Changes in / [46438e4:05d55ff]


Ignore:
Files:
14 deleted
11 edited

Legend:

Unmodified
Added
Removed

  • doc/LaTeXmacros/common.tex

    r46438e4 r05d55ff  
    1111%% Created On       : Sat Apr  9 10:06:17 2016
    1212%% Last Modified By : Peter A. Buhr
    13 %% Last Modified On : Fri May 24 07:59:54 2019
    14 %% Update Count     : 382
     13%% Last Modified On : Mon Jul  9 08:28:05 2018
     14%% Update Count     : 380
    1515%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
    1616
    … …  
    6161\newlength{\gcolumnposn}                                % temporary hack because lstlisting does not handle tabs correctly
    6262\newlength{\columnposn}
    63 \setlength{\gcolumnposn}{2.75in}
     63\setlength{\gcolumnposn}{2.5in}
    6464\setlength{\columnposn}{\gcolumnposn}
    6565\newcommand{\C}[2][\@empty]{\ifx#1\@empty\else\global\setlength{\columnposn}{#1}\global\columnposn=\columnposn\fi\hfill\makebox[\textwidth-\columnposn][l]{\lst@basicstyle{\LstCommentStyle{#2}}}}

  • doc/bibliography/pl.bib

    r46438e4 r05d55ff  
    25852585
    25862586@article{Tarjan75,
    2587     keywords    = {union-find},
    2588     contributer = {a3moss@uwaterloo.ca},
    2589     author      = {Tarjan, Robert Endre},
    2590     title       = {Efficiency of a Good But Not Linear Set Union Algorithm},
    2591     journal     = {J. ACM},
    2592     issue_date  = {April 1975},
    2593     volume      = {22},
    2594     number      = {2},
    2595     month       = apr,
    2596     year        = {1975},
    2597     issn        = {0004-5411},
    2598     pages       = {215--225},
    2599     numpages    = {11},
    2600     url         = {http://doi.acm.org/10.1145/321879.321884},
    2601     doi         = {10.1145/321879.321884},
    2602     acmid       = {321884},
    2603     publisher   = {ACM},
    2604     address     = {New York, NY, USA},
     2587 keywords = {union-find},
     2588 contributer = {a3moss@uwaterloo.ca},
     2589 author = {Tarjan, Robert Endre},
     2590 title = {Efficiency of a Good But Not Linear Set Union Algorithm},
     2591 journal = {J. ACM},
     2592 issue_date = {April 1975},
     2593 volume = {22},
     2594 number = {2},
     2595 month = apr,
     2596 year = {1975},
     2597 issn = {0004-5411},
     2598 pages = {215--225},
     2599 numpages = {11},
     2600 url = {http://doi.acm.org/10.1145/321879.321884},
     2601 doi = {10.1145/321879.321884},
     2602 acmid = {321884},
     2603 publisher = {ACM},
     2604 address = {New York, NY, USA},
    26052605}
    26062606
    … …  
    26232623    journal     = ipl,
    26242624    year        = 1980,
    2625     month       = apr,
    2626     volume      = 10,
    2627     number      = 3,
    2628     pages        = {120-123},
     2625    month       = apr, volume = 10, number = 3, pages = {120-123},
    26292626    comment     = {
    26302627        The ``two-pass'' algorithm.  An upward pass over a parse tree
    … …  
    26602657}
    26612658
    2662 @inproceedings{chambers89a,
     2659@InProceedings{chambers89a,
    26632660    keywords    = {maps, delegation},
    26642661    author      = "Craig Chambers and David Ungar and Elgin Lee",
    2665     title       = "An Efficient Implementation of {SELF}, a Dynamically-Typed Object-Oriented Language Based on Prototypes",
     2662    title       = "An Efficient Implementation of {SELF}, a Dynamically-Typed
     2663                 Object-Oriented Language Based on Prototypes",
    26662664    crossref    = "OOPSLA89",
    26672665    pages       = {49-70}
    26682666}
    26692667
    2670 @misc{Turley99,
    2671     keywords    = {embedded system, micrprocessor},
    2672     contributer = {pabuhr@plg},
    2673     author      = {Jim Turley},
    2674     title       = {Embedded Processors by the Numbers},
    2675     year        = 1999,
    2676     month       = may,
    2677     note        = {Electronic Engineering Times},
    2678     howpublished= {\href{https://www.eetimes.com/author.asp?sectionid=36&doc_id=1287712}
    2679                   {https://\-www.eetimes.com/\-author.asp?sectionid=\-36&doc_id=1287712}},
    2680 }
    2681 
    26822668@article{oop:encapsulation,
    26832669    keywords    = {Encapsulation, Inheritance, Subclasses, Multiple Inheritance},
    26842670    contributer = {gjditchfield@plg},
    26852671    author      = {Alan Snyder},
    2686     title       = {Encapsulation and Inheritance in Object-Oriented Programming Languages},
     2672    title       = {Encapsulation and Inheritance in Object-Oriented Programming
     2673        Languages},
    26872674    journal     = sigplan,
    26882675    volume      = {21},    number = {11},
    … …  
    27192706    title       = {Encapsulators: A New Software Paradigm in Smalltalk-80},
    27202707    journal     = sigplan,
    2721     volume      = {21},
    2722     number      = {11},
     2708    volume      = {21},    number       = {11},
    27232709    pages       = {341-346},
    2724     month       = nov,
    2725     year        = 1986,
     2710    month       = nov, year = 1986,
    27262711    comment     = {
    27272712        Encapsulators are objects that surround other objects.

  • doc/papers/concurrency/Makefile

    r46438e4 r05d55ff  
    2424
    2525PICTURES = ${addsuffix .pstex, \
    26 FullProdConsStack \
    27 FullCoroutinePhases \
    28 corlayout \
    2926monitor \
    3027ext_monitor \

  • doc/papers/concurrency/Paper.tex

    r46438e4 r05d55ff  
    153153                auto, _Bool, catch, catchResume, choose, _Complex, __complex, __complex__, __const, __const__,
    154154                coroutine, disable, dtype, enable, exception, __extension__, fallthrough, fallthru, finally,
    155                 __float80, float80, __float128, float128, forall, ftype, generator, _Generic, _Imaginary, __imag, __imag__,
     155                __float80, float80, __float128, float128, forall, ftype, _Generic, _Imaginary, __imag, __imag__,
    156156                inline, __inline, __inline__, __int128, int128, __label__, monitor, mutex, _Noreturn, one_t, or,
    157                 otype, restrict, __restrict, __restrict__, __signed, __signed__, _Static_assert, suspend, thread,
     157                otype, restrict, __restrict, __restrict__, __signed, __signed__, _Static_assert, thread,
    158158                _Thread_local, throw, throwResume, timeout, trait, try, ttype, typeof, __typeof, __typeof__,
    159159                virtual, __volatile, __volatile__, waitfor, when, with, zero_t},
    … …  
    231231}
    232232
    233 \newbox\myboxA
    234 \newbox\myboxB
    235 \newbox\myboxC
    236 \newbox\myboxD
    237 
    238233\title{\texorpdfstring{Advanced Control-flow and Concurrency in \protect\CFA}{Advanced Control-flow in Cforall}}
    239234
    … …  
    252247This paper discusses the design philosophy and implementation of its advanced control-flow and concurrent/parallel features, along with the supporting runtime.
    253248These features are created from scratch as ISO C has only low-level and/or unimplemented concurrency, so C programmers continue to rely on library features like C pthreads.
    254 \CFA introduces modern language-level control-flow mechanisms, like generators, coroutines, user-level threading, and monitors for mutual exclusion and synchronization.
     249\CFA introduces modern language-level control-flow mechanisms, like coroutines, user-level threading, and monitors for mutual exclusion and synchronization.
    255250Library extension for executors, futures, and actors are built on these basic mechanisms.
    256 The runtime provides significant programmer simplification and safety by eliminating spurious wakeup and optional monitor barging.
     251The runtime provides significant programmer simplification and safety by eliminating spurious wakeup and reducing monitor barging.
    257252The runtime also ensures multiple monitors can be safely acquired \emph{simultaneously} (deadlock free), and this feature is fully integrated with all monitor synchronization mechanisms.
    258 All control-flow features integrate with the \CFA polymorphic type-system and exception handling, while respecting the expectations and style of C programmers.
     253All language features integrate with the \CFA polymorphic type-system and exception handling, while respecting the expectations and style of C programmers.
    259254Experimental results show comparable performance of the new features with similar mechanisms in other concurrent programming-languages.
    260255}%
    261256
    262 \keywords{generator, coroutine, concurrency, parallelism, thread, monitor, runtime, C, \CFA (Cforall)}
     257\keywords{coroutines, concurrency, parallelism, threads, monitors, runtime, C, \CFA (Cforall)}
    263258
    264259
    … …  
    290285As a result, languages like Java, Scala~\cite{Scala}, Objective-C~\cite{obj-c-book}, \CCeleven~\cite{C11}, and C\#~\cite{Csharp} adopt the 1:1 kernel-threading model, with a variety of presentation mechanisms.
    291286From 2000 onwards, languages like Go~\cite{Go}, Erlang~\cite{Erlang}, Haskell~\cite{Haskell}, D~\cite{D}, and \uC~\cite{uC++,uC++book} have championed the M:N user-threading model, and many user-threading libraries have appeared~\cite{Qthreads,MPC,BoostThreads}, including putting green threads back into Java~\cite{Quasar}.
    292 The main argument for user-level threading is that they are lighter weight than kernel threads (locking and context switching do not cross the kernel boundary), so there is less restriction on programming styles that encourage large numbers of threads performing medium work-units to facilitate load balancing by the runtime~\cite{Verch12}.
     287The main argument for user-level threading is that they are lighter weight than kernel threads (locking and context switching do not cross the kernel boundary), so there is less restriction on programming styles that encourage large numbers of threads performing smaller work-units to facilitate load balancing by the runtime~\cite{Verch12}.
    293288As well, user-threading facilitates a simpler concurrency approach using thread objects that leverage sequential patterns versus events with call-backs~\cite{vonBehren03}.
    294289Finally, performant user-threading implementations (both time and space) are largely competitive with direct kernel-threading implementations, while achieving the programming advantages of high concurrency levels and safety.
    … …  
    305300
    306301Finally, it is important for a language to provide safety over performance \emph{as the default}, allowing careful reduction of safety for performance when necessary.
    307 Two concurrency violations of this philosophy are \emph{spurious wakeup} and \emph{barging}, i.e., random wakeup~\cite[\S~8]{Buhr05a} and signals-as-hints~\cite[\S~8]{Buhr05a}, where one is a consequence of the other, i.e., once there is spurious wakeup, signals-as-hints follows.
    308 However, spurious wakeup is \emph{not} a foundational concurrency property~\cite[\S~8]{Buhr05a}, it is a performance design choice.
    309 Similarly, signals-as-hints is also a performance decision.
    310 We argue removing spurious wakeup and signals-as-hints makes concurrent programming significantly safer because it removes local non-determinism and matches with programmer expectation.
