpy( (char *)ptr + i, Fill.at, size );
+                        }
+                } else if ( Fill.tag == 'T' ) {
+                        memcpy( (char *)ptr + copy_end, Fill.at, Dim * size );
+                }
 …
         } // $alloc_internal
         forall( ttype TT | { T * $alloc_internal( void *, T *, size_t, size_t, S_fill(T), TT ); } ) {
+        forall( TT... | { T * $alloc_internal( void *, T *, size_t, size_t, S_fill(T), TT ); } ) {
                 T * $alloc_internal( void *       , T * Realloc, size_t Align, size_t Dim, S_fill(T) Fill, T_resize Resize, TT rest) {
 …
 } // distribution T
 static inline forall( dtype T | sized(T) ) {
+static inline forall( T & | sized(T) ) {
         // CFA safe initialization/copy, i.e., implicit size specification, non-array types
         T * memset( T * dest, char fill ) {
 …
 // CFA deallocation for multiple objects
 static inline forall( dtype T )                                                 // FIX ME, problems with 0p in list
+static inline forall( T & )                                                     // FIX ME, problems with 0p in list
 void free( T * ptr ) {
         free( (void *)ptr );                                                            // C free
 } // free
 static inline forall( dtype T, ttype TT | { void free( TT ); } )
+static inline forall( T &, TT... | { void free( TT ); } )
 void free( T * ptr, TT rest ) {
         free( ptr );
 …
 // CFA allocation/deallocation and constructor/destructor, non-array types
 static inline forall( dtype T | sized(T), ttype TT | { void ?{}( T &, TT ); } )
+static inline forall( T & | sized(T), TT... | { void ?{}( T &, TT ); } )
 T * new( TT p ) {
         return &(*(T *)malloc()){ p };                                                  // run constructor
 } // new
 static inline forall( dtype T | { void ^?{}( T & ); } )
+static inline forall( T & | { void ^?{}( T & ); } )
 void delete( T * ptr ) {
         // special case for 0-sized object => always call destructor
 …
         free( ptr );                                                                            // always call free
 } // delete
 static inline forall( dtype T, ttype TT | { void ^?{}( T & ); void delete( TT ); } )
+static inline forall( T &, TT... | { void ^?{}( T & ); void delete( TT ); } )
 void delete( T * ptr, TT rest ) {
         delete( ptr );
 …
 // CFA allocation/deallocation and constructor/destructor, array types
 forall( dtype T | sized(T), ttype TT | { void ?{}( T &, TT ); } ) T * anew( size_t dim, TT p );
 forall( dtype T | sized(T) | { void ^?{}( T & ); } ) void adelete( T arr[] );
 forall( dtype T | sized(T) | { void ^?{}( T & ); }, ttype TT | { void adelete( TT ); } ) void adelete( T arr[], TT rest );
+forall( T & | sized(T), TT... | { void ?{}( T &, TT ); } ) T * anew( size_t dim, TT p );
+forall( T & | sized(T) | { void ^?{}( T & ); } ) void adelete( T arr[] );
+forall( T & | sized(T) | { void ^?{}( T & ); }, TT... | { void adelete( TT ); } ) void adelete( T arr[], TT rest );
 //---------------------------------------
 …
 //---------------------------------------
 forall( otype E | { int ?<?( E, E ); } ) {
+forall( E | { int ?<?( E, E ); } ) {
         E * bsearch( E key, const E * vals, size_t dim );
         size_t bsearch( E key, const E * vals, size_t dim );
 …
 } // distribution
 forall( otype K, otype E | { int ?<?( K, K ); K getKey( const E & ); } ) {
+forall( K, E | { int ?<?( K, K ); K getKey( const E & ); } ) {
         E * bsearch( K key, const E * vals, size_t dim );
         size_t bsearch( K key, const E * vals, size_t dim );
 …
 } // distribution
 forall( otype E | { int ?<?( E, E ); } ) {
+forall( E | { int ?<?( E, E ); } ) {
         void qsort( E * vals, size_t dim );
 } // distribution

libcfa/src/time.cfa

-              r92bfda0
+              rdafbde8
 forall( dtype ostype | ostream( ostype ) ) {
+forall( ostype & | ostream( ostype ) ) {
         ostype & ?|?( ostype & os, Duration dur ) with( dur ) {
                 (ostype &)(os | tn / TIMEGRAN);                                 // print seconds
 …
 } // strftime
 forall( dtype ostype | ostream( ostype ) ) {
+forall( ostype & | ostream( ostype ) ) {
         ostype & ?|?( ostype & os, Time time ) with( time ) {
                 char buf[32];                                                                   // at least 26

libcfa/src/vec/vec.hfa

-              r92bfda0
+              rdafbde8
 #include <math.hfa>
 trait fromint(otype T) {
+trait fromint(T) {
     void ?{}(T&, int);
 };
 trait zeroinit(otype T) {
+trait zeroinit(T) {
     void ?{}(T&, zero_t);
 };
 trait zero_assign(otype T) {
+trait zero_assign(T) {
     T ?=?(T&, zero_t);
 };
 trait subtract(otype T) {
+trait subtract(T) {
     T ?-?(T, T);
 };
 trait negate(otype T) {
+trait negate(T) {
     T -?(T);
 };
 trait add(otype T) {
+trait add(T) {
     T ?+?(T, T);
 };
 trait multiply(otype T) {
+trait multiply(T) {
     T ?*?(T, T);
 };
 trait divide(otype T) {
+trait divide(T) {
     T ?/?(T, T);
 };
 trait lessthan(otype T) {
+trait lessthan(T) {
     int ?<?(T, T);
 };
 trait equality(otype T) {
+trait equality(T) {
     int ?==?(T, T);
 };
 trait sqrt(otype T) {
+trait sqrt(T) {
     T sqrt(T);
 };
 …
+}
 trait dottable(otype V, otype T) {
+trait dottable(V, T) {
     T dot(V, V);
 };
 …
 static inline {
 forall(otype T | sqrt(T), otype V | dottable(V, T))
+forall(T | sqrt(T), V | dottable(V, T))
 T length(V v) {
    return sqrt(dot(v, v));
+}
 forall(otype T, otype V | dottable(V, T))
+forall(T, V | dottable(V, T))
 T length_squared(V v) {
    return dot(v, v);
+}
 forall(otype T, otype V | { T length(V); } | subtract(V))
+forall(T, V | { T length(V); } | subtract(V))
 T distance(V v1, V v2) {
     return length(v1 - v2);
+}
 forall(otype T, otype V | { T length(V); V ?/?(V, T); })
+forall(T, V | { T length(V); V ?/?(V, T); })
 V normalize(V v) {
     return v / length(v);
 …
 // Project vector u onto vector v
 forall(otype T, otype V | dottable(V, T) | { V normalize(V); V ?*?(V, T); })
+forall(T, V | dottable(V, T) | { V normalize(V); V ?*?(V, T); })
 V project(V u, V v) {
     V v_norm = normalize(v);
 …
 // Reflect incident vector v with respect to surface with normal n
 forall(otype T | fromint(T), otype V | { V project(V, V); V ?*?(T, V); V ?-?(V,V); })
+forall(T | fromint(T), V | { V project(V, V); V ?*?(T, V); V ?-?(V,V); })
 V reflect(V v, V n) {
     return v - (T){2} * project(v, n);
 …
 // entering material (i.e., from air to water, eta = 1/1.33)
 // v and n must already be normalized
 forall(otype T | fromint(T) | subtract(T) | multiply(T) | add(T) | lessthan(T) | sqrt(T),
        otype V | dottable(V, T) | { V ?*?(T, V); V ?-?(V,V); void ?{}(V&, zero_t); })
+forall(T | fromint(T) | subtract(T) | multiply(T) | add(T) | lessthan(T) | sqrt(T),
+       V | dottable(V, T) | { V ?*?(T, V); V ?-?(V,V); void ?{}(V&, zero_t); })
 V refract(V v, V n, T eta) {
     T dotValue = dot(n, v);
 …
 // i is the incident vector
 // ng is the geometric normal of the surface
 forall(otype T | lessthan(T) | zeroinit(T), otype V | dottable(V, T) | negate(V))
+forall(T | lessthan(T) | zeroinit(T), V | dottable(V, T) | negate(V))
 V faceforward(V n, V i, V ng) {
     return dot(ng, i) < (T){0} ? n : -n;

libcfa/src/vec/vec2.hfa

-              r92bfda0
+              rdafbde8
 #include "vec.hfa"
 forall (otype T) {
+forall (T) {
     struct vec2 {
         T x, y;
 …
+}
 forall (otype T) {
+forall (T) {
     static inline {
 …
+}
 forall(dtype ostype, otype T | writeable(T, ostype)) {
+forall(ostype &, T | writeable(T, ostype)) {
     ostype & ?|?(ostype & os, vec2(T) v) with (v) {
         return os | '<' | x | ',' | y | '>';

libcfa/src/vec/vec3.hfa

-              r92bfda0
+              rdafbde8
 #include "vec.hfa"
 forall (otype T) {
+forall (T) {
     struct vec3 {
         T x, y, z;
 …
+}
 forall (otype T) {
+forall (T) {
     static inline {
 …
+}
 forall(dtype ostype, otype T | writeable(T, ostype)) {
+forall(ostype &, T | writeable(T, ostype)) {
     ostype & ?|?(ostype & os, vec3(T) v) with (v) {
         return os | '<' | x | ',' | y | ',' | z | '>';