    311 (Experience by the authors teaching concurrency is that students are highly confused by these semantics.)
    312 Clawing back performance, when local non-determinism is unimportant, should be an option not the default.
     302Two concurrency violations of this philosophy are \emph{spurious wakeup} and \emph{barging}, i.e., random wakeup~\cite[\S~8]{Buhr05a} and signalling-as-hints~\cite[\S~8]{Buhr05a}, where one begats the other.
     303If you believe spurious wakeup is a foundational concurrency property, than unblocking (signalling) a thread is always a hint.
     304If you \emph{do not} believe spurious wakeup is foundational, than signalling-as-hints is a performance decision.
     305Most importantly, removing spurious wakeup and signals-as-hints makes concurrent programming significantly safer because it removes local non-determinism.
     306Clawing back performance where the local non-determinism is unimportant, should be an option not the default.
    313307
    314308\begin{comment}
    … …  
    333327
    334328\CFA embraces user-level threading, language extensions for advanced control-flow, and safety as the default.
    335 We present comparative examples so the reader can judge if the \CFA control-flow extensions are better and safer than those in or proposed for \Celeven, \CC, and other concurrent, imperative programming-languages, and perform experiments to show the \CFA runtime is competitive with other similar mechanisms.
     329We present comparative examples so the reader can judge if the \CFA control-flow extensions are better and safer than those in or proposed for \Celeven, \CC and other concurrent, imperative programming languages, and perform experiments to show the \CFA runtime is competitive with other similar mechanisms.
    336330The main contributions of this work are:
    337331\begin{itemize}
    338332\item
    339 language-level generators, coroutines and user-level threading, which respect the expectations of C programmers.
     333expressive language-level coroutines and user-level threading, which respect the expectations of C programmers.
    340334\item
    341 monitor synchronization without barging, and the ability to safely acquiring multiple monitors \emph{simultaneously} (deadlock free), while seamlessly integrating these capability with all monitor synchronization mechanisms.
     335monitor synchronization without barging.
     336\item
     337safely acquiring multiple monitors \emph{simultaneously} (deadlock free), while seamlessly integrating this capability with all monitor synchronization mechanisms.
    342338\item
    343339providing statically type-safe interfaces that integrate with the \CFA polymorphic type-system and other language features.
    … …  
    347343a runtime system with no spurious wakeup.
    348344\item
    349 a dynamic partitioning mechanism to segregate the execution environment for specialized requirements.
    350 \item
    351 a non-blocking I/O library
    352 \item
    353 experimental results showing comparable performance of the new features with similar mechanisms in other programming languages.
     345experimental results showing comparable performance of the new features with similar mechanisms in other concurrent programming-languages.
    354346\end{itemize}
    355347
    356 
    357 \section{Stateful Function}
    358 
    359 The generator/coroutine provides a stateful function, which is an old idea~\cite{Conway63,Marlin80} that is new again~\cite{C++20Coroutine19}.
    360 A stateful function allows execution to be temporarily suspended and later resumed, e.g., plugin, device driver, finite-state machine.
    361 Hence, a stateful function may not end when it returns to its caller, allowing it to be restarted with the data and execution location present at the point of suspension.
    362 This capability is accomplished by retaining a data/execution \emph{closure} between invocations.
    363 If the closure is fixed size, we call it a \emph{generator} (or \emph{stackless}), and its control flow is restricted, e.g., suspending outside the generator is prohibited.
    364 If the closure is variable sized, we call it a \emph{coroutine} (or \emph{stackful}), and as the names implies, often implemented with a separate stack with no programming restrictions.
    365 Hence, refactoring a stackless coroutine may require changing it to stackful.
    366 A foundational property of all \emph{stateful functions} is that resume/suspend \emph{do not} cause incremental stack growth, i.e., resume/suspend operations are remembered through the closure not the stack.
    367 As well, activating a stateful function is \emph{asymmetric} or \emph{symmetric}, identified by resume/suspend (no cycles) and resume/resume (cycles).
    368 A fixed closure activated by modified call/return is faster than a variable closure activated by context switching.
    369 Additionally, any storage management for the closure (especially in unmanaged languages, i.e., no garbage collection) must also be factored into design and performance.
    370 Therefore, selecting between stackless and stackful semantics is a tradeoff between programming requirements and performance, where stackless is faster and stackful is more general.
    371 Note, creation cost is amortized across usage, so activation cost is usually the dominant factor.
    372 
    373 
    374 \begin{figure}
    375 \centering
    376 \begin{lrbox}{\myboxA}
    377 \begin{cfa}[aboveskip=0pt,belowskip=0pt]
    378 typedef struct {
    379         int fn1, fn;
    380 } Fib;
    381 #define FibCtor { 1, 0 }
    382 int fib( Fib * f ) {
    383 
    384 
    385 
    386         int fn = f->fn; f->fn = f->fn1;
    387                 f->fn1 = f->fn + fn;
    388         return fn;
    389 
    390 }
    391 int main() {
    392         Fib f1 = FibCtor, f2 = FibCtor;
    393         for ( int i = 0; i < 10; i += 1 )
    394                 printf( "%d %d\n",
    395                            fib( &f1 ), fib( &f2 ) );
    396 }
    397 \end{cfa}
    398 \end{lrbox}
    399 
    400 \begin{lrbox}{\myboxB}
    401 \begin{cfa}[aboveskip=0pt,belowskip=0pt]
    402 `generator` Fib {
    403         int fn1, fn;
     348\begin{comment}
     349This paper provides a minimal concurrency \newterm{Application Program Interface} (API) that is simple, efficient and can be used to build other concurrency features.
     350While the simplest concurrency system is a thread and a lock, this low-level approach is hard to master.
     351An easier approach for programmers is to support higher-level constructs as the basis of concurrency.
     352Indeed, for highly-productive concurrent-programming, high-level approaches are much more popular~\cite{Hochstein05}.
     353Examples of high-level approaches are jobs (thread pool)~\cite{TBB}, implicit threading~\cite{OpenMP}, monitors~\cite{Java}, channels~\cite{CSP,Go}, and message passing~\cite{Erlang,MPI}.
     354
     355The following terminology is used.
     356A \newterm{thread} is a fundamental unit of execution that runs a sequence of code and requires a stack to maintain state.
     357Multiple simultaneous threads give rise to \newterm{concurrency}, which requires locking to ensure access to shared data and safe communication.
     358\newterm{Locking}, and by extension \newterm{locks}, are defined as a mechanism to prevent progress of threads to provide safety.
     359\newterm{Parallelism} is running multiple threads simultaneously.
     360Parallelism implies \emph{actual} simultaneous execution, where concurrency only requires \emph{apparent} simultaneous execution.
     361As such, parallelism only affects performance, which is observed through differences in space and/or time at runtime.
     362Hence, there are two problems to be solved: concurrency and parallelism.
     363While these two concepts are often combined, they are distinct, requiring different tools~\cite[\S~2]{Buhr05a}.
     364Concurrency tools handle mutual exclusion and synchronization, while parallelism tools handle performance, cost, and resource utilization.
     365
     366The proposed concurrency API is implemented in a dialect of C, called \CFA (pronounced C-for-all).
     367The paper discusses how the language features are added to the \CFA translator with respect to parsing, semantics, and type checking, and the corresponding high-performance runtime-library to implement the concurrent features.
     368\end{comment}
     369
     370
     371\begin{comment}
     372\section{\CFA Overview}
     373
     374The following is a quick introduction to the \CFA language, specifically tailored to the features needed to support concurrency.
     375Extended versions and explanation of the following code examples are available at the \CFA website~\cite{Cforall} or in Moss~\etal~\cite{Moss18}.
     376
     377\CFA is a non-object-oriented extension of ISO-C, and hence, supports all C paradigms.
     378Like C, the building blocks of \CFA are structures and routines.
     379Virtually all of the code generated by the \CFA translator respects C memory layouts and calling conventions.
     380While \CFA is not object oriented, lacking the concept of a receiver (\eg @this@) and nominal inheritance-relationships, C has a notion of objects: ``region of data storage in the execution environment, the contents of which can represent values''~\cite[3.15]{C11}.
     381While some object-oriented features appear in \CFA, they are independent capabilities, allowing \CFA to adopt them while maintaining a procedural paradigm.
     382
     383
     384\subsection{References}
     385
     386\CFA provides multi-level rebindable references, as an alternative to pointers, which significantly reduces syntactic noise.
     387\begin{cfa}
     388int x = 1, y = 2, z = 3;
     389int * p1 = &x, ** p2 = &p1,  *** p3 = &p2,      $\C{// pointers to x}$
     390    `&` r1 = x,   `&&` r2 = r1,   `&&&` r3 = r2;        $\C{// references to x}$
     391int * p4 = &z, `&` r4 = z;
     392
     393*p1 = 3; **p2 = 3; ***p3 = 3;       // change x
     394r1 =  3;     r2 = 3;      r3 = 3;        // change x: implicit dereferences *r1, **r2, ***r3
     395**p3 = &y; *p3 = &p4;                // change p1, p2
     396`&`r3 = &y; `&&`r3 = &`&`r4;             // change r1, r2: cancel implicit dereferences (&*)**r3, (&(&*)*)*r3, &(&*)r4
     397\end{cfa}
     398A reference is a handle to an object, like a pointer, but is automatically dereferenced the specified number of levels.
     399Referencing (address-of @&@) a reference variable cancels one of the implicit dereferences, until there are no more implicit references, after which normal expression behaviour applies.
     400
     401
     402\subsection{\texorpdfstring{\protect\lstinline{with} Statement}{with Statement}}
     403\label{s:WithStatement}
     404
     405Heterogeneous data is aggregated into a structure/union.
     406To reduce syntactic noise, \CFA provides a @with@ statement (see Pascal~\cite[\S~4.F]{Pascal}) to elide aggregate field-qualification by opening a scope containing the field identifiers.
     407\begin{cquote}
     408\vspace*{-\baselineskip}%???
     409\lstDeleteShortInline@%
     410\begin{cfa}
     411struct S { char c; int i; double d; };
     412struct T { double m, n; };
     413// multiple aggregate parameters
     414\end{cfa}
     415\begin{tabular}{@{}l@{\hspace{2\parindentlnth}}|@{\hspace{2\parindentlnth}}l@{}}
     416\begin{cfa}
     417void f( S & s, T & t ) {
     418        `s.`c; `s.`i; `s.`d;
     419        `t.`m; `t.`n;
     420}
     421\end{cfa}
     422&
     423\begin{cfa}
     424void f( S & s, T & t ) `with ( s, t )` {
     425        c; i; d;                // no qualification
     426        m; n;
     427}
     428\end{cfa}
     429\end{tabular}
     430\lstMakeShortInline@%
     431\end{cquote}
     432Object-oriented programming languages only provide implicit qualification for the receiver.
     433
     434In detail, the @with@-statement syntax is:
     435\begin{cfa}
     436$\emph{with-statement}$:
     437        'with' '(' $\emph{expression-list}$ ')' $\emph{compound-statement}$
     438\end{cfa}
     439and may appear as the body of a routine or nested within a routine body.
     440Each expression in the expression-list provides a type and object.
     441The type must be an aggregate type.
     442(Enumerations are already opened.)
     443The object is the implicit qualifier for the open structure-fields.
     444All expressions in the expression list are opened in parallel within the compound statement, which is different from Pascal, which nests the openings from left to right.
     445
     446
     447\subsection{Overloading}
     448
     449\CFA maximizes the ability to reuse names via overloading to aggressively address the naming problem.
     450Both variables and routines may be overloaded, where selection is based on number and types of returns and arguments (as in Ada~\cite{Ada}).
     451\newpage
     452\vspace*{-2\baselineskip}%???
     453\begin{cquote}
     454\begin{cfa}
     455// selection based on type
     456\end{cfa}
     457\lstDeleteShortInline@%
     458\begin{tabular}{@{}l@{\hspace{2\parindentlnth}}|@{\hspace{2\parindentlnth}}l@{}}
     459\begin{cfa}
     460const short int `MIN` = -32768;
     461const int `MIN` = -2147483648;
     462const long int `MIN` = -9223372036854775808L;
     463\end{cfa}
     464&
     465\begin{cfa}
     466short int si = `MIN`;
     467int i = `MIN`;
     468long int li = `MIN`;
     469\end{cfa}
     470\end{tabular}
     471\begin{cfa}
     472// selection based on type and number of parameters
     473\end{cfa}
     474\begin{tabular}{@{}l@{\hspace{2.7\parindentlnth}}|@{\hspace{2\parindentlnth}}l@{}}
     475\begin{cfa}
     476void `f`( void );
     477void `f`( char );
     478void `f`( int, double );
     479\end{cfa}
     480&
     481\begin{cfa}
     482`f`();
     483`f`( 'a' );
     484`f`( 3, 5.2 );
     485\end{cfa}
     486\end{tabular}
     487\begin{cfa}
     488// selection based on type and number of returns
     489\end{cfa}
     490\begin{tabular}{@{}l@{\hspace{2\parindentlnth}}|@{\hspace{2\parindentlnth}}l@{}}
     491\begin{cfa}
     492char `f`( int );
     493double `f`( int );
     494[char, double] `f`( int );
     495\end{cfa}
     496&
     497\begin{cfa}
     498char c = `f`( 3 );
     499double d = `f`( 3 );
     500[d, c] = `f`( 3 );
     501\end{cfa}
     502\end{tabular}
     503\lstMakeShortInline@%
     504\end{cquote}
     505Overloading is important for \CFA concurrency since the runtime system relies on creating different types to represent concurrency objects.
     506Therefore, overloading eliminates long prefixes and other naming conventions to prevent name clashes.
     507As seen in Section~\ref{s:Concurrency}, routine @main@ is heavily overloaded.
     508As another example, variable overloading is useful in the parallel semantics of the @with@ statement for fields with the same name:
     509\begin{cfa}
     510struct S { int `i`; int j; double m; } s;
     511struct T { int `i`; int k; int m; } t;
     512with ( s, t ) {
     513        j + k;                                                                  $\C{// unambiguous, s.j + t.k}$
     514        m = 5.0;                                                                $\C{// unambiguous, s.m = 5.0}$
     515        m = 1;                                                                  $\C{// unambiguous, t.m = 1}$
     516        int a = m;                                                              $\C{// unambiguous, a = t.m }$
     517        double b = m;                                                   $\C{// unambiguous, b = s.m}$
     518        int c = `s.i` + `t.i`;                                  $\C{// unambiguous, qualification}$
     519        (double)m;                                                              $\C{// unambiguous, cast s.m}$
     520}
     521\end{cfa}
     522For parallel semantics, both @s.i@ and @t.i@ are visible with the same type, so only @i@ is ambiguous without qualification.
     523
     524
     525\subsection{Operators}
     526
     527Overloading also extends to operators.
     528Operator-overloading syntax creates a routine name with an operator symbol and question marks for the operands:
     529\begin{cquote}
     530\lstDeleteShortInline@%
     531\begin{tabular}{@{}ll@{\hspace{\parindentlnth}}|@{\hspace{\parindentlnth}}l@{}}
     532\begin{cfa}
     533int ++?(int op);
     534int ?++(int op);
     535int `?+?`(int op1, int op2);
     536int ?<=?(int op1, int op2);
     537int ?=? (int & op1, int op2);
     538int ?+=?(int & op1, int op2);
     539\end{cfa}
     540&
     541\begin{cfa}
     542// unary prefix increment
     543// unary postfix increment
     544// binary plus
     545// binary less than
     546// binary assignment
     547// binary plus-assignment
     548\end{cfa}
     549&
     550\begin{cfa}
     551struct S { int i, j; };
     552S `?+?`( S op1, S op2) { // add two structures
     553        return (S){op1.i + op2.i, op1.j + op2.j};
     554}
     555S s1 = {1, 2}, s2 = {2, 3}, s3;
     556s3 = s1 `+` s2;         // compute sum: s3 == {2, 5}
     557\end{cfa}
     558\end{tabular}
     559\lstMakeShortInline@%
     560\end{cquote}
     561
     562
     563\subsection{Constructors / Destructors}
     564
     565Object lifetime is a challenge in non-managed programming languages.
     566\CFA responds with \CC-like constructors and destructors, using a different operator-overloading syntax.
     567\begin{cfa}
     568struct VLA { int len, * data; };                        $\C{// variable length array of integers}$
     569void ?{}( VLA & vla ) with ( vla ) { len = 10;  data = alloc( len ); }  $\C{// default constructor}$
     570void ?{}( VLA & vla, int size, char fill ) with ( vla ) { len = size;  data = alloc( len, fill ); } // initialization
     571void ?{}( VLA & vla, VLA other ) { vla.len = other.len;  vla.data = other.data; } $\C{// copy, shallow}$
     572void ^?{}( VLA & vla ) with ( vla ) { free( data ); } $\C{// destructor}$
     573{
     574        VLA  x,            y = { 20, 0x01 },     z = y; $\C{// z points to y}$
     575        // $\LstCommentStyle{\color{red}\ \ \ x\{\};\ \ \ \ \ \ \ \ \ y\{ 20, 0x01 \};\ \ \ \ \ \ \ \ \ \ z\{ z, y \};\ \ \ \ \ \ \ implicit calls}$
     576        ^x{};                                                                   $\C{// deallocate x}$
     577        x{};                                                                    $\C{// reallocate x}$
     578        z{ 5, 0xff };                                                   $\C{// reallocate z, not pointing to y}$
     579        ^y{};                                                                   $\C{// deallocate y}$
     580        y{ x };                                                                 $\C{// reallocate y, points to x}$
     581        x{};                                                                    $\C{// reallocate x, not pointing to y}$
     582}       //  $\LstCommentStyle{\color{red}\^{}z\{\};\ \ \^{}y\{\};\ \ \^{}x\{\};\ \ \ implicit calls}$
     583\end{cfa}
     584Like \CC, construction is implicit on allocation (stack/heap) and destruction is implicit on deallocation.
     585The object and all their fields are constructed/destructed.
     586\CFA also provides @new@ and @delete@ as library routines, which behave like @malloc@ and @free@, in addition to constructing and destructing objects:
     587\begin{cfa}
     588{
     589        ... struct S s = {10}; ...                              $\C{// allocation, call constructor}$
     590}                                                                                       $\C{// deallocation, call destructor}$
     591struct S * s = new();                                           $\C{// allocation, call constructor}$
     592...
     593delete( s );                                                            $\C{// deallocation, call destructor}$
     594\end{cfa}
     595\CFA concurrency uses object lifetime as a means of mutual exclusion and/or synchronization.
     596
     597
     598\subsection{Parametric Polymorphism}
     599\label{s:ParametricPolymorphism}
     600
     601The signature feature of \CFA is parametric-polymorphic routines~\cite{Cforall} with routines generalized using a @forall@ clause (giving the language its name), which allow separately compiled routines to support generic usage over multiple types.
     602For example, the following sum routine works for any type that supports construction from 0 and addition:
     603\begin{cfa}
     604forall( otype T | { void `?{}`( T *, zero_t ); T `?+?`( T, T ); } ) // constraint type, 0 and +
     605T sum( T a[$\,$], size_t size ) {
     606        `T` total = { `0` };                                    $\C{// initialize by 0 constructor}$
     607        for ( size_t i = 0; i < size; i += 1 )
     608                total = total `+` a[i];                         $\C{// select appropriate +}$
     609        return total;
     610}
     611S sa[5];
     612int i = sum( sa, 5 );                                           $\C{// use S's 0 construction and +}$
     613\end{cfa}
     614Type variables can be @otype@ or @dtype@.
     615@otype@ refers to a \emph{complete type}, \ie, a type with size, alignment, default constructor, copy constructor, destructor, and assignment operator.
     616@dtype@ refers to an \emph{incomplete type}, \eg, void or a forward-declared type.
     617The builtin types @zero_t@ and @one_t@ overload constant 0 and 1 for a new types, where both 0 and 1 have special meaning in C.