libcfa/src/vec/vec4.hfa

-              r92bfda0
+              rdafbde8
 #include "vec.hfa"
 forall (otype T) {
+forall (T) {
     struct vec4 {
         T x, y, z, w;
 …
+}
 forall (otype T) {
+forall (T) {
     static inline {
 …
+}
 forall(dtype ostype, otype T | writeable(T, ostype)) {
+forall(ostype &, T | writeable(T, ostype)) {
     ostype & ?|?(ostype & os, vec4(T) v) with (v) {
         return os | '<' | x | ',' | y | ',' | z | ',' | w | '>';

src/Parser/parser.yy

-              r92bfda0
+              rdafbde8
 type_parameter:                                                                                 // CFA
         type_class identifier_or_type_name
+                { typedefTable.addToScope( *$2, TYPEDEFname, "9" ); }
+                {   typedefTable.addToScope( *$2, TYPEDEFname, "9" );
+                        if ( $1 == TypeDecl::Otype ) { SemanticError( yylloc, "otype keyword is deprecated" ); }
+                        if ( $1 == TypeDecl::Dtype ) { SemanticError( yylloc, "dtype keyword is deprecated" ); }
+                        if ( $1 == TypeDecl::Ttype ) { SemanticError( yylloc, "ttype keyword is deprecated" ); }
+                }
           type_initializer_opt assertion_list_opt
                 { $$ = DeclarationNode::newTypeParam( $1, $2 )->addTypeInitializer( $4 )->addAssertions( $5 ); }

src/ResolvExpr/PolyCost.cc

-              r92bfda0
+              rdafbde8
                 PassVisitor<PolyCost> coster( env, indexer );
                 type->accept( coster );
                 return coster.pass.result;
+                return (coster.pass.result > 0) ? 1 : 0;
+        }
 …
         ast::Pass<PolyCost_new> costing( symtab, env );
         type->accept( costing );
         return costing.core.result;
+        return (costing.core.result > 0) ? 1 : 0;
+}

src/ResolvExpr/SpecCost.cc

-              r92bfda0
+              rdafbde8
                 // mark specialization of base type
                 void postvisit(ReferenceType*) { if ( count >= 0 ) ++count; }
+                void postvisit(StructInstType*) { if ( count >= 0 ) ++count; }
+                void postvisit(UnionInstType*) { if ( count >= 0 ) ++count; }
         private:
 …
                 void previsit(StructInstType* sty) {
                         count = minover( sty->parameters );
-                        visit_children = false;
+                }
 …
                 void previsit(UnionInstType* uty) {
                         count = minover( uty->parameters );
-                        visit_children = false;
+                }
 …
                 void postvisit( const ast::ArrayType * ) { if ( count >= 0 ) ++count; }
                 void postvisit( const ast::ReferenceType * ) { if ( count >= 0 ) ++count; }
+                void postvisit( const ast::StructInstType * ) { if ( count >= 0 ) ++count; }
+                void postvisit( const ast::UnionInstType * ) { if ( count >= 0 ) ++count; }
                 // Use the minimal specialization value over returns and params.
 …
                 void previsit( const ast::StructInstType * sty ) {
                         count = minimumPresent( sty->params, expr_result );
-                        visit_children = false;
+                }
 …
                 void previsit( const ast::UnionInstType * uty ) {
                         count = minimumPresent( uty->params, expr_result );
-                        visit_children = false;
+                }

tests/avltree/avl-private.cfa

-              r92bfda0
+              rdafbde8
 // an AVL tree's height is easy to compute
 // just follow path with the larger balance
 forall(otype K | Comparable(K), otype V)
+forall(K | Comparable(K), V)
 int height(tree(K, V) * t){
   int helper(tree(K, V) * t, int ht){
 …
+}
 forall(otype K | Comparable(K), otype V)
+forall(K | Comparable(K), V)
 int calcBalance(tree(K, V) * t){
   int l = height(t->left);
 …
 // re-establish the link between parent and child
 forall(otype K | Comparable(K), otype V)
+forall(K | Comparable(K), V)
 void relinkToParent(tree(K, V) * t){
   tree(K, V) * parent = t->parent; // FIX ME!!
 …
 // rotate left from t
 forall(otype K | Comparable(K), otype V)
+forall(K | Comparable(K), V)
 tree(K, V) * rotateLeft(tree(K, V) * t){
   tree(K, V) * newRoot = t->right;
 …
 // rotate right from t
 forall(otype K | Comparable(K), otype V)
+forall(K | Comparable(K), V)
 tree(K, V) * rotateRight(tree(K, V) * t){
   tree(K, V) * newRoot = t->left;
 …
 // balances a node that has balance factor -2 or 2
 forall(otype K | Comparable(K), otype V)
+forall(K | Comparable(K), V)
 tree(K, V) * fix(tree(K, V) * t){
   // ensure that t's balance factor is one of
 …
 // attempt to fix the tree, if necessary
 forall(otype K | Comparable(K), otype V)
+forall(K | Comparable(K), V)
 tree(K, V) * tryFix(tree(K, V) * t){
   int b = calcBalance(t);
 …
 // sets parent field of c to be p
 forall(otype K | Comparable(K), otype V)
+forall(K | Comparable(K), V)
 void setParent(tree(K, V) * c, tree(K, V) * p){
   if (! empty(c)){

tests/avltree/avl-private.h

-              r92bfda0
+              rdafbde8
 // attempt to fix the tree, if necessary
 forall(otype K | Comparable(K), otype V)
+forall(K | Comparable(K), V)
 tree(K, V) * tryFix(tree(K, V) * t);
 // sets parent field of c to be p
 forall(otype K | Comparable(K), otype V)
+forall(K | Comparable(K), V)
 void setParent(tree(K, V) * c, tree(K, V) * p);
 forall(otype K | Comparable(K), otype V)
+forall(K | Comparable(K), V)
 int height(tree(K, V) * t);

tests/avltree/avl.h

-              r92bfda0
+              rdafbde8
 // #include <lib.h>
 trait Comparable(otype T) {
+trait Comparable(T) {
   int ?<?(T, T);
 };
 forall(otype T | Comparable(T))
+forall(T | Comparable(T))
 int ?==?(T t1, T t2);
 forall(otype T | Comparable(T))
+forall(T | Comparable(T))
 int ?>?(T t1, T t2);
 …
 // temporary: need forward decl to get around typedef problem
 forall(otype K | Comparable(K), otype V)
+forall(K | Comparable(K), V)
 struct tree;
 forall(otype K | Comparable(K), otype V)
+forall(K | Comparable(K), V)
 struct tree {
   K key;
 …
 };
 forall(otype K | Comparable(K), otype V)
+forall(K | Comparable(K), V)
 void ?{}(tree(K, V) &t, K key, V value);
 forall(otype K, otype V)
+forall(K | Comparable(K), V)
 void ^?{}(tree(K, V) & t);
 forall(otype K | Comparable(K), otype V)
+forall(K | Comparable(K), V)
 tree(K, V) * create(K key, V value);
 forall(otype K | Comparable(K), otype V)
+forall(K | Comparable(K), V)
 V * find(tree(K, V) * t, K key);
 forall(otype K | Comparable(K), otype V)
+forall(K | Comparable(K), V)
 int empty(tree(K, V) * t);
 // returns the root of the tree
 forall(otype K | Comparable(K), otype V)
+forall(K | Comparable(K), V)
 int insert(tree(K, V) ** t, K key, V value);
 forall(otype K | Comparable(K), otype V)
+forall(K | Comparable(K), V)
 int remove(tree(K, V) ** t, K key);
 forall(otype K | Comparable(K), otype V)
+forall(K | Comparable(K), V)
 void copy(tree(K, V) * src, tree(K, V) ** ret);
 forall(otype K | Comparable(K), otype V)
+forall(K | Comparable(K), V)
 void for_each(tree(K, V) * t, void (*func)(V));

tests/avltree/avl0.cfa

-              r92bfda0
+              rdafbde8
 #include "avl.h"
 forall(otype T | Comparable(T))
+forall(T | Comparable(T))
 int ?==?(T t1, T t2) {
   return !(t1 < t2) && !(t2 < t1);
+}
 forall(otype T | Comparable(T))
+forall(T | Comparable(T))
 int ?>?(T t1, T t2) {
   return t2 < t1;

tests/avltree/avl1.cfa

-              r92bfda0
+              rdafbde8
 #include <stdlib.hfa>
 forall(otype K | Comparable(K), otype V)
+forall(K | Comparable(K), V)
 void ?{}(tree(K, V) &t, K key, V value){
   (t.key) { key };
 …
+}
 forall(otype K, otype V)
+forall(K| Comparable(K), V)
 void ^?{}(tree(K, V) & t){
   delete(t.left);
 …
+}
 forall(otype K | Comparable(K), otype V)
+forall(K | Comparable(K), V)
 tree(K, V) * create(K key, V value) {
   // infinite loop trying to resolve ... t = malloc();

tests/avltree/avl2.cfa

-              r92bfda0
+              rdafbde8
 #include "avl-private.h"
 forall(otype K | Comparable(K), otype V)
+forall(K | Comparable(K), V)
 V * find(tree(K, V) * t, K key){
   if (empty(t)){
 …
+}
 forall(otype K | Comparable(K), otype V)
+forall(K | Comparable(K), V)
 int empty(tree(K, V) * t){
   return t == NULL;
 …
 // returns the root of the tree
 forall(otype K | Comparable(K), otype V)
+forall(K | Comparable(K), V)
 int insert(tree(K, V) ** t, K key, V value) {
   // handles a non-empty tree

tests/avltree/avl3.cfa

-              r92bfda0
+              rdafbde8
 // swaps the data within two tree nodes
 forall(otype K | Comparable(K), otype V)
+forall(K | Comparable(K), V)
 void node_swap(tree(K, V) * t, tree(K, V) * t2){
         swap( t->key,  t2->key);
 …
 // go left as deep as possible from within the right subtree
 forall(otype K | Comparable(K), otype V)
+forall(K | Comparable(K), V)
 tree(K, V) * find_successor(tree(K, V) * t){
         tree(K, V) * find_successor_helper(tree(K, V) * t){
 …
 // cleanup - don't want to deep delete, so set children to NULL first.
 forall(otype K | Comparable(K), otype V)
+forall(K | Comparable(K), V)
 void deleteSingleNode(tree(K, V) * t) {
         t->left = NULL;
 …
 // does the actual remove operation once we've found the node in question
 forall(otype K | Comparable(K), otype V)
+forall(K | Comparable(K), V)
 tree(K, V) * remove_node(tree(K, V) * t){
         // is the node a leaf?
 …
 // finds the node that needs to be removed
 forall(otype K | Comparable(K), otype V)
+forall(K | Comparable(K), V)
 tree(K, V) * remove_helper(tree(K, V) * t, K key, int * worked){
         if (empty(t)){
 …
+}
 forall(otype K | Comparable(K), otype V)
+forall(K | Comparable(K), V)
 int remove(tree(K, V) ** t, K key){
         int worked = 0;

tests/avltree/avl4.cfa

-              r92bfda0
+              rdafbde8
 // Perform a shallow copy of src, return the
 // new tree in ret
 forall(otype K | Comparable(K), otype V)
+forall(K | Comparable(K), V)
 int copy(tree(K, V) * src, tree(K, V) ** ret){
   tree(K, V) * helper(tree(K, V) * t, int * worked){
 …
 // Apply func to every value element in t, using an in order traversal
 forall(otype K | Comparable(K), otype V)
+forall(K | Comparable(K), V)
 void for_each(tree(K, V) * t, int (*func)(V)) {
   if (t == NULL) {

tests/bugs/10.cfa

r92bfda0	rdafbde8
2	2	// https://cforall.uwaterloo.ca/trac/ticket/10
3	3
4		forall(~~otype~~ T)
	4	forall(T)
5	5	struct result {
6	6	union {

tests/bugs/104.cfa

r92bfda0	rdafbde8
4	4	[ float, float ] modf_( float x );
5	5
6		forall(~~otype~~ T \| { [T, T] modf_(T); })
	6	forall(T \| { [T, T] modf_(T); })
7	7	void modf(T);
8	8

tests/bugs/194.cfa

-              r92bfda0
+              rdafbde8
 // https://cforall.uwaterloo.ca/trac/ticket/194
 forall( dtype T | sized(T) ) T * foo( void ) {
+forall( T & | sized(T) ) T * foo( void ) {
       printf( "foo1\n" );
         return (T *)0;
+}
 forall( dtype T | sized(T) ) T & foo( void ) {
+forall( T & | sized(T) ) T & foo( void ) {
         printf( "foo2\n" );
         return (T &)*(T *)0;

tests/bugs/196.cfa

-              r92bfda0
+              rdafbde8
 // https://cforall.uwaterloo.ca/trac/ticket/196
 forall(dtype T)
+forall(T &)
 struct link;
 forall(dtype T)
+forall(T &)
 struct link {
         link(T) * next;
 …
 // -----
 forall(dtype T)
+forall(T &)
 struct foo;
 forall(dtype U)
+forall(U &)
 struct bar {
         foo(U) * data;
 };
 forall(dtype T)
+forall(T &)
 struct foo {};

tests/bugs/203-2.cfa

-              r92bfda0
+              rdafbde8
 // Trac ticket: https://cforall.uwaterloo.ca/trac/ticket/203
 forall(dtype A)
+forall(A &)
 struct empty {
         // Nothing.
 };
 forall(dtype C)
+forall(C &)
 struct wrap_e {
         empty(C) field;

tests/bugs/203-7.cfa

-              r92bfda0
+              rdafbde8
 // Trac ticket: https://cforall.uwaterloo.ca/trac/ticket/203
 forall(dtype A)
+forall(A &)
 struct empty {
         // Nothing.
 };
 forall(dtype C)
+forall(C &)
 struct wrap_e {
         empty(C) field;

tests/bugs/203-9.cfa

-              r92bfda0
+              rdafbde8
 // Trac ticket: https://cforall.uwaterloo.ca/trac/ticket/203
 forall(dtype A)
+forall(A &)
 struct empty {
         // Nothing.
 };
 forall(dtype C)
+forall(C &)
 struct wrap_e {
         empty(C) field;

tests/bugs/7.cfa

-              r92bfda0
+              rdafbde8
 // (Bug 1 unresolved as of this test.)
 forall(otype T)
+forall(T)
 struct stack_node;
 forall(otype T)
+forall(T)
 struct stack_node {
     stack_node(T) * next;
 …
 };
 forall(otype T)
+forall(T)
 struct stack {
     stack_node(T) * head;
 };
 trait stack_errors(otype T) {
+trait stack_errors(T) {
     T emptyStackHandler (stack(T) * this);
 };
 forall(otype T | stack_errors(T))
+forall(T | stack_errors(T))
 T pop (stack(T) * this) {
     return (T){};

tests/castError.cfa

r92bfda0	rdafbde8
14	14	//
15	15
16		forall(~~otype~~ T) struct S { T p; };
	16	forall(T) struct S { T p; };
17	17	int f;
18	18	S(int) sint;

tests/concurrent/examples/boundedBufferEXT.cfa

r92bfda0	rdafbde8
24	24	enum { BufferSize = 50 };
25	25
26		forall( ~~otype~~ T ) {
	26	forall( T ) {
27	27	monitor Buffer {
28	28	int front, back, count;

tests/concurrent/examples/boundedBufferINT.cfa

r92bfda0	rdafbde8
24	24	enum { BufferSize = 50 };
25	25
26		forall( ~~otype~~ T ) {
	26	forall( T ) {
27	27	monitor Buffer {
28	28	condition full, empty;

tests/concurrent/examples/quickSort.generic.cfa

r92bfda0	rdafbde8
21	21	#include <string.h> // strcmp
22	22
23		forall( ~~otype~~ T \| { int ?<?( T, T ); } ) {
	23	forall( T \| { int ?<?( T, T ); } ) {
24	24	thread Quicksort {
25	25	T * values; // communication variables

tests/concurrent/multi-monitor.cfa

r92bfda0	rdafbde8
38	38	}
39	39
40		forall(~~dtype T~~ \| sized(T) \| { void ^?{}(T & mutex); })
	40	forall(T & \| sized(T) \| { void ^?{}(T & mutex); })
41	41	void delete_mutex(T * x) {
42	42	^(*x){};

tests/errors/completeType.cfa

-              r92bfda0
+              rdafbde8
 void foo(int *) {}
 void bar(void *) {}
 forall(otype T) void baz(T *);
 forall(dtype T) void qux(T *);
 forall(dtype T | sized(T)) void quux(T *);
+forall(T) void baz(T *);
+forall(T &) void qux(T *);
+forall(T & | sized(T)) void quux(T *);
 struct A;       // incomplete
 …
 forall(otype T)
+forall(T)
 void baz(T * x) {
         // okay
 …
+}
 forall(dtype T)
+forall(T &)
 void qux(T * y) {
         // okay
 …
+}
 forall(dtype T | sized(T))
+forall(T & | sized(T))
 void quux(T * z) {
         // okay

tests/exceptions/defaults.cfa

r92bfda0	rdafbde8
55	55
56	56	void unhandled_test(void) {
57		forall(~~dtype T, dtype V~~ \| is_exception(T, V))
	57	forall(T &, V & \| is_exception(T, V))
58	58	void defaultTerminationHandler(T &) {
59	59	throw (unhandled_exception){};

tests/exceptions/polymorphic.cfa

-              r92bfda0
+              rdafbde8
 #include <exception.hfa>
 FORALL_TRIVIAL_EXCEPTION(proxy, (otype T), (T));
 FORALL_TRIVIAL_INSTANCE(proxy, (otype U), (U))
+FORALL_TRIVIAL_EXCEPTION(proxy, (T), (T));
+FORALL_TRIVIAL_INSTANCE(proxy, (U), (U))
 const char * msg(proxy(int) * this) { return "proxy(int)"; }
 …
+}
 FORALL_DATA_EXCEPTION(cell, (otype T), (T))(
+FORALL_DATA_EXCEPTION(cell, (T), (T))(
         T data;
 );
 FORALL_DATA_INSTANCE(cell, (otype T), (T))
+FORALL_DATA_INSTANCE(cell, (T), (T))
 const char * msg(cell(int) * this) { return "cell(int)"; }