     618
     619\CFA provides \newterm{traits} to name a group of type assertions, where the trait name allows specifying the same set of assertions in multiple locations, preventing repetition mistakes at each routine declaration:
     620\begin{cfa}
     621trait `sumable`( otype T ) {
     622        void `?{}`( T &, zero_t );                              $\C{// 0 literal constructor}$
     623        T `?+?`( T, T );                                                $\C{// assortment of additions}$
     624        T ?+=?( T &, T );
     625        T ++?( T & );
     626        T ?++( T & );
    404627};
    405 
    406 void `main(Fib & fib)` with(fib) {
    407 
    408         [fn1, fn] = [1, 0];
    409         for () {
    410                 `suspend;`
    411                 [fn1, fn] = [fn, fn + fn1];
    412 
    413         }
    414 }
    415 int main() {
    416         Fib f1, f2;
    417         for ( 10 )
    418                 sout | `resume( f1 )`.fn
    419                          | `resume( f2 )`.fn;
    420 }
    421 \end{cfa}
    422 \end{lrbox}
    423 
    424 \begin{lrbox}{\myboxC}
    425 \begin{cfa}[aboveskip=0pt,belowskip=0pt]
    426 typedef struct {
    427         int fn1, fn;  void * `next`;
    428 } Fib;
    429 #define FibCtor { 1, 0, NULL }
    430 Fib * comain( Fib * f ) {
    431         if ( f->next ) goto *f->next;
    432         f->next = &&s1;
    433         for ( ;; ) {
    434                 return f;
    435           s1:; int fn = f->fn + f->fn1;
    436                         f->fn1 = f->fn; f->fn = fn;
    437         }
    438 }
    439 int main() {
    440         Fib f1 = FibCtor, f2 = FibCtor;
    441         for ( int i = 0; i < 10; i += 1 )
    442                 printf("%d %d\n",comain(&f1)->fn,
    443                                  comain(&f2)->fn);
    444 }
    445 \end{cfa}
    446 \end{lrbox}
    447 
    448 \subfloat[C asymmetric generator]{\label{f:CFibonacci}\usebox\myboxA}
    449 \hspace{3pt}
    450 \vrule
    451 \hspace{3pt}
    452 \subfloat[\CFA asymmetric generator]{\label{f:CFAFibonacciGen}\usebox\myboxB}
    453 \hspace{3pt}
    454 \vrule
    455 \hspace{3pt}
    456 \subfloat[C generator implementation]{\label{f:CFibonacciSim}\usebox\myboxC}
    457 \caption{Fibonacci (output) Asymmetric Generator}
    458 \label{f:FibonacciAsymmetricGenerator}
    459 
    460 \bigskip
    461 
    462 \begin{lrbox}{\myboxA}
    463 \begin{cfa}[aboveskip=0pt,belowskip=0pt]
    464 `generator Fmt` {
    465         char ch;
    466         int g, b;
    467 };
    468 void ?{}( Fmt & fmt ) { `resume(fmt);` } // constructor
    469 void ^?{}( Fmt & f ) with(f) { $\C[1.75in]{// destructor}$
    470         if ( g != 0 || b != 0 ) sout | nl; }
    471 void `main( Fmt & f )` with(f) {
    472         for () { $\C{// until destructor call}$
    473                 for ( ; g < 5; g += 1 ) { $\C{// groups}$
    474                         for ( ; b < 4; b += 1 ) { $\C{// blocks}$
    475                                 `suspend;` $\C{// wait for character}$
    476                                 while ( ch == '\n' ) `suspend;` // ignore
    477                                 sout | ch;                                              // newline
    478                         } sout | " ";  // block spacer
    479                 } sout | nl; // group newline
    480         }
    481 }
    482 int main() {
    483         Fmt fmt; $\C{// fmt constructor called}$
    484         for () {
    485                 sin | fmt.ch; $\C{// read into generator}$
    486           if ( eof( sin ) ) break;
    487                 `resume( fmt );`
    488         }
    489 
    490 } $\C{// fmt destructor called}\CRT$
    491 \end{cfa}
    492 \end{lrbox}
    493 
    494 \begin{lrbox}{\myboxB}
    495 \begin{cfa}[aboveskip=0pt,belowskip=0pt]
    496 typedef struct {
    497         void * next;
    498         char ch;
    499         int g, b;
    500 } Fmt;
    501 void comain( Fmt * f ) {
    502         if ( f->next ) goto *f->next;
    503         f->next = &&s1;
    504         for ( ;; ) {
    505                 for ( f->g = 0; f->g < 5; f->g += 1 ) {
    506                         for ( f->b = 0; f->b < 4; f->b += 1 ) {
    507                                 return;
    508                           s1:;  while ( f->ch == '\n' ) return;
    509                                 printf( "%c", f->ch );
    510                         } printf( " " );
    511                 } printf( "\n" );
    512         }
    513 }
    514 int main() {
    515         Fmt fmt = { NULL };  comain( &fmt ); // prime
    516         for ( ;; ) {
    517                 scanf( "%c", &fmt.ch );
    518           if ( feof( stdin ) ) break;
    519                 comain( &fmt );
    520         }
    521         if ( fmt.g != 0 || fmt.b != 0 ) printf( "\n" );
    522 }
    523 \end{cfa}
    524 \end{lrbox}
    525 
    526 \subfloat[\CFA asymmetric generator]{\label{f:CFAFormatGen}\usebox\myboxA}
    527 \hspace{3pt}
    528 \vrule
    529 \hspace{3pt}
    530 \subfloat[C generator simulation]{\label{f:CFormatSim}\usebox\myboxB}
    531 \hspace{3pt}
    532 \caption{Formatter (input) Asymmetric Generator}
    533 \label{f:FormatterAsymmetricGenerator}
    534 \end{figure}
    535 
    536 
    537 \subsection{Generator}
    538 
    539 Stackless generators have the potential to be very small and fast, \ie as small and fast as function call/return for both creation and execution.
    540 The \CFA goal is to achieve this performance target, possibly at the cost of some semantic complexity.
    541 How this goal is accomplished is done through a series of different kinds of generators and their implementation.
    542 
    543 Figure~\ref{f:FibonacciAsymmetricGenerator} shows an unbounded asymmetric generator for an infinite sequence of Fibonacci numbers written in C and \CFA, with a simple C implementation for the \CFA version.
    544 This kind of generator is an \emph{output generator}, producing a new result on each resumption.
    545 To compute Fibonacci, the previous two values in the sequence are retained to generate the next value, \ie @fn1@ and @fn@, plus the execution location where control restarts when the generator is resumed, \ie top or middle.
    546 An additional requirement is the ability to create an arbitrary number of generators (of any kind), \ie retaining state in global variables is insufficient;
    547 hence, state is retained in a closure between calls.
    548 Figure~\ref{f:CFibonacci} shows the C approach of manually creating the closure in structure @Fib@, and multiple instances of this closure provide multiple Fibonacci generators.
    549 The C version only has the middle execution state because the top execution state becomes declaration initialization.
    550 Figure~\ref{f:CFAFibonacciGen} shows the \CFA approach, which also has a manual closure, but replaces the structure with a special \CFA @generator@ type.
    551 This generator type is then connected to a function named @main@ that takes as its only parameter a reference to the generator type, called a \emph{generator main}.
    552 The generator main contains @suspend@ statements that suspend execution without ending the generator versus @return@.
    553 For the Fibonacci generator-main,\footnote{
    554 The \CFA \lstinline|with| opens an aggregate scope making its fields directly accessible, like Pascal \lstinline|with|, but using parallel semantics.
    555 Multiple aggregates may be opened.}
    556 the top initialization state appears at the start and the middle execution state is denoted by statement @suspend@.
    557 Any local variables in @main@ \emph{are not retained} between calls;
    558 hence local variable are only for temporary computations \emph{between} suspends.
    559 All retained state \emph{must} appear in the generator's type.
    560 As well, generator code containing a @suspend@ cannot be refactored into a helper function called by the generator, because @suspend@ is implemented via @return@, so a return from the helper function goes back to the current generator not the resumer.
    561 The generator is started by calling function @resume@ with a generator instance, which begins execution at the top of the generator main, and subsequent @resume@ calls restart the generator at its point of last suspension.
    562 Resuming an ended (returned) generator is undefined.
    563 Function @resume@ returns its argument generator so it can be cascaded in an expression, in this case to print the next Fibonacci value @fn@ computed in the generator instance.
    564 Figure~\ref{f:CFibonacciSim} shows the C implementation of the \CFA generator only needs one additional field, @next@, to handle retention of execution state.
    565 The computed @goto@ at the start of the generator main, which branches after the previous suspend, adds very little cost to the resume call.
    566 Finally, an explicit generator type provides both design and performance benefits, such as multiple type-safe interface functions taking and returning arbitrary types.
    567 \begin{cfa}
    568 int ?()( Fib & fib ) with( fib ) { return `resume( fib )`.fn; }   // function-call interface
    569 int ?()( Fib & fib, int N ) with( fib ) { for ( N - 1 ) `fib()`; return `fib()`; }   // use simple interface
    570 double ?()( Fib & fib ) with( fib ) { return (int)`fib()` / 3.14159; } // cast prevents recursive call
    571 sout | (int)f1() | (double)f1() | f2( 2 );   // simple interface, cast selects call based on return type, step 2 values
    572 \end{cfa}
    573 Now, the generator can be a separately-compiled opaque-type only accessed through its interface functions.
    574 For contrast, Figure~\ref{f:PythonFibonacci} shows the equivalent Python Fibonacci generator, which does not use a generator type, and hence only has a single interface, but an implicit closure.
    575 
    576 Having to manually create the generator closure by moving local-state variables into the generator type is an additional programmer burden and possible point of error.
    577 (This restriction is removed by the coroutine, see Section~\ref{s:Coroutine}.)
    578 Our concern is that static analysis to discriminate local state from temporary variables is complex and beyond the current scope of the \CFA project.
    579 As well, supporting variable-size local-state, like variable-length arrays, requires dynamic allocation of the local state, which significantly increases the cost of generator creation/destruction, and is a show-stopper for embedded programming.
    580 But more importantly, the size of the generator type is tied to the local state in the generator main, which precludes separate compilation of the generator main, \ie a generator must be inlined or local state must be dynamically allocated.
    581 Our current experience is that most generator problems have simple data state, including local state, but complex execution state.
    582 As well, C programmers are not afraid with this kind of semantic programming requirement, if it results in very small, fast generators.
    583 
    584 Figure~\ref{f:CFAFormatGen} shows an asymmetric \newterm{input generator}, @Fmt@, for restructuring text into groups of characters of fixed-size blocks, \ie the input on the left is reformatted into the output on the right, where newlines are ignored.
    585 \begin{center}
    586 \tt
    587 \begin{tabular}{@{}l|l@{}}
    588 \multicolumn{1}{c|}{\textbf{\textrm{input}}} & \multicolumn{1}{c}{\textbf{\textrm{output}}} \\
    589 \begin{tabular}[t]{@{}ll@{}}
    590 abcdefghijklmnopqrstuvwxyz \\
    591 abcdefghijklmnopqrstuvwxyz
    592 \end{tabular}
    593 &
    594 \begin{tabular}[t]{@{}lllll@{}}
    595 abcd    & efgh  & ijkl  & mnop  & qrst  \\
    596 uvwx    & yzab  & cdef  & ghij  & klmn  \\
    597 opqr    & stuv  & wxyz  &               &
    598 \end{tabular}
    599 \end{tabular}
    600 \end{center}
    601 The example takes advantage of resuming a generator in the constructor to prime the loops so the first character sent for formatting appears inside the nested loops.
    602 The destructor provides a newline, if formatted text ends with a full line.
    603 Figure~\ref{f:CFormatSim} shows the C implementation of the \CFA input generator with one additional field and the computed @goto@.
    604 For contrast, Figure~\ref{f:PythonFormatter} shows the equivalent Python format generator with the same properties as the Fibonacci generator.
    605 
    606 Figure~\ref{f:DeviceDriverGen} shows a \emph{killer} asymmetric generator, a device-driver, because device drivers caused 70\%-85\% of failures in Windows/Linux~\cite{Swift05}.
    607 Device drives follow the pattern of simple data state but complex execution state, \ie finite state-machine (FSM) parsing a protocol.
    608 For example, the following protocol:
    609 \begin{center}
    610 \ldots\, STX \ldots\, message \ldots\, ESC ETX \ldots\, message \ldots\, ETX 2-byte crc \ldots
    611 \end{center}
    612 is a network message beginning with the control character STX, ending with an ETX, and followed by a 2-byte cyclic-redundancy check.
    613 Control characters may appear in a message if preceded by an ESC.
    614 When a message byte arrives, it triggers an interrupt, and the operating system services the interrupt by calling the device driver with the byte read from a hardware register.
    615 The device driver returns a status code of its current state, and when a complete message is obtained, the operating system knows the message is in the message buffer.