tests/exceptions/virtual-poly.cfa

-              r92bfda0
+              rdafbde8
 };
 forall(otype T)
+forall(T)
 struct mono_child_vtable {
         mono_base_vtable const * const parent;
 };
 forall(otype T)
+forall(T)
 struct mono_child {
         mono_child_vtable(T) const * virtual_table;
 …
+}
 forall(otype U)
+forall(U)
 struct poly_base_vtable {
         poly_base_vtable(U) const * const parent;
 };
 forall(otype U)
+forall(U)
 struct poly_base {
         poly_base_vtable(U) const * virtual_table;
 };
 forall(otype V)
+forall(V)
 struct poly_child_vtable {
         poly_base_vtable(V) const * const parent;
 };
 forall(otype V)
+forall(V)
 struct poly_child {
         poly_child_vtable(V) const * virtual_table;

tests/forall.cfa

-              r92bfda0
+              rdafbde8
 void g1() {
         forall( otype T ) T f( T ) {};
+        forall( T ) T f( T ) {};
         void f( int ) {};
         void h( void (*p)(void) ) {};
 …
 void g2() {
         forall( otype T ) void f( T, T ) {}
         forall( otype T, otype U ) void f( T, U ) {}
+        forall( T ) void f( T, T ) {}
+        forall( T, U ) void f( T, U ) {}
         int x;
 …
+}
 typedef forall ( otype T ) int (* f)( int );
 forall( otype T )
+typedef forall ( T ) int (* f)( int );
+forall( T )
 void swap( T left, T right ) {
         T temp = left;
 …
+}
 trait sumable( otype T ) {
+trait sumable( T ) {
         void ?{}( T &, zero_t );                                                        // 0 literal constructor
         T ?+?( T, T );                                                                          // assortment of additions
 …
 }; // sumable
 forall( otype T | sumable( T ) )                                                // use trait
+forall( T | sumable( T ) )                                              // use trait
 T sum( size_t size, T a[] ) {
         T total = 0;                                                                            // initialize by 0 constructor
 …
 } // sum
 forall( otype T | { T ?+?( T, T ); T ?++( T & ); [T] ?+=?( T &,T ); } )
+forall( T | { T ?+?( T, T ); T ?++( T & ); [T] ?+=?( T &,T ); } )
 T twice( T t ) {
         return t + t;
+}
 forall( otype T | { int ?<?(T, T); } )
+forall( T | { int ?<?(T, T); } )
 T min( T t1, T t2 ) {
         return t1 < t2 ? t1 : t2;
 …
 // Multiple forall
 forall( otype T ) forall( otype S ) struct { int i; };
 forall( otype T ) struct { int i; } forall( otype S );
 struct { int i; } forall( otype T ) forall( otype S );
 forall( otype W ) struct { int i; } forall( otype T ) forall( otype S );
+forall( T ) forall( S ) struct { int i; };
+forall( T ) struct { int i; } forall( S );
+struct { int i; } forall( T ) forall( S );
+forall( W ) struct { int i; } forall( T ) forall( S );
 // Distribution
 struct P { int i; };
 forall( otype T ) struct Q { T i; };
 forall( otype T ) struct { int i; };
+forall( T ) struct Q { T i; };
+forall( T ) struct { int i; };
 struct KK { int i; };
 inline static {
         void RT1() {}
+}
 forall( otype T ) {
+forall( T ) {
         T RT2( T ) {
                 typedef int TD1;
                 struct S1 { T t; };
+        }
         forall( otype X ) {
+        forall( X ) {
                 typedef int TD2;
                 struct S2 {};
 …
+        }
         extern "C" {
                 forall( otype W ) {
+                forall( W ) {
                         W RT3( W ) {}
                         struct S3 {};
 …
+        }
         void RT4() {
                 forall( otype W ) struct S4 {};
+                forall( W ) struct S4 {};
                 typedef int TD3;
+        }
 …
 static inline {
         forall( otype T ) {
+        forall( T ) {
                 int RT6( T p );
+        }
         forall( otype T, otype U ) {
+        forall( T, U ) {
                 int RT7( T, U );
+        }
+}
 static forall( otype T ) {
+static forall( T ) {
         int RT8( T );
+}
 forall( otype T ) inline static {
+forall( T ) inline static {
         int RT9( T ) { T t; return 3; }
+}
 forall( otype T | { T ?+?( T, T ); } ) {
         forall( otype S | { T ?+?( T, S ); } ) {
                 forall( otype W ) T bar( T t, S s ) { return t + s; }
                 forall( otype W | { W ?+?( T, W ); } ) W baz( T t, S s, W w ) { return t + s + w; }
+forall( T | { T ?+?( T, T ); } ) {
+        forall( S | { T ?+?( T, S ); } ) {
+                forall( W ) T bar( T t, S s ) { return t + s; }
+                forall( W | { W ?+?( T, W ); } ) W baz( T t, S s, W w ) { return t + s + w; }
                 struct W { T t; } (int,int) ww;
                 struct P pp;
 …
+}
 forall( otype T | { T ?+?( T, T ); } ) forall( otype S | { T ?+?( T, S ); } )
+forall( T | { T ?+?( T, T ); } ) forall( S | { T ?+?( T, S ); } )
 struct XW { T t; };
 XW(int,int) xww;
 forall( otype T ) struct S { T t; } (int) x, y, z;
 forall( otype T ) struct { T t; } (int) a, b, c;
 forall( otype T ) static forall( otype S ) {
     forall( otype X ) struct U {
+forall( T ) struct S { T t; } (int) x, y, z;
+forall( T ) struct { T t; } (int) a, b, c;
+forall( T ) static forall( S ) {
+    forall( X ) struct U {
                 T x;
     };
+}
 forall( otype T ) {
+forall( T ) {
         extern "C" {
                 struct SS { T t; };

tests/function-operator.cfa

-              r92bfda0
+              rdafbde8
 // STL-like Algorithms
 trait Assignable(dtype T, dtype U) { T ?=?(T &, U); };
 trait Copyable(dtype T) { void ?{}(T &, T); };
 trait Destructable(dtype T) { void ^?{}(T &); };
+trait Assignable(T &, U &) { T ?=?(T &, U); };
+trait Copyable(T &) { void ?{}(T &, T); };
+trait Destructable(T &) { void ^?{}(T &); };
 trait Iterator(dtype iter | sized(iter) | Copyable(iter) | Destructable(iter), otype T) {
+trait Iterator(iter & | sized(iter) | Copyable(iter) | Destructable(iter), T) {
         T & *?(iter);
         iter ++?(iter &);
 …
 };
 forall(otype Tin, dtype Input | Iterator(Input, Tin), otype Tout, dtype Output | Iterator(Output, Tout) | Assignable(Tout, Tin))
+forall(Tin, Input & | Iterator(Input, Tin), Tout, Output & | Iterator(Output, Tout) | Assignable(Tout, Tin))
 Output copy(Input first, Input last, Output result) {
         while (first != last) {
 …
 // test ?()(T *, ...) -- ?() with function call-by-pointer
 forall(otype Tin, dtype Input | Iterator(Input, Tin), otype Tout, dtype Output | Iterator(Output, Tout), otype FuncRet, dtype Func | { FuncRet ?()(Func *, Tin); } | Assignable(Tout, FuncRet))
+forall(Tin, Input & | Iterator(Input, Tin), Tout, Output & | Iterator(Output, Tout), FuncRet, Func & | { FuncRet ?()(Func *, Tin); } | Assignable(Tout, FuncRet))
 Output transform (Input first, Input last, Output result, Func * op) {
         while (first != last) {
 …
 // test ?()(T, ...) -- ?() with function call-by-value
 forall(dtype Iter, otype T | Iterator(Iter, T), otype Pred | { int ?()(Pred, T); })
+forall(Iter &, T | Iterator(Iter, T), Pred | { int ?()(Pred, T); })
 Iter find_if (Iter first, Iter last, Pred pred) {
         while (first != last) {
 …
 // test ?()(T, ...) -- ?() with function call-by-reference
 forall(otype Generator, otype GenRet | { GenRet ?()(Generator &); }, dtype Iter, otype T | Iterator(Iter, T) | Assignable(T, GenRet))
+forall(Generator, GenRet | { GenRet ?()(Generator &); }, Iter &, T | Iterator(Iter, T) | Assignable(T, GenRet))
 void generate(Iter first, Iter last, Generator & gen) {
         int i = 0;
 …
+}
 forall(otype T | { int ?==?(T, T); })
+forall(T | { int ?==?(T, T); })
 struct Equals {
         T val;
 };
 forall(otype T | { int ?==?(T, T); })
+forall(T | { int ?==?(T, T); })
 int ?()(Equals(T) eq, T x) {
         return eq.val == x;
+}
 forall(otype T | { T ?*?(T, T); })
+forall(T | { T ?*?(T, T); })
 struct Multiply {
         T val;
 };
 forall(otype T | { T ?*?(T, T); })
+forall(T | { T ?*?(T, T); })
 T ?()(Multiply(T) * mult, T x) {
         return mult->val * x;
 …
 // TODO: generalize to ttype return; doesn't work yet
 // like std::function
 forall(otype Return, ttype Args)
+forall(Return, Args...)
 struct function {
         Return (*f)(Args);

tests/genericUnion.cfa

-              r92bfda0
+              rdafbde8
 #include <limits.hfa>
 forall(otype T)
+forall(T)
 union ByteView {
         T val;
 …
 };
 forall(otype T)
+forall(T)
 void print(ByteView(T) x) {
         for (int i = 0; i < sizeof(int); i++) {                         // want to change to sizeof(T)
 …
+}
 forall(otype T)
+forall(T)
 void f(ByteView(T) x, T val) {
         print(x);

tests/global-monomorph.cfa

-              r92bfda0
+              rdafbde8
 // Create monomorphic instances of polymorphic types at global scope.
 forall(dtype T)
+forall(T &)
 void poly0(T &) {}
 forall(dtype T | sized(T))
+forall(T & | sized(T))
 void poly1(T &) {}
 forall(otype T)
+forall(T)
 void poly2(T &) {}

tests/identity.cfa

r92bfda0	rdafbde8
16	16	#include <fstream.hfa>
17	17
18		forall( ~~otype~~ T )
	18	forall( T )
19	19	T identity( T t ) {
20	20	return t;

tests/init1.cfa

-              r92bfda0
+              rdafbde8
+}
 forall (dtype T, dtype S)
+forall (T &, S &)
 T & anycvt( S & s ) {
     return s;               // mismatched referenced type
+}
 forall (dtype T, dtype S)
+forall (T &, S &)
 T * anycvt( S * s ) {
     return s;               // mismatched referenced type

tests/nested-types.cfa

r92bfda0	rdafbde8
16	16	typedef int N;
17	17	struct A {
18		forall(~~otype~~ T)
	18	forall(T)
19	19	struct N {
20	20	T x;

tests/poly-d-cycle.cfa

-              r92bfda0
+              rdafbde8
 // Check that a cycle of polymorphic dtype structures can be instancated.
 forall(dtype T)
+forall(T &)
 struct func_table;
 forall(dtype U)
+forall(U &)
 struct object {
         func_table(U) * virtual_table;
 };
 forall(dtype T)
+forall(T &)
 struct func_table {
         void (*object_func)(object(T) *);

tests/poly-o-cycle.cfa

-              r92bfda0
+              rdafbde8
 // Check that a cycle of polymorphic otype structures can be instancated.
 forall(otype T)
+forall(T)
 struct func_table;
 forall(otype U)
+forall(U)
 struct object {
         func_table(U) * virtual_table;
 };
 forall(otype T)
+forall(T)
 struct func_table {
         void (*object_func)(object(T) *);

tests/polymorphism.cfa

-              r92bfda0
+              rdafbde8
 #include <fstream.hfa>
 forall(otype T)
+forall(T)
 T f(T x, T y) {
         x = y;
 …
+}
 forall(otype T) T ident(T x) {
+forall(T) T ident(T x) {
         return x;
+}
 forall( otype T, otype U )
+forall( T, U )
 size_t struct_size( T i, U j ) {
         struct S { T i; U j; };
 …
+}
 forall( otype T, otype U )
+forall( T, U )
 size_t union_size( T i, U j ) {
         union B { T i; U j; };
 …
 // perform some simple operations on aggregates of T and U
 forall( otype T | { void print(T); int ?==?(T, T); }, otype U | { void print(U); U ?=?(U&, zero_t); } )
+forall( T | { void print(T); int ?==?(T, T); }, U | { void print(U); U ?=?(U&, zero_t); } )
 U foo(T i, U j) {
         struct S { T i; U j; };

tests/raii/ctor-autogen.cfa

-              r92bfda0
+              rdafbde8
 // dtype-static generic type is otype
 forall(dtype T)
+forall(T &)
 struct DtypeStaticStruct {
   T * data;
 …
 };
 forall(dtype T)
+forall(T &)
 union DtypeStaticUnion {
   T * data;
 …
 // dynamic generic type is otype
 forall(otype T)
+forall(T)
 struct DynamicStruct {
         T x;
 };
 forall(otype T)
+forall(T)
 union DynamicUnion {
         T x;
 …
 forall(otype T)
+forall(T)
 T identity(T x) { return x; }

tests/simpleGenericTriple.cfa

-              r92bfda0
+              rdafbde8
 //
 forall(otype T)
+forall(T)
 struct T3 {
         T f0, f1, f2;
 };
 forall(otype T | { T ?+?(T, T); })
+forall(T | { T ?+?(T, T); })
 T3(T) ?+?(T3(T) x, T3(T) y) {
         T3(T) z = { x.f0+y.f0, x.f1+y.f1, x.f2+y.f2 };

tests/sum.cfa

-              r92bfda0
+              rdafbde8
 #include <stdlib.hfa>
 trait sumable( otype T ) {
+trait sumable( T ) {
         void ?{}( T &, zero_t );                                                        // 0 literal constructor
         T ?+?( T, T );                                                                          // assortment of additions
 …
 }; // sumable
 forall( otype T | sumable( T ) )                                                // use trait
+forall( T | sumable( T ) )                                              // use trait
 T sum( size_t size, T a[] ) {
         T total = 0;                                                                            // initialize by 0 constructor
 …
                  | sum( size, (S *)a ) | ", check" | (S)s;
         forall( otype Impl | sumable( Impl ) )
+        forall( Impl | sumable( Impl ) )
         struct GS {
                 Impl * x, * y;
 …
                  sum( size, (S *)a ).[i, j], s.[i, j] );
         forall( otype Impl | sumable( Impl ) )
+        forall( Impl | sumable( Impl ) )
         struct GS {
                 Impl * x, * y;

tests/tuple/tuplePolymorphism.cfa

-              r92bfda0
+              rdafbde8
 // ensure that f is a viable candidate for g, even though its parameter structure does not exactly match
 [A] f([A, B] x, B y) { printf("%g %c %g %lld %c %lld %lld %c %lld\n", x.0.[x,y,z], x.1.[x,y,z], y.[x,y,z]); return x.0; }
 forall(otype T, otype U | { T f(T, U, U); })
+forall(T, U | { T f(T, U, U); })
 void g(T x, U y) { f(x, y, y); }
 // add two triples
 forall(otype T | { T ?+?(T, T); })
+forall(T | { T ?+?(T, T); })
 [T, T, T] ?+?([T, T, T] x, [T, T, T] y) {
         return [x.0+y.0, x.1+y.1, x.2+y.2];
 …
+}
 forall(otype T)
+forall(T)
 [T, T] foo([T, T] y) {
         [T, T] x;

tests/tuple/tupleVariadic.cfa

-              r92bfda0
+              rdafbde8
         printf("called func(void)\n");
+}
 forall(otype T, ttype Params | { void process(T); void func(Params); })
+forall(T, Params... | { void process(T); void func(Params); })
 void func(T arg1, Params p) {
         process(arg1);
 …
+}
 forall(otype T)
+forall(T)
 T * copy(T x) {
         // test calling new inside a polymorphic function
 …
+}
 forall(ttype T | { void foo(T); }) void bar(T x) {}
+forall(T... | { void foo(T); }) void bar(T x) {}
 void foo(int) {}

tests/zombies/ArrayN.c

r92bfda0	rdafbde8
6	6	// }
7	7
8		forall(~~otype~~ index_t)
	8	forall(index_t)
9	9	index_t offset_to_index(unsigned offset, index_t size) {
10	10	return [offset / size.0, offset % size.1];

tests/zombies/Members.c

-              r92bfda0
+              rdafbde8
 int ?=?( int*, int );
 float ?=?( float*, float );
 forall( dtype DT ) DT * ?=?( DT**, DT* );
 forall(otype T) lvalue T *?( T* );
+forall( DT & ) DT * ?=?( DT**, DT* );
+forall(T) lvalue T *?( T* );
 char *__builtin_memcpy();

tests/zombies/Rank2.c

-              r92bfda0
+              rdafbde8
 int ?=?( int &, int );
 forall(dtype DT) DT * ?=?( DT *&, DT * );
+forall(DT &) DT * ?=?( DT *&, DT * );
 void a() {
         forall( otype T ) void f( T );
         void g( forall( otype U ) void p( U ) );
+        forall( T ) void f( T );
+        void g( forall( U ) void p( U ) );
         g( f );
+}
 …
 void g() {
         void h( int *null );
         forall( otype T ) T id( T );
+        forall( T ) T id( T );
 //      forall( dtype T ) T *0;
 //      int 0;

tests/zombies/abstype.c

-              r92bfda0
+              rdafbde8
+}
 forall( otype T ) T *?( T * );
+forall( T ) T *?( T * );
 int ?++( int * );
 int ?=?( int *, int );
 forall( dtype DT ) DT * ?=?( DT **, DT * );
+forall( DT & ) DT * ?=?( DT **, DT * );
 otype U = int *;

tests/zombies/context.cfa

-              r92bfda0
+              rdafbde8
 // trait declaration
 trait has_q( otype T ) {
+trait has_q( T ) {
         T q( T );
 };
 forall( otype z | has_q( z ) ) void f() {
         trait has_r( otype T, otype U ) {
+forall( z | has_q( z ) ) void f() {
+        trait has_r( T, U ) {
                 T r( T, T (T,U) );
         };