    616 Hence, the device driver is an input/output generator.
    617 
    618 Note, the cost of creating and resuming the device-driver generator, @Driver@, is virtually identical to call/return, so performance in an operating-system kernel is excellent.
    619 As well, the data state is small, where variables @byte@ and @msg@ are communication variables for passing in message bytes and returning the message, and variables @lnth@, @crc@, and @sum@ are local variable that must be retained between calls and are hoisted into the generator type.
    620 Manually, detecting and hoisting local-state variables is easy when the number is small.
    621 Finally, the execution state is large, with one @resume@ and seven @suspend@s.
    622 Hence, the key benefits of the generator are correctness, safety, and maintenance because the execution states of the FSM are transcribed directly into the programming language rather than using a table-driven approach.
    623 Because FSMs can be complex and occur frequently in important domains, direct support of the generator is crucial in a systems programming-language.
    624 
    625 \begin{figure}
    626 \centering
    627 \newbox\myboxA
    628 \begin{lrbox}{\myboxA}
    629 \begin{python}[aboveskip=0pt,belowskip=0pt]
    630 def Fib():
    631     fn1, fn = 0, 1
    632     while True:
    633         `yield fn1`
    634         fn1, fn = fn, fn1 + fn
    635 f1 = Fib()
    636 f2 = Fib()
    637 for i in range( 10 ):
    638         print( next( f1 ), next( f2 ) )
    639 
    640 
    641 
    642 
    643 
    644 
    645 \end{python}
    646 \end{lrbox}
    647 
    648 \newbox\myboxB
    649 \begin{lrbox}{\myboxB}
    650 \begin{python}[aboveskip=0pt,belowskip=0pt]
    651 def Fmt():
    652         try:
    653                 while True:
    654                         for g in range( 5 ):
    655                                 for b in range( 4 ):
    656                                         print( `(yield)`, end='' )
    657                                 print( '  ', end='' )
    658                         print()
    659         except GeneratorExit:
    660                 if g != 0 | b != 0:
    661                         print()
    662 fmt = Fmt()
    663 `next( fmt )`                    # prime, next prewritten
    664 for i in range( 41 ):
    665         `fmt.send( 'a' );`      # send to yield
    666 \end{python}
    667 \end{lrbox}
    668 \subfloat[Fibonacci]{\label{f:PythonFibonacci}\usebox\myboxA}
    669 \hspace{3pt}
    670 \vrule
    671 \hspace{3pt}
    672 \subfloat[Formatter]{\label{f:PythonFormatter}\usebox\myboxB}
    673 \caption{Python Generator}
    674 \label{f:PythonGenerator}
    675 
    676 \bigskip
    677 
    678 \begin{tabular}{@{}l|l@{}}
    679 \begin{cfa}[aboveskip=0pt,belowskip=0pt]
    680 enum Status { CONT, MSG, ESTX,
    681                                 ELNTH, ECRC };
    682 `generator` Driver {
    683         Status status;
    684         unsigned char byte, * msg; // communication
    685         unsigned int lnth, sum;      // local state
    686         unsigned short int crc;
    687 };
    688 void ?{}( Driver & d, char * m ) { d.msg = m; }
    689 Status next( Driver & d, char b ) with( d ) {
    690         byte = b; `resume( d );` return status;
    691 }
    692 void main( Driver & d ) with( d ) {
    693         enum { STX = '\002', ESC = '\033',
    694                         ETX = '\003', MaxMsg = 64 };
    695   msg: for () { // parse message
    696                 status = CONT;
    697                 lnth = 0; sum = 0;
    698                 while ( byte != STX ) `suspend;`
    699           emsg: for () {
    700                         `suspend;` // process byte
    701 \end{cfa}
    702 &
    703 \begin{cfa}[aboveskip=0pt,belowskip=0pt]
    704                         choose ( byte ) { // switch with implicit break
    705                           case STX:
    706                                 status = ESTX; `suspend;` continue msg;
    707                           case ETX:
    708                                 break emsg;
    709                           case ESC:
    710                                 `suspend;`
    711                         }
    712                         if ( lnth >= MaxMsg ) { // buffer full ?
    713                                 status = ELNTH; `suspend;` continue msg; }
    714                         msg[lnth++] = byte;
    715                         sum += byte;
    716                 }
    717                 msg[lnth] = '\0'; // terminate string
    718                 `suspend;`
    719                 crc = byte << 8;
    720                 `suspend;`
    721                 status = (crc | byte) == sum ? MSG : ECRC;
    722                 `suspend;`
    723         }
    724 }
    725 \end{cfa}
    726 \end{tabular}
    727 \caption{Device-driver generator for communication protocol}
    728 \label{f:DeviceDriverGen}
    729 \end{figure}
    730 
    731 Figure~\ref{f:CFAPingPongGen} shows a symmetric generator, where the generator resumes another generator, forming a resume/resume cycle.
    732 (The trivial cycle is a generator resuming itself.)
    733 This control flow is similar to recursion for functions, but without stack growth.
    734 The steps for symmetric control-flow are creating, executing, and terminating the cycle.
    735 Constructing the cycle must deal with definition-before-use to close the cycle, \ie, the first generator must know about the last generator, which is not within scope.
    736 (This issues occurs for any cyclic data-structure.)
    737 % The example creates all the generators and then assigns the partners that form the cycle.
    738 % Alternatively, the constructor can assign the partners as they are declared, except the first, and the first-generator partner is set after the last generator declaration to close the cycle.
    739 Once the cycle is formed, the program main resumes one of the generators, and the generators can then traverse an arbitrary cycle using @resume@ to activate partner generator(s).
    740 Terminating the cycle is accomplished by @suspend@ or @return@, both of which go back to the stack frame that started the cycle (program main in the example).
    741 The starting stack-frame is below the last active generator because the resume/resume cycle does not grow the stack.
    742 Also, since local variables are not retained in the generator function, it does not contain an objects with destructors that must be called, so the  cost is the same as a function return.
    743 Destructor cost occurs when the generator instance is deallocated, which is easily controlled by the programmer.
    744 
    745 Figure~\ref{f:CPingPongSim} shows the implementation of the symmetric generator, where the complexity is the @resume@, which needs an extension to the calling convention to perform a forward rather than backward jump.
    746 This jump starts at the top of the next generator main to re-execute the normal calling convention to make space on the stack for its local variables.
    747 However, before the jump, the caller must reset its stack (and any registers) equivalent to a @return@, but subsequently jump forward not backwards.
    748 This semantics is basically a tail-call optimization, which compilers already perform.
    749 Hence, assembly code is manually insert in the example to undo the generator's entry code before the direct jump.
    750 This assembly code is fragile as it depends on what entry code is generated, specifically if there are local variables and the level of optimization.
    751 To provide this new calling convention requires a mechanism built into the compiler, which is beyond the scope of \CFA at this time.
    752 Nevertheless, it is possible to hand generate any symmetric generators for proof of concept and performance testing.
    753 A compiler could also eliminate other artifacts in the generator simulation to further increase performance.
    754 
    755 \begin{figure}
    756 \centering
    757 \begin{lrbox}{\myboxA}
    758 \begin{cfa}[aboveskip=0pt,belowskip=0pt]
    759 `generator PingPong` {
    760         const char * name;
    761         PingPong & partner; // rebindable reference
    762         int N, i;
    763 };
    764 void ?{}(PingPong & pp, char * nm, int N) with(pp) {
    765         name = nm;  partner = 0p;  pp.N = N;  i = 0; }
    766 void `main( PingPong & pp )` with(pp) {
    767         for ( ; i < N; i += 1 ) {
    768                 sout | name | i;
    769                 `resume( partner );`
    770         }
    771 }
    772 int main() {
    773         enum { N = 5 };
    774         PingPong ping = {"ping", N}, pong = {"pong", N};
    775         &ping.partner = &pong;  &pong.partner = &ping;
    776         `resume( ping );`
    777 }
    778 \end{cfa}
    779 \end{lrbox}
    780 
    781 \begin{lrbox}{\myboxB}
    782 \begin{cfa}[escapechar={},aboveskip=0pt,belowskip=0pt]
    783 typedef struct PingPong {
    784         const char * name;
    785         struct PingPong * partner;
    786         int N, i;
    787         void * next;
    788 } PingPong;
    789 #define PPCtor(name, N) { name, NULL, N, 0, NULL }
    790 void comain( PingPong * pp ) {
    791         if ( pp->next ) goto *pp->next;
    792         pp->next = &&cycle;
    793         for ( ; pp->i < pp->N; pp->i += 1 ) {
    794                 printf( "%s %d\n", pp->name, pp->i );
    795                 asm( "mov  %0,%%rdi" : "=m" (pp->partner) );
    796                 asm( "mov  %rdi,%rax" );
    797                 asm( "popq %rbx" );
    798                 asm( "jmp  comain" );
    799           cycle: ;
    800         }
    801 }
    802 \end{cfa}
    803 \end{lrbox}
    804 
    805 \subfloat[\CFA symmetric generator]{\label{f:CFAPingPongGen}\usebox\myboxA}
    806 \hspace{3pt}
    807 \vrule
    808 \hspace{3pt}
    809 \subfloat[C generator simulation]{\label{f:CPingPongSim}\usebox\myboxB}
    810 \hspace{3pt}
    811 \caption{Ping-Pong Symmetric Generator}
    812 \label{f:PingPongSymmetricGenerator}
    813 \end{figure}
    814 
    815 Finally, part of this generator work was inspired by the recent \CCtwenty generator proposal~\cite{C++20Coroutine19} (which they call coroutines).
    816 Our work provides the same high-performance asymmetric-generators as \CCtwenty, and extends their work with symmetric generators.
    817 An additional \CCtwenty generator feature allows @suspend@ and @resume@ to be followed by a restricted compound-statement that is executed after the current generator has reset its stack but before calling the next generator, specified with the following \CFA syntax.
    818 \begin{cfa}
    819 ... suspend`{ ... }`;
    820 ... resume( C )`{ ... }` ...
    821 \end{cfa}
    822 Since the current generator's stack is released before calling the compound statement, the compound statement can only reference variables in the generator's type.
    823 This feature is useful when a generator is used in a concurrent context to ensure it is stopped before releasing a lock in the compound statement, which might immediately allow another thread to resume the generator.
    824 Hence, this mechanism provides a general and safe handoff of the generator among competing threads.
    825 
    826 
    827 \subsection{Coroutine}
    828 \label{s:Coroutine}
    829 
    830 Stackful coroutines extend generator semantics, \ie there is an implicit closure and @suspend@ may appear in a helper function called from the coroutine main.
    831 A coroutine is specified by replacing @generator@ with @coroutine@ for the type.
    832 This generality results in higher cost for creation, due to dynamic stack allocation, execution, due to context switching among stacks, and terminating, due to possible stack unwinding and dynamic stack deallocation.
    833 How coroutines differ from generators is done through a series of examples.
    834 
    835 First, the previous generator examples are converted to their coroutine counterparts, allowing local-state variables to be moved from the generator type into the coroutine main.
    836 \begin{description}
    837 \item[Fibonacci]
    838 Move the declaration of @fn1@ to the start of coroutine main.
    839 \begin{cfa}[xleftmargin=0pt]
    840 void main( Fib & fib ) with(fib) {
    841         `int fn1;`
    842 \end{cfa}
    843 \item[Formatter]
    844 Move the declaration of @g@ and @b@ to the for loops in the coroutine main.
    845 \begin{cfa}[xleftmargin=0pt]
    846 for ( `g`; 5 ) {
    847         for ( `b`; 4 ) {
    848 \end{cfa}
    849 \item[Device Driver]
    850 Move the declaration of @lnth@ and @sum@ to their points of initialization.
    851 \begin{cfa}[xleftmargin=0pt]
    852         status = CONT;
    853         `unsigned int lnth = 0, sum = 0;`
    854         ...
    855         `unsigned short int crc = byte << 8;`
    856 \end{cfa}
    857 \item[PingPong]
    858 Move the declaration of @i@ to the for loop in the coroutine main.
    859 \begin{cfa}[xleftmargin=0pt]
    860 void main( PingPong & pp ) with(pp) {
    861         for ( `i`; N ) {
    862 \end{cfa}
    863 \end{description}
    864 It is also possible to refactor code containing local-state and @suspend@ statements into a helper routine, like the computation of the CRC for the device driver.
    865 \begin{cfa}
    866 unsigned int Crc() {
    867         `suspend;`
    868         unsigned short int crc = byte << 8;
    869         `suspend;`
    870         status = (crc | byte) == sum ? MSG : ECRC;
    871         return crc;
    872 }
    873 \end{cfa}
    874 A call to this function is placed at the end of the driver's coroutine-main.
    875 For complex finite-state machines, refactoring is part of normal program abstraction, especially when code is used in multiple places.
    876 Again, this complexity is usually associated with execution state rather than data state.
    877 
    878 \begin{comment}
    879 Figure~\ref{f:Coroutine3States} creates a @coroutine@ type, @`coroutine` Fib { int fn; }@, which provides communication, @fn@, for the \newterm{coroutine main}, @main@, which runs on the coroutine stack, and possibly multiple interface routines, \eg @next@.
    880 Like the structure in Figure~\ref{f:ExternalState}, the coroutine type allows multiple instances, where instances of this type are passed to the (overloaded) coroutine main.
    881 The coroutine main's stack holds the state for the next generation, @f1@ and @f2@, and the code represents the three states in the Fibonacci formula via the three suspend points, to context switch back to the caller's @resume@.
    882 The interface routine @next@, takes a Fibonacci instance and context switches to it using @resume@;
    883 on restart, the Fibonacci field, @fn@, contains the next value in the sequence, which is returned.
    884 The first @resume@ is special because it allocates the coroutine stack and cocalls its coroutine main on that stack;
    885 when the coroutine main returns, its stack is deallocated.
    886 Hence, @Fib@ is an object at creation, transitions to a coroutine on its first resume, and transitions back to an object when the coroutine main finishes.
    887 Figure~\ref{f:Coroutine1State} shows the coroutine version of the C version in Figure~\ref{f:ExternalState}.
    888 Coroutine generators are called \newterm{output coroutines} because values are only returned.
     628forall( otype T `| sumable( T )` )                      $\C{// use trait}$
     629T sum( T a[$\,$], size_t size );
     630\end{cfa}
     631
     632Using the return type for overload discrimination, it is possible to write a type-safe @alloc@ based on the C @malloc@:
     633\begin{cfa}
     634forall( dtype T | sized(T) ) T * alloc( void ) { return (T *)malloc( sizeof(T) ); }
     635int * ip = alloc();                                                     $\C{// select type and size from left-hand side}$
     636double * dp = alloc();
     637struct S {...} * sp = alloc();
     638\end{cfa}
     639where the return type supplies the type/size of the allocation, which is impossible in most type systems.
     640\end{comment}
     641
     642
     643\section{Coroutines: Stepping Stone}
     644\label{coroutine}
     645
     646Coroutines are generalized routines allowing execution to be temporarily suspended and later resumed.
     647Hence, unlike a normal routine, a coroutine may not terminate when it returns to its caller, allowing it to be restarted with the values and execution location present at the point of suspension.
     648This capability is accomplished via the coroutine's stack, where suspend/resume context switch among stacks.
     649Because threading design-challenges are present in coroutines, their design effort is relevant, and this effort can be easily exposed to programmers giving them a useful new programming paradigm because a coroutine handles the class of problems that need to retain state between calls, \eg plugins, device drivers, and finite-state machines.
     650Therefore, the two fundamental features of the core \CFA coroutine-API are independent call-stacks and @suspend@/@resume@ operations.
     651
     652For example, a problem made easier with coroutines is unbounded generators, \eg generating an infinite sequence of Fibonacci numbers
     653\begin{displaymath}
     654\mathsf{fib}(n) = \left \{
     655\begin{array}{ll}
     6560                                       & n = 0         \\
     6571                                       & n = 1         \\
     658\mathsf{fib}(n-1) + \mathsf{fib}(n-2)   & n \ge 2       \\
     659\end{array}
     660\right.
     661\end{displaymath}
     662where Figure~\ref{f:C-fibonacci} shows conventional approaches for writing a Fibonacci generator in C.
     663Figure~\ref{f:GlobalVariables} illustrates the following problems: unique unencapsulated global variables necessary to retain state between calls, only one Fibonacci generator, and execution state must be explicitly retained via explicit state variables.
     664Figure~\ref{f:ExternalState} addresses these issues: unencapsulated program global variables become encapsulated structure variables, unique global variables are replaced by multiple Fibonacci objects, and explicit execution state is removed by precomputing the first two Fibonacci numbers and returning $\mathsf{fib}(n-2)$.
    889665
    890666\begin{figure}
    … …  
    913689\begin{lrbox}{\myboxA}
    914690\begin{cfa}[aboveskip=0pt,belowskip=0pt]
    915 #define FibCtor { 0, 1 }
     691#define FIB_INIT { 0, 1 }
    916692typedef struct { int fn1, fn; } Fib;
    917693int fib( Fib * f ) {
    … …  
    926702
    927703int main() {
    928         Fib f1 = FibCtor, f2 = FibCtor;
     704        Fib f1 = FIB_INIT, f2 = FIB_INIT;
    929705        for ( int i = 0; i < 10; i += 1 ) {
    930706                printf( "%d %d\n",
    … …  
    943719        [fn1, fn] = [0, 1];
    944720        for () {
    945                 `suspend;`
     721                `suspend();`
    946722                [fn1, fn] = [fn, fn1 + fn];
    947723        }
    948724}
    949725int ?()( Fib & fib ) with( fib ) {
    950         return `resume( fib )`.fn1;
     726        `resume( fib );`  return fn1;
    951727}
    952728int main() {
    … …  
    996772\caption{Fibonacci Generator}
    997773\label{f:C-fibonacci}
     774
     775% \bigskip
     776%
     777% \newbox\myboxA
     778% \begin{lrbox}{\myboxA}
     779% \begin{cfa}[aboveskip=0pt,belowskip=0pt]
     780% `coroutine` Fib { int fn; };
     781% void main( Fib & fib ) with( fib ) {
     782%       fn = 0;  int fn1 = fn; `suspend()`;
     783%       fn = 1;  int fn2 = fn1;  fn1 = fn; `suspend()`;
     784%       for () {
     785%               fn = fn1 + fn2; fn2 = fn1; fn1 = fn; `suspend()`; }
     786% }
     787% int next( Fib & fib ) with( fib ) { `resume( fib );` return fn; }
     788% int main() {
     789%       Fib f1, f2;
     790%       for ( 10 )
     791%               sout | next( f1 ) | next( f2 );
     792% }
     793% \end{cfa}
     794% \end{lrbox}
     795% \newbox\myboxB
     796% \begin{lrbox}{\myboxB}
     797% \begin{python}[aboveskip=0pt,belowskip=0pt]
     798%
     799% def Fibonacci():
     800%       fn = 0; fn1 = fn; `yield fn`  # suspend
     801%       fn = 1; fn2 = fn1; fn1 = fn; `yield fn`
     802%       while True:
     803%               fn = fn1 + fn2; fn2 = fn1; fn1 = fn; `yield fn`
     804%
     805%
     806% f1 = Fibonacci()
     807% f2 = Fibonacci()
     808% for i in range( 10 ):
     809%       print( `next( f1 )`, `next( f2 )` ) # resume
     810%
     811% \end{python}
     812% \end{lrbox}
     813% \subfloat[\CFA]{\label{f:Coroutine3States}\usebox\myboxA}
     814% \qquad
     815% \subfloat[Python]{\label{f:Coroutine1State}\usebox\myboxB}
     816% \caption{Fibonacci input coroutine, 3 states, internal variables}
     817% \label{f:cfa-fibonacci}
    998818\end{figure}
    999819
    1000 \bigskip
    1001 
     820Using a coroutine, it is possible to express the Fibonacci formula directly without any of the C problems.
     821Figure~\ref{f:Coroutine3States} creates a @coroutine@ type, @`coroutine` Fib { int fn; }@, which provides communication, @fn@, for the \newterm{coroutine main}, @main@, which runs on the coroutine stack, and possibly multiple interface routines, \eg @next@.
     822Like the structure in Figure~\ref{f:ExternalState}, the coroutine type allows multiple instances, where instances of this type are passed to the (overloaded) coroutine main.
     823The coroutine main's stack holds the state for the next generation, @f1@ and @f2@, and the code represents the three states in the Fibonacci formula via the three suspend points, to context switch back to the caller's @resume@.
     824The interface routine @next@, takes a Fibonacci instance and context switches to it using @resume@;
     825on restart, the Fibonacci field, @fn@, contains the next value in the sequence, which is returned.
     826The first @resume@ is special because it allocates the coroutine stack and cocalls its coroutine main on that stack;
     827when the coroutine main returns, its stack is deallocated.
     828Hence, @Fib@ is an object at creation, transitions to a coroutine on its first resume, and transitions back to an object when the coroutine main finishes.
     829Figure~\ref{f:Coroutine1State} shows the coroutine version of the C version in Figure~\ref{f:ExternalState}.
     830Coroutine generators are called \newterm{output coroutines} because values are only returned.
     831
     832Figure~\ref{f:CFAFmt} shows an \newterm{input coroutine}, @Format@, for restructuring text into groups of characters of fixed-size blocks.
     833For example, the input of the left is reformatted into the output on the right.
     834\begin{quote}
     835\tt
     836\begin{tabular}{@{}l|l@{}}
     837\multicolumn{1}{c|}{\textbf{\textrm{input}}} & \multicolumn{1}{c}{\textbf{\textrm{output}}} \\
     838abcdefghijklmnopqrstuvwxyzabcdefghijklmnopqrstuvwxyz
     839&
     840\begin{tabular}[t]{@{}lllll@{}}
     841abcd    & efgh  & ijkl  & mnop  & qrst  \\
     842uvwx    & yzab  & cdef  & ghij  & klmn  \\
     843opqr    & stuv  & wxyz  &               &
     844\end{tabular}
     845\end{tabular}
     846\end{quote}
     847The example takes advantage of resuming a coroutine in the constructor to prime the loops so the first character sent for formatting appears inside the nested loops.