tests/zombies/gc_no_raii/bug-repro/blockers/explicit_cast.c

-              r92bfda0
+              rdafbde8
 };
 forall(otype T)
+forall(T)
 struct gcpointer
+{
 …
 };
 forall(otype T)
+forall(T)
 static inline gcpointer(T) gcmalloc()
+{

tests/zombies/gc_no_raii/bug-repro/blockers/recursive_realloc.c

-              r92bfda0
+              rdafbde8
 #include <stdlib.hfa>
 trait allocator_c(otype T, otype allocator_t)
+trait allocator_c(T, allocator_t)
+{
         void realloc(allocator_t* const, size_t);
 };
 forall(otype T)
+forall(T)
 struct heap_allocator
+{
 …
 };
 forall(otype T)
+forall(T)
 inline void realloc(heap_allocator(T) *const this, size_t size)
+{

tests/zombies/gc_no_raii/bug-repro/deref.c

-              r92bfda0
+              rdafbde8
     forall(otype T)
+    forall(T)
     struct wrap
+    {
 …
     };
     forall(otype T)
+    forall(T)
     T *? (wrap(T) rhs)
+    {

tests/zombies/gc_no_raii/bug-repro/field.c

-              r92bfda0
+              rdafbde8
 //------------------------------------------------------------------------------
 //Declaration
 trait allocator_c(otype T, otype allocator_t)
+trait allocator_c(T, allocator_t)
+{
         void ctor(allocator_t* const);
 …
 };
 forall(otype T, otype allocator_t | allocator_c(T, allocator_t))
+forall(T, allocator_t | allocator_c(T, allocator_t))
 struct vector
+{

tests/zombies/gc_no_raii/bug-repro/malloc.c

r92bfda0	rdafbde8
1		forall(~~otype~~ T)
	1	forall(T)
2	2	struct wrapper
3	3	{
…	…
5	5	};
6	6
7		forall(~~otype~~ T)
	7	forall(T)
8	8	void ctor(wrapper(T)* this)
9	9	{
…	…
11	11	}
12	12
13		forall(~~otype~~ T)
	13	forall(T)
14	14	wrapper(T) gcmalloc()
15	15	{
…	…
19	19	}
20	20
21		forall(~~otype~~ T)
	21	forall(T)
22	22	wrapper(T)* ?=? (wrapper(T)* lhs, wrapper(T)* rhs)
23	23	{

tests/zombies/gc_no_raii/bug-repro/oddtype.c

-              r92bfda0
+              rdafbde8
 forall(dtype T)
+forall(T &)
 struct wrap {
         int i;
 };
 forall(otype T) void ?{}(wrap(T)* this) {}
 forall(otype T) void ?=?(wrap(T)* this) {}
 forall(otype T) void ^?{}(wrap(T)* this) {}
+forall(T) void ?{}(wrap(T)* this) {}
+forall(T) void ?=?(wrap(T)* this) {}
+forall(T) void ^?{}(wrap(T)* this) {}
 struct List_t {

tests/zombies/gc_no_raii/bug-repro/push_back.h

-              r92bfda0
+              rdafbde8
 //------------------------------------------------------------------------------
 //Declaration
 trait allocator_c(otype T, otype allocator_t) {
+trait allocator_c(T, allocator_t) {
         void ctor(allocator_t* const);
         void dtor(allocator_t* const);
 …
 };
 forall(otype T, otype allocator_t | allocator_c(T, allocator_t))
+forall(T, allocator_t | allocator_c(T, allocator_t))
 struct vector
+{
 …
 //------------------------------------------------------------------------------
 //Initialization
 forall(otype T, otype allocator_t | allocator_c(T, allocator_t))
+forall(T, allocator_t | allocator_c(T, allocator_t))
 void vector_ctor(vector(T, allocator_t) *const this);
 forall(otype T, otype allocator_t | allocator_c(T, allocator_t))
+forall(T, allocator_t | allocator_c(T, allocator_t))
 void dtor(vector(T, allocator_t) *const this);
 //------------------------------------------------------------------------------
 //Allocator
 forall(otype T)
+forall(T)
 struct heap_allocator
+{
 …
 };
 forall(otype T)
+forall(T)
 void ctor(heap_allocator(T) *const this);
 forall(otype T)
+forall(T)
 void dtor(heap_allocator(T) *const this);
 forall(otype T)
+forall(T)
 void realloc(heap_allocator(T) *const this, size_t size);
 forall(otype T)
+forall(T)
 inline T* data(heap_allocator(T) *const this)
+{
 …
 //------------------------------------------------------------------------------
 //Capacity
 forall(otype T, otype allocator_t | allocator_c(T, allocator_t))
+forall(T, allocator_t | allocator_c(T, allocator_t))
 inline bool empty(vector(T, allocator_t) *const this)
+{
 …
+}
 forall(otype T, otype allocator_t | allocator_c(T, allocator_t))
+forall(T, allocator_t | allocator_c(T, allocator_t))
 inline bool size(vector(T, allocator_t) *const this)
+{
 …
+}
 forall(otype T, otype allocator_t | allocator_c(T, allocator_t))
+forall(T, allocator_t | allocator_c(T, allocator_t))
 inline void reserve(vector(T, allocator_t) *const this, size_t size)
+{
 …
 //------------------------------------------------------------------------------
 //Modifiers
 forall(otype T, otype allocator_t | allocator_c(T, allocator_t))
+forall(T, allocator_t | allocator_c(T, allocator_t))
 void push_back(vector(T, allocator_t) *const this, T value);

tests/zombies/gc_no_raii/bug-repro/realloc.c

-              r92bfda0
+              rdafbde8
 void* realloc(void*, unsigned long int);
 forall(otype T)
+forall(T)
 struct wrap
+{
 …
 };
 forall(otype T)
+forall(T)
 static inline void realloc(wrap(T) *const this, unsigned long int size)
+{

tests/zombies/gc_no_raii/bug-repro/return.c

r92bfda0	rdafbde8
1		forall(~~otype~~ T)
	1	forall(T)
2	2	struct wrapper
3	3	{
…	…
5	5	};
6	6
7		forall(~~otype~~ T)
	7	forall(T)
8	8	wrapper(T) create()
9	9	{
…	…
12	12	}
13	13
14		forall(~~otype~~ T)
	14	forall(T)
15	15	wrapper(T)* ?=?(wrapper(T)* lhs, wrapper(T)* rhs)
16	16	{

tests/zombies/gc_no_raii/bug-repro/return_template.c

-              r92bfda0
+              rdafbde8
 forall(otype T)
+forall(T)
 struct wrap
+{
 …
 };
 forall(otype T) void ?{}(wrap(T)* this);
 forall(otype T) void ?{}(wrap(T)* this, wrap(T)* rhs);
 forall(otype T) void ^?{}(wrap(T)* this);
 forall(otype T) void ?=?(wrap(T)* this, wrap(T)* rhs);
+forall(T) void ?{}(wrap(T)* this);
+forall(T) void ?{}(wrap(T)* this, wrap(T)* rhs);
+forall(T) void ^?{}(wrap(T)* this);
+forall(T) void ?=?(wrap(T)* this, wrap(T)* rhs);
 forall(otype T)
+forall(T)
 wrap(T) test()
+{

tests/zombies/gc_no_raii/bug-repro/slow_malloc.c

r92bfda0	rdafbde8
1	1	#include <stdlib.hfa>
2	2
3		forall(~~otype~~ T)
	3	forall(T)
4	4	struct heap_allocator
5	5	{

tests/zombies/gc_no_raii/bug-repro/zero.c

-              r92bfda0
+              rdafbde8
 forall(otype T)
+forall(T)
 struct wrap
+{
 …
 };
 forall(otype T)
+forall(T)
 int ?==? (wrap(T) lhs, wrap(T) rhs)
+{
 …
 struct wrap(int) 0;
 /*/
 forall(otype T)
+forall(T)
 struct wrap(T) 0;
 //*/

tests/zombies/gc_no_raii/src/gc.h

r92bfda0	rdafbde8
13	13	// }
14	14
15		forall(~~otype~~ T)
	15	forall(T)
16	16	static inline void gcmalloc(gcpointer(T)* ptr)
17	17	{

tests/zombies/gc_no_raii/src/gcpointers.c

-              r92bfda0
+              rdafbde8
 #endif
 forall(otype T) void ?{}(gcpointer(T)* this) {
+forall(T) void ?{}(gcpointer(T)* this) {
         (&this->internal) {};
+}
 forall(otype T) void ?{}(gcpointer(T)* this, void* address) {
+forall(T) void ?{}(gcpointer(T)* this, void* address) {
         (&this->internal) { address };
+}
 forall(otype T) void ?{}(gcpointer(T)* this, gcpointer(T) other) {
+forall(T) void ?{}(gcpointer(T)* this, gcpointer(T) other) {
         (&this->internal) { other.internal };
+}
 forall(otype T) void ^?{}(gcpointer(T)* this) {
+forall(T) void ^?{}(gcpointer(T)* this) {
         ^?{}(&this->internal);
+}
 forall(otype T) gcpointer(T) ?=?(gcpointer(T)* this, gcpointer(T) rhs) {
+forall(T) gcpointer(T) ?=?(gcpointer(T)* this, gcpointer(T) rhs) {
         this->internal = rhs.internal;
         return *this;
 …
 // forall(otype T) T *?(gcpointer(T) this);
 forall(otype T) T* get(gcpointer(T)* this) {
+forall(T) T* get(gcpointer(T)* this) {
         return (T*)this->internal.ptr;
+}
 //
 // //Logical operators
 forall(otype T) int ?!=?(gcpointer(T) this, int zero) {
+forall(T) int ?!=?(gcpointer(T) this, int zero) {
         return this.internal.ptr != 0;
+}