     848The destructor provides a newline, if formatted text ends with a full line.
     849Figure~\ref{f:CFmt} shows the C equivalent formatter, where the loops of the coroutine are flattened (linearized) and rechecked on each call because execution location is not retained between calls.
     850(Linearized code is the bane of device drivers.)
     851
     852\begin{figure}
     853\centering
    1002854\newbox\myboxA
    1003855\begin{lrbox}{\myboxA}
    1004856\begin{cfa}[aboveskip=0pt,belowskip=0pt]
    1005 `coroutine` Fib { int fn; };
    1006 void main( Fib & fib ) with( fib ) {
    1007         fn = 0;  int fn1 = fn; `suspend`;
    1008         fn = 1;  int fn2 = fn1;  fn1 = fn; `suspend`;
     857`coroutine` Fmt {
     858        char ch;   // communication variables
     859        int g, b;   // needed in destructor
     860};
     861void main( Fmt & fmt ) with( fmt ) {
    1009862        for () {
    1010                 fn = fn1 + fn2; fn2 = fn1; fn1 = fn; `suspend`; }
    1011 }
    1012 int next( Fib & fib ) with( fib ) { `resume( fib );` return fn; }
     863                for ( g = 0; g < 5; g += 1 ) { // groups
     864                        for ( b = 0; b < 4; b += 1 ) { // blocks
     865                                `suspend();`
     866                                sout | ch; } // print character
     867                        sout | "  "; } // block separator
     868                sout | nl; }  // group separator
     869}
     870void ?{}( Fmt & fmt ) { `resume( fmt );` } // prime
     871void ^?{}( Fmt & fmt ) with( fmt ) { // destructor
     872        if ( g != 0 || b != 0 ) // special case
     873                sout | nl; }
     874void send( Fmt & fmt, char c ) { fmt.ch = c; `resume( fmt )`; }
    1013875int main() {
    1014         Fib f1, f2;
    1015         for ( 10 )
    1016                 sout | next( f1 ) | next( f2 );
     876        Fmt fmt;
     877        sout | nlOff;   // turn off auto newline
     878        for ( 41 )
     879                send( fmt, 'a' );
    1017880}
    1018881\end{cfa}
    1019882\end{lrbox}
     883
    1020884\newbox\myboxB
    1021885\begin{lrbox}{\myboxB}
    1022886\begin{python}[aboveskip=0pt,belowskip=0pt]
    1023887
    1024 def Fibonacci():
    1025         fn = 0; fn1 = fn; `yield fn`  # suspend
    1026         fn = 1; fn2 = fn1; fn1 = fn; `yield fn`
    1027         while True:
    1028                 fn = fn1 + fn2; fn2 = fn1; fn1 = fn; `yield fn`
    1029 
    1030 
    1031 f1 = Fibonacci()
    1032 f2 = Fibonacci()
    1033 for i in range( 10 ):
    1034         print( `next( f1 )`, `next( f2 )` ) # resume
     888
     889
     890def Fmt():
     891        try:
     892                while True:
     893                        for g in range( 5 ):
     894                                for b in range( 4 ):
     895
     896                                        print( `(yield)`, end='' )
     897                                print( '  ', end='' )
     898                        print()
     899
     900
     901        except GeneratorExit:
     902                if g != 0 | b != 0:
     903                        print()
     904
     905
     906fmt = Fmt()
     907`next( fmt )`                    # prime
     908for i in range( 41 ):
     909        `fmt.send( 'a' );`      # send to yield
    1035910
    1036911\end{python}
    1037912\end{lrbox}
    1038 \subfloat[\CFA]{\label{f:Coroutine3States}\usebox\myboxA}
     913\subfloat[\CFA]{\label{f:CFAFmt}\usebox\myboxA}
    1039914\qquad
    1040 \subfloat[Python]{\label{f:Coroutine1State}\usebox\myboxB}
    1041 \caption{Fibonacci input coroutine, 3 states, internal variables}
    1042 \label{f:cfa-fibonacci}
     915\subfloat[Python]{\label{f:CFmt}\usebox\myboxB}
     916\caption{Output formatting text}
     917\label{f:fmt-line}
    1043918\end{figure}
    1044 \end{comment}
     919
     920The previous examples are \newterm{asymmetric (semi) coroutine}s because one coroutine always calls a resuming routine for another coroutine, and the resumed coroutine always suspends back to its last resumer, similar to call/return for normal routines.
     921However, @resume@ and @suspend@ context switch among existing stack-frames, rather than create new ones so there is no stack growth.
     922\newterm{Symmetric (full) coroutine}s have a coroutine call to a resuming routine for another coroutine, and its coroutine main calls another resuming routine, which eventually forms a resuming-call cycle.
     923(The trivial cycle is a coroutine resuming itself.)
     924This control flow is similar to recursion for normal routines, but again there is no stack growth from the context switch.
    1045925
    1046926\begin{figure}
    … …  
    1050930\begin{cfa}
    1051931`coroutine` Prod {
    1052         Cons & c;                       // communication
     932        Cons & c;
    1053933        int N, money, receipt;
    1054934};
    … …  
    1071951}
    1072952void start( Prod & prod, int N, Cons &c ) {
    1073         &prod.c = &c;
     953        &prod.c = &c; // reassignable reference
    1074954        prod.[N, receipt] = [N, 0];
    1075955        `resume( prod );`
    … …  
    1084964\begin{cfa}
    1085965`coroutine` Cons {
    1086         Prod & p;                       // communication
     966        Prod & p;
    1087967        int p1, p2, status;
    1088968        bool done;
    … …  
    1092972        cons.[status, done ] = [0, false];
    1093973}
     974void ^?{}( Cons & cons ) {}
    1094975void main( Cons & cons ) with( cons ) {
    1095976        // 1st resume starts here
    … …  
    1113994        `resume( cons );`
    1114995}
    1115 
    1116996\end{cfa}
    1117997\end{tabular}
    1118998\caption{Producer / consumer: resume-resume cycle, bi-directional communication}
    1119999\label{f:ProdCons}
    1120 
    1121 \medskip
    1122 
    1123 \begin{center}
    1124 \input{FullProdConsStack.pstex_t}
    1125 \end{center}
    1126 \vspace*{-10pt}
    1127 \caption{Producer / consumer runtime stacks}
    1128 \label{f:ProdConsRuntimeStacks}
    1129 
    1130 \medskip
    1131 
    1132 \begin{center}
    1133 \input{FullCoroutinePhases.pstex_t}
    1134 \end{center}
    1135 \vspace*{-10pt}
    1136 \caption{Ping / Pong coroutine steps}
    1137 \label{f:PingPongFullCoroutineSteps}
    11381000\end{figure}
    11391001
    1140 Figure~\ref{f:ProdCons} shows the ping-pong example in Figure~\ref{f:CFAPingPongGen} extended into a producer/consumer symmetric-coroutine performing bidirectional communication.
    1141 This example is illustrative because both producer/consumer have two interface functions with @resume@s that suspend execution in these interface (helper) functions.
    1142 The program main creates the producer coroutine, passes it to the consumer coroutine in its initialization, and closes the cycle at the call to @start@ along with the number of items to be produced.
    1143 The first @resume@ of @prod@ creates @prod@'s stack with a frame for @prod@'s coroutine main at the top, and context switches to it.
    1144 @prod@'s coroutine main starts, creates local-state variables that are retained between coroutine activations, and executes $N$ iterations, each generating two random values, calling the consumer to deliver the values, and printing the status returned from the consumer.
     1002Figure~\ref{f:ProdCons} shows a producer/consumer symmetric-coroutine performing bi-directional communication.
     1003Since the solution involves a full-coroutining cycle, the program main creates one coroutine in isolation, passes this coroutine to its partner, and closes the cycle at the call to @start@.
     1004The @start@ routine communicates both the number of elements to be produced and the consumer into the producer's coroutine-structure.
     1005Then the @resume@ to @prod@ creates @prod@'s stack with a frame for @prod@'s coroutine main at the top, and context switches to it.
     1006@prod@'s coroutine main starts, creates local variables that are retained between coroutine activations, and executes $N$ iterations, each generating two random values, calling the consumer to deliver the values, and printing the status returned from the consumer.
    11451007
    11461008The producer call to @delivery@ transfers values into the consumer's communication variables, resumes the consumer, and returns the consumer status.
    1147 On the first resume, @cons@'s stack is created and initialized, holding local-state variables retained between subsequent activations of the coroutine.
     1009For the first resume, @cons@'s stack is initialized, creating local variables retained between subsequent activations of the coroutine.
    11481010The consumer iterates until the @done@ flag is set, prints the values delivered by the producer, increments status, and calls back to the producer via @payment@, and on return from @payment@, prints the receipt from the producer and increments @money@ (inflation).
    11491011The call from the consumer to @payment@ introduces the cycle between producer and consumer.
    11501012When @payment@ is called, the consumer copies values into the producer's communication variable and a resume is executed.
    11511013The context switch restarts the producer at the point where it last context switched, so it continues in @delivery@ after the resume.
     1014
    11521015@delivery@ returns the status value in @prod@'s coroutine main, where the status is printed.
    11531016The loop then repeats calling @delivery@, where each call resumes the consumer coroutine.
    … …  
    11551018The consumer increments and returns the receipt to the call in @cons@'s coroutine main.
    11561019The loop then repeats calling @payment@, where each call resumes the producer coroutine.
    1157 Figure~\ref{f:ProdConsRuntimeStacks} shows the runtime stacks of the program main, and the coroutine mains for @prod@ and @cons@ during the cycling.
    1158 
    1159 Terminating a coroutine cycle is more complex than a generator cycle, because it requires context switching to the program main's \emph{stack} to shutdown the program, whereas generators started by the program main run on its stack.
    1160 Furthermore, each deallocated coroutine must guarantee all destructors are run for object allocated in the coroutine type \emph{and} allocated on the coroutine's stack at the point of suspension, which can be arbitrarily deep.
    1161 When a coroutine's main ends, its stack is already unwound so any stack allocated objects with destructors have been finalized.
    1162 The na\"{i}ve semantics for coroutine-cycle termination is context switch to the last resumer, like executing a @suspend@/@return@ in a generator.
    1163 However, for coroutines, the last resumer is \emph{not} implicitly below the current stack frame, as for generators, because each coroutine's stack is independent.
    1164 Unfortunately, it is impossible to determine statically if a coroutine is in a cycle and unrealistic to check dynamically (graph-cycle problem).
    1165 Hence, a compromise solution is necessary that works for asymmetric (acyclic) and symmetric (cyclic) coroutines.
    1166 
    1167 Our solution for coroutine termination works well for the most common asymmetric and symmetric coroutine usage-patterns.
    1168 For asymmetric coroutines, it is common for the first resumer (starter) coroutine to be the only resumer.
    1169 All previous generators converted to coroutines have this property.
    1170 For symmetric coroutines, it is common for the cycle creator to persist for the life-time of the cycle.
    1171 Hence, the starter coroutine is remembered on the first resume and ending the coroutine resumes the starter.
    1172 Figure~\ref{f:ProdConsRuntimeStacks} shows this semantic by the dashed lines from the end of the coroutine mains: @prod@ starts @cons@ so @cons@ resumes @prod@ at the end, and the program main starts @prod@ so @prod@ resumes the program main at the end.
    1173 For other scenarios, it is always possible to devise a solution with additional programming effort.
    1174 
    1175 The producer/consumer example does not illustrate the full power of the starter semantics because @cons@ always ends first.
    1176 Assume generator @PingPong@ is converted to a coroutine.
    1177 Figure~\ref{f:PingPongFullCoroutineSteps} shows the creation, starter, and cyclic execution steps of the coroutine version.
    1178 The program main creates (declares) coroutine instances @ping@ and @pong@.
    1179 Next, program main resumes @ping@, making it @ping@'s starter, and @ping@'s main resumes @pong@'s main, making it @pong@'s starter.
    1180 Execution forms a cycle when @pong@ resumes @ping@, and cycles $N$ times.
    1181 By adjusting $N$ for either @ping@/@pong@, it is possible to have either one finish first, instead of @pong@ always ending first.
    1182 If @pong@ ends first, it resumes its starter @ping@ in its coroutine main, then @ping@ ends and resumes its starter the program main in function @start@.
    1183 If @ping@ ends first, it resumes its starter the program main in function @start@.
    1184 Regardless of the cycle complexity, the starter stack always leads back to the program main, but the stack can be entered at an arbitrary point.
    1185 Once back at the program main, coroutines @ping@ and @pong@ are deallocated.
    1186 For generators, deallocation runs the destructors for all objects in the generator type.
    1187 For coroutines, deallocation deals with objects in the coroutine type and must also run the destructors for any objects pending on the coroutine's stack for any unterminated coroutine.
    1188 Hence, if a coroutine's destructor detects the coroutine is not ended, it implicitly raises a cancellation exception (uncatchable exception) at the coroutine and resumes it so the cancellation exception can propagate to the root of the coroutine's stack destroying all local variable on the stack.
    1189 So the \CFA semantics for the generator and coroutine, ensure both can be safely deallocated at any time, regardless of their current state, like any other aggregate object.
    1190 Explicitly raising normal exceptions at another coroutine can replace flag variables, like @stop@, \eg @prod@ raises a @stop@ exception at @cons@ after it finishes generating values and resumes @cons@, which catches the @stop@exception to terminate its loop.
    1191 
    1192 Finally, there is an interesting effect for @suspend@ with symmetric coroutines.
    1193 A coroutine must retain its last resumer to suspend back because the resumer is on a different stack.
    1194 These reverse pointers allow @suspend@ to cycle \emph{backwards}, which may be useful in certain cases.
    1195 However, there is an anomaly if a coroutine resumes itself, because it overwrites its last resumer with itself, losing the ability to resume the last external resumer.
    1196 To prevent losing this information, a self-resume does not overwrite the last resumer.
     1020
     1021After iterating $N$ times, the producer calls @stop@.
     1022The @done@ flag is set to stop the consumer's execution and a resume is executed.
     1023The context switch restarts @cons@ in @payment@ and it returns with the last receipt.
     1024The consumer terminates its loops because @done@ is true, its @main@ terminates, so @cons@ transitions from a coroutine back to an object, and @prod@ reactivates after the resume in @stop@.
     1025@stop@ returns and @prod@'s coroutine main terminates.
     1026The program main restarts after the resume in @start@.
     1027@start@ returns and the program main terminates.
     1028
     1029One \emph{killer} application for a coroutine is device drivers, which at one time caused 70\%-85\% of failures in Windows/Linux~\cite{Swift05}.
     1030Many device drivers are a finite state-machine parsing a protocol, e.g.:
     1031\begin{tabbing}
     1032\ldots STX \= \ldots message \ldots \= ESC \= ETX \= \ldots message \ldots  \= ETX \= 2-byte crc \= \ldots      \kill
     1033\ldots STX \> \ldots message \ldots \> ESC \> ETX \> \ldots message \ldots  \> ETX \> 2-byte crc \> \ldots
     1034\end{tabbing}
     1035where a network message begins with the control character STX and ends with an ETX, followed by a 2-byte cyclic-redundancy check.
     1036Control characters may appear in a message if preceded by an ESC.
     1037Because FSMs can be complex and occur frequently in important domains, direct support of the coroutine is crucial in a systems programminglanguage.
     1038
     1039\begin{figure}
     1040\begin{cfa}
     1041enum Status { CONT, MSG, ESTX, ELNTH, ECRC };
     1042`coroutine` Driver {
     1043        Status status;
     1044        char * msg, byte;
     1045};
     1046void ?{}( Driver & d, char * m ) { d.msg = m; }         $\C[3.0in]{// constructor}$
     1047Status next( Driver & d, char b ) with( d ) {           $\C{// 'with' opens scope}$
     1048        byte = b; `resume( d );` return status;
     1049}
     1050void main( Driver & d ) with( d ) {
     1051        enum { STX = '\002', ESC = '\033', ETX = '\003', MaxMsg = 64 };
     1052        unsigned short int crc;                                                 $\C{// error checking}$
     1053  msg: for () {                                                                         $\C{// parse message}$
     1054                status = CONT;
     1055                unsigned int lnth = 0, sum = 0;
     1056                while ( byte != STX ) `suspend();`
     1057          emsg: for () {
     1058                        `suspend();`                                                    $\C{// process byte}$
     1059                        choose ( byte ) {                                               $\C{// switch with default break}$
     1060                          case STX:
     1061                                status = ESTX; `suspend();` continue msg;
     1062                          case ETX:
     1063                                break emsg;
     1064                          case ESC:
     1065                                suspend();
     1066                        } // choose
     1067                        if ( lnth >= MaxMsg ) {                                 $\C{// buffer full ?}$
     1068                                status = ELNTH; `suspend();` continue msg; }
     1069                        msg[lnth++] = byte;
     1070                        sum += byte;
     1071                } // for
     1072                msg[lnth] = '\0';                                                       $\C{// terminate string}\CRT$
     1073                `suspend();`
     1074                crc = (unsigned char)byte << 8; // prevent sign extension for signed char
     1075                `suspend();`
     1076                status = (crc | (unsigned char)byte) == sum ? MSG : ECRC;
     1077                `suspend();`
     1078        } // for
     1079}
     1080\end{cfa}
     1081\caption{Device driver for simple communication protocol}
     1082\end{figure}
    11971083
    11981084
    … …  
    12001086
    12011087A significant implementation challenge for coroutines (and threads, see Section~\ref{threads}) is adding extra fields and executing code after/before the coroutine constructor/destructor and coroutine main to create/initialize/de-initialize/destroy extra fields and the stack.
    1202 There are several solutions to this problem and the chosen option directed the \CFA coroutine design.
    1203 
    1204 For object-oriented languages, inheritance can be used to provide extra fields and code, but it requires explicitly writing the inheritance:
     1088There are several solutions to this problem and the chosen option forced the \CFA coroutine design.
     1089
     1090Object-oriented inheritance provides extra fields and code in a restricted context, but it requires programmers to explicitly perform the inheritance:
    12051091\begin{cfa}[morekeywords={class,inherits}]
    12061092class mycoroutine inherits baseCoroutine { ... }
    12071093\end{cfa}
    1208 In addition, the programming language and possibly its tool set, \eg debugger, @valgrind@ need to understand @baseCoroutine@ because of special stack property of coroutines.
     1094and the programming language (and possibly its tool set, \eg debugger) may need to understand @baseCoroutine@ because of the stack.
    12091095Furthermore, the execution of constructors/destructors is in the wrong order for certain operations.
    12101096For example, for threads if the thread is implicitly started, it must start \emph{after} all constructors, because the thread relies on a completely initialized object, but the inherited constructor runs \emph{before} the derived.
    … …  
    12171103}
    12181104\end{cfa}
    1219 which also requires an explicit declaration that must be last to ensure correct initialization order.
    1220 However, there is nothing preventing wrong placement or multiple declarations of @baseCoroutine@.
    1221 
    1222 For coroutines, as for threads, many implementations are based on function pointers or function objects~\cite{Butenhof97, C++14, MS:VisualC++, BoostCoroutines15}.
     1105which also requires an explicit declaration that must be the last one to ensure correct initialization order.
     1106However, there is nothing preventing wrong placement or multiple declarations.
     1107
     1108For coroutines as for threads, many implementations are based on routine pointers or routine objects~\cite{Butenhof97, C++14, MS:VisualC++, BoostCoroutines15}.
    12231109For example, Boost implements coroutines in terms of four functor object-types:
    12241110\begin{cfa}
    … …  
    12281114symmetric_coroutine<>::yield_type
    12291115\end{cfa}
    1230 
    1231 Figure~\ref{f:CoroutineMemoryLayout} shows different memory-layout options for a coroutine (where a task is similar).
    1232 The coroutine handle is the @coroutine@ instance containing programmer specified global/communication variables.
    1233 The coroutine descriptor contains all implicit declarations needed by the runtime, \eg @suspend@/@resume@, and can be part of the coroutine handle or separate.\footnote{
    1234 We are examining variable-sized structures (VLS), where fields can be variable-sized structures or arrays.
    1235 Once allocated, a VLS is fixed sized.}
    1236 The coroutine stack can appear in a number of locations and forms, fixed or variable sized.
    1237 Hence, the coroutine's stack could be a VLS on the allocating stack, provided the allocating stack is large enough.
    1238 For stack allocation, allocation/deallocation only requires a few arithmetic operation to compute the size and adjust the stack point, modulo any constructor costs.
    1239 For heap allocation, allocation/deallocation is an expensive heap operation (where the heap can be a shared resource), modulo any constructor costs.
    1240 For heap stack allocation, it is also possible to use a split (segmented) stack calling-convention, available with gcc and clang, so the stack is variable sized.
    1241 Currently, \CFA supports stack/heap allocated descriptors but only heap allocated stacks;
    1242 split-stack allocation is under development.
    1243 In \CFA debug-mode, a fixed-sized stack is terminated with a write-only page, which catches most stack overflows.
    1244 
    1245 \begin{figure}
    1246 \centering
    1247 \input{corlayout.pstex_t}
    1248 \caption{Coroutine memory layout}
    1249 \label{f:CoroutineMemoryLayout}
    1250 \end{figure}
    1251 
    1252 Similarly, the canonical threading paradigm is often based on function pointers, \eg @pthreads@~\cite{Butenhof97}, \Csharp~\cite{Csharp}, Go~\cite{Go}, and Scala~\cite{Scala}.
    1253 However, the generic thread-