tests/zombies/gc_no_raii/src/gcpointers.h

-              r92bfda0
+              rdafbde8
 #include <stdint.h>
 forall(dtype T)
+forall(T &)
 struct gcpointer;
 …
 #endif
 forall(dtype T)
+forall(T &)
 struct gcpointer
+{
 …
 //
 forall(otype T) void ?{}(gcpointer(T)* this);
 forall(otype T) void ?{}(gcpointer(T)* this, void* address);
 forall(otype T) void ?{}(gcpointer(T)* this, gcpointer(T) other);
 forall(otype T) void ^?{}(gcpointer(T)* this);
 forall(otype T) gcpointer(T) ?=?(gcpointer(T)* this, gcpointer(T) rhs);
+forall(T) void ?{}(gcpointer(T)* this);
+forall(T) void ?{}(gcpointer(T)* this, void* address);
+forall(T) void ?{}(gcpointer(T)* this, gcpointer(T) other);
+forall(T) void ^?{}(gcpointer(T)* this);
+forall(T) gcpointer(T) ?=?(gcpointer(T)* this, gcpointer(T) rhs);
 // forall(otype T) T *?(gcpointer(T) this);
 forall(otype T) T* get(gcpointer(T)* this);
+forall(T) T* get(gcpointer(T)* this);
 //Logical operators
 forall(otype T) int ?!=?(gcpointer(T) this, int zero);
 forall(otype T) int ?!=?(gcpointer(T) this, gcpointer(T) rhs);
 forall(otype T) int ?==?(gcpointer(T) this, gcpointer(T) rhs);
+forall(T) int ?!=?(gcpointer(T) this, int zero);
+forall(T) int ?!=?(gcpointer(T) this, gcpointer(T) rhs);
+forall(T) int ?==?(gcpointer(T) this, gcpointer(T) rhs);

tests/zombies/gc_no_raii/src/tools.h

-              r92bfda0
+              rdafbde8
 // }
 trait has_equal(otype T)
+trait has_equal(T)
+{
         signed int ?==?(T a, T b);
 };
 trait InputIterator_t(otype T, otype InputIterator)
+trait InputIterator_t(T, InputIterator)
+{
         signed int ?==?(InputIterator a, InputIterator b);
 …
 };
 forall(otype T | has_equal(T), otype InputIterator | InputIterator_t(T, InputIterator))
+forall(T | has_equal(T), InputIterator | InputIterator_t(T, InputIterator))
 static inline InputIterator find( InputIterator first, const InputIterator* const last, T val)
+{

tests/zombies/hashtable.cfa

-              r92bfda0
+              rdafbde8
 trait has_hash( otype K ) {
+trait has_hash( K ) {
     size_t hash(K);
     int ?==?( K, K );
 };
 trait hkey( otype K, dtype tN | has_hash(K) ) {
+trait hkey( K, tN & | has_hash(K) ) {
     K key(tN &);
 };
 forall( otype K, dtype tN, dtype tE | $dlistable(tN, tE) | hkey(K, tN) ) {
+forall( K, tN &, tE & | $dlistable(tN, tE) | hkey(K, tN) ) {
     struct hashtable {
 …
+}
 forall( otype K, dtype tN, dtype tE | $dlistable(tN, tE) | hkey(K, tN) | { void defaultResumptionHandler(ht_fill_limit_crossed &); } ) {
+forall( K, tN &, tE & | $dlistable(tN, tE) | hkey(K, tN) | { void defaultResumptionHandler(ht_fill_limit_crossed &); } ) {
     void ?{}( hashtable(K, tN, tE) & this, size_t n_buckets, dlist(tN, tE) *buckets ) {
 …
+}
 forall( otype K, dtype tN, dtype tE | $dlistable(tN, tE) | hkey(K, tN) ) {
+forall( K, tN &, tE & | $dlistable(tN, tE) | hkey(K, tN) ) {
     float fill_frac( hashtable(K, tN, tE) & this ) with(this) {
 …
 trait heaped(dtype T) {
+trait heaped(T &) {
     T * alloc( size_t );
     void free( void * );
 …
+}
 forall( otype K, dtype tN, dtype tE | $dlistable(tN, tE) | hkey(K, tN) | heaped( dlist(tN, tE) ) ) {
+forall( K, tN &, tE & | $dlistable(tN, tE) | hkey(K, tN) | heaped( dlist(tN, tE) ) ) {
     struct hashtable_dynamic {

tests/zombies/hashtable2.cfa

-              r92bfda0
+              rdafbde8
 trait pretendsToMatter( dtype TTT ) {
+trait pretendsToMatter( TTT & ) {
     void actsmart(TTT &);
 };
 forall( dtype TTTx )
+forall( TTTx & )
 void actsmart(TTTx &) {}
 …
 //   2. shows up in -CFA output as hashtable_rbs(), which is bad C; expecting hashtable_rbs*
 forall( otype Tt_unused | pretendsToMatter(Tt_unused) ) {
+forall( Tt_unused | pretendsToMatter(Tt_unused) ) {
     // hashtable of request by source
 …
+}
 forall( otype Tt_unused | pretendsToMatter(Tt_unused) | { void defaultResumptionHandler(ht_fill_limit_crossed &); } ) {
+forall( Tt_unused | pretendsToMatter(Tt_unused) | { void defaultResumptionHandler(ht_fill_limit_crossed &); } ) {
     void ?{}( hashtable_rbs(Tt_unused) & this, size_t n_buckets, dlist(request_in_ht_by_src, request) *buckets,
 …
 void defaultResumptionHandler( ht_auto_resize_pending & ex );
 forall( otype Tt_unused | pretendsToMatter(Tt_unused) ) {
+forall( Tt_unused | pretendsToMatter(Tt_unused) ) {
     float fill_frac( hashtable_rbs(Tt_unused) & this ) with(this) {
 …
 trait heaped(dtype T) {
+trait heaped(T &) {
     T * alloc( size_t );
     void free( void * );
 …
 void __dynamic_defaultResumptionHandler(ht_fill_limit_crossed &);
 forall( otype Tt_unused ) {
+forall( Tt_unused ) {
     struct hashtable_rbs_dynamic {
 …
 forall( otype Tt_unused | heaped( dlist(request_in_ht_by_src, request) ) ) {
+forall( Tt_unused | heaped( dlist(request_in_ht_by_src, request) ) ) {
     void ?{}( hashtable_rbs_dynamic(Tt_unused).resize_policy & this, size_t nbuckets_floor ) {
 …
+}
 forall( otype Tt_unused ) {
+forall( Tt_unused ) {
     void rehashToLarger_STEP( hashtable_rbs_dynamic(Tt_unused) & this, size_t new_n_buckets ) with (this) {
         rehashToLarger( this, new_n_buckets );

tests/zombies/huge.c

r92bfda0	rdafbde8
14	14	//
15	15
16		int huge( int n, forall( ~~otype~~ T ) T (*f)( T ) ) {
	16	int huge( int n, forall( T ) T (*f)( T ) ) {
17	17	if ( n <= 0 )
18	18	return f( 0 );

tests/zombies/it_out.c

-              r92bfda0
+              rdafbde8
 typedef unsigned long streamsize_type;
 trait ostream( dtype os_type ) {
+trait ostream( os_type & ) {
         os_type *write( os_type *, const char *, streamsize_type );
         int fail( os_type * );
 };
 trait writeable( otype T ) {
         forall( dtype os_type | ostream( os_type ) ) os_type * ?<<?( os_type *, T );
+trait writeable( T ) {
+        forall( os_type & | ostream( os_type ) ) os_type * ?<<?( os_type *, T );
 };
 forall( dtype os_type | ostream( os_type ) ) os_type * ?<<?( os_type *, char );
 forall( dtype os_type | ostream( os_type ) ) os_type * ?<<?( os_type *, int );
 forall( dtype os_type | ostream( os_type ) ) os_type * ?<<?( os_type *, const char * );
+forall( os_type & | ostream( os_type ) ) os_type * ?<<?( os_type *, char );
+forall( os_type & | ostream( os_type ) ) os_type * ?<<?( os_type *, int );
+forall( os_type & | ostream( os_type ) ) os_type * ?<<?( os_type *, const char * );
 trait istream( dtype is_type ) {
+trait istream( is_type & ) {
         is_type *read( is_type *, char *, streamsize_type );
         is_type *unread( is_type *, char );
 …
 };
 trait readable( otype T ) {
         forall( dtype is_type | istream( is_type ) ) is_type * ?<<?( is_type *, T );
+trait readable( T ) {
+        forall( is_type & | istream( is_type ) ) is_type * ?<<?( is_type *, T );
 };
 forall( dtype is_type | istream( is_type ) ) is_type * ?>>?( is_type *, char* );
 forall( dtype is_type | istream( is_type ) ) is_type * ?>>?( is_type *, int* );
+forall( is_type & | istream( is_type ) ) is_type * ?>>?( is_type *, char* );
+forall( is_type & | istream( is_type ) ) is_type * ?>>?( is_type *, int* );
 trait iterator( otype iterator_type, otype elt_type ) {
+trait iterator( iterator_type, elt_type ) {
         iterator_type ?++( iterator_type* );
         iterator_type ++?( iterator_type* );
 …
 };
 forall( otype elt_type | writeable( elt_type ),
                 otype iterator_type | iterator( iterator_type, elt_type ),
                 dtype os_type | ostream( os_type ) )
+forall( elt_type | writeable( elt_type ),
+                iterator_type | iterator( iterator_type, elt_type ),
+                os_type & | ostream( os_type ) )
 void write_all( iterator_type begin, iterator_type end, os_type *os );
 forall( otype elt_type | writeable( elt_type ),
                 otype iterator_type | iterator( iterator_type, elt_type ),
                 dtype os_type | ostream( os_type ) )
+forall( elt_type | writeable( elt_type ),
+                iterator_type | iterator( iterator_type, elt_type ),
+                os_type & | ostream( os_type ) )
 void write_all( elt_type begin, iterator_type end, os_type *os ) {
         os << begin;

tests/zombies/new.c

r92bfda0	rdafbde8
14	14	//
15	15
16		forall( ~~otype~~ T )
	16	forall( T )
17	17	void f( T *t ) {
18	18	t--;

tests/zombies/occursError.cfa

-              r92bfda0
+              rdafbde8
 forall( otype T ) void f( void (*)( T, T * ) );
 forall( otype U ) void g( U,  U * );
 forall( otype U ) void h( U *, U );
+forall( T ) void f( void (*)( T, T * ) );
+forall( U ) void g( U,  U * );
+forall( U ) void h( U *, U );
 void test() {

tests/zombies/prolog.c

-              r92bfda0
+              rdafbde8
 void is_integer( int x ) {}
 trait ArithmeticType( otype T ) {
+trait ArithmeticType( T ) {
         void is_arithmetic( T );
 };
 trait IntegralType( otype T | ArithmeticType( T ) ) {
+trait IntegralType( T | ArithmeticType( T ) ) {
         void is_integer( T );
 };
 forall( otype T | IntegralType( T ) | { void printResult( T ); } )
+forall( T | IntegralType( T ) | { void printResult( T ); } )
 void hornclause( T param ) {
         printResult( param );

tests/zombies/quad.c

-              r92bfda0
+              rdafbde8
 #include <fstream.hfa>
 forall( otype T | { T ?*?( T, T ); } )
+forall( T | { T ?*?( T, T ); } )
 T square( T t ) {
         return t * t;
+}
 forall( otype U | { U square( U ); } )
+forall( U | { U square( U ); } )
 U quad( U u ) {
         return square( square( u ) );

tests/zombies/scope.cfa

-              r92bfda0
+              rdafbde8
 y p;
 trait has_u( otype z ) {
+trait has_u( z ) {
         z u(z);
 };
 forall( otype t | has_u( t ) )
+forall( t | has_u( t ) )
 y q( t the_t ) {
         t y = u( the_t );

tests/zombies/simplePoly.c

r92bfda0	rdafbde8
14	14	//
15	15
16		forall( ~~otype T, otype~~ U \| { T f( T, U ); } )
	16	forall( T, U \| { T f( T, U ); } )
17	17	T q( T t, U u ) {
18	18	return f( t, u );

tests/zombies/simpler.c

r92bfda0	rdafbde8
14	14	//
15	15
16		forall( ~~otype~~ T ) T id( T, T );
	16	forall( T ) T id( T, T );
17	17
18	18	int main() {

tests/zombies/specialize.c

r92bfda0	rdafbde8
39	39	}
40	40
41		forall( ~~otype~~ T ) T f( T t )
	41	forall( T ) T f( T t )
42	42	{
43	43	printf( "in f; sizeof T is %d\n", sizeof( T ) );

tests/zombies/square.c

r92bfda0	rdafbde8
16	16	#include <fstream.hfa>
17	17
18		forall( ~~otype~~ T \| { T ?*?( T, T ); } )
	18	forall( T \| { T ?*?( T, T ); } )
19	19	T square( T t ) {
20	20	return t * t;

tests/zombies/structMember.cfa

r92bfda0	rdafbde8
66	66	S.T;
67	67	.S.T;
68		forall( ~~otype S, otype~~ T ) struct W {
	68	forall( S, T ) struct W {
69	69	struct X {};
70	70	};

tests/zombies/subrange.cfa

-              r92bfda0
+              rdafbde8
 // A small context defining the notion of an ordered otype.  (The standard
 // library should probably contain a context for this purpose.)
 trait ordered(otype T) {
+trait ordered(T) {
     int ?<?(T, T), ?<=?(T, T);
 };
 …
 // A subrange otype resembling an Ada subotype with a base otype and a range
 // constraint.
 otype subrange(otype base_t | ordered(base_t), base_t low = 0, base_t high = 8) = base_t;
+otype subrange(base_t | ordered(base_t), base_t low = 0, base_t high = 8) = base_t;
 // Note that subrange() can be applied to floating-point and pointer otypes, not
 …
 // Convenient access to subrange bounds, for instance for iteration:
 forall (otype T, T low, T high)
+forall (T, T low, T high)
 T lbound( subrange(T, low, high) v) {
     return low;
+}
 forall (otype T, T low, T high)
+forall (T, T low, T high)
 T hbound( subrange(T, low, high) v) {
     return high;
 …
 // of exception handling here.  Inlining allows the compiler to eliminate
 // bounds checks.
 forall (otype T | ordered(T), T low, T high)
+forall (T | ordered(T), T low, T high)
 inline subrange(T, low, high) ?=?(subrange(T, low, high)* target, T source) {
     if (low <= source && source <= high) *((T*)target) = source;
 …
 // compares range bounds so that the compiler can optimize checks away when the
 // ranges are known to overlap.
 forall (otype T | ordered(T), T t_low, T t_high, T s_low, T s_high)
+forall (T | ordered(T), T t_low, T t_high, T s_low, T s_high)
 inline subrange(T, t_low, t_high) ?=?(subrange(T, t_low, t_high)* target,
                                       subrange(T, s_low, s_high) source) {

tests/zombies/twice.c

r92bfda0	rdafbde8
16	16	#include <fstream.hfa>
17	17
18		forall( ~~otype~~ T \| { T ?+?( T, T ); } )
	18	forall( T \| { T ?+?( T, T ); } )
19	19	T twice( const T t ) {
20	20	return t + t;

tests/zombies/typeGenerator.cfa

-              r92bfda0
+              rdafbde8
 context addable( otype T ) {
+context addable( T ) {
         T ?+?( T,T );
         T ?=?( T*, T);
 };
 otype List1( otype T | addable( T ) ) = struct { T data; List1( T ) *next; } *;
+otype List1( T | addable( T ) ) = struct { T data; List1( T ) *next; } *;
 typedef List1( int ) ListOfIntegers;
 //List1( int ) li;
 …
 [int] h( * List1( int ) p );                                                    // new declaration syntax
 struct( otype T ) S2 { T i; };                                                  // actual definition
+struct( T ) S2 { T i; };                                                        // actual definition
 struct( int ) S3 v1, *p;                                                                // expansion and instantiation
 struct( otype T )( int ) S24 { T i; } v2;                               // actual definition, expansion and instantiation
 struct( otype T )( int ) { T i; } v2;                                   // anonymous actual definition, expansion and instantiation
+struct( T )( int ) S24 { T i; } v2;                             // actual definition, expansion and instantiation
+struct( T )( int ) { T i; } v2;                                 // anonymous actual definition, expansion and instantiation
 struct( otype T | addable( T ) ) node { T data; struct( T ) node *next; };
 otype List( otype T ) = struct( T ) node *;
+struct( T | addable( T ) ) node { T data; struct( T ) node *next; };
+otype List( T ) = struct( T ) node *;
 List( int ) my_list;

tests/zombies/withStatement.cfa

-              r92bfda0
+              rdafbde8
+}
 forall( otype T )
+forall( T )
 struct Box {
         T x;
 };
 forall( otype T )
+forall( T )
 void ?{}( Box(T) & this ) with( this ) { // with clause in polymorphic function
         x{};
 …
 void print( int i ) { sout | i; }
 forall( otype T | { void print( T ); })
+forall( T | { void print( T ); })
 void foo( T t ) {
         Box( T ) b = { t };

tests/zombies/wrapper/src/pointer.h

r92bfda0	rdafbde8
8	8	// type safe malloc / free
9	9
10		forall(~~otype~~ T)
	10	forall(T)
11	11	T* new()
12	12	{
…	…
16	16	}
17	17
18		forall(~~otype~~ T)
	18	forall(T)
19	19	void delete(T* p)
20	20	{

Context Navigation

Legend:

Download in other formats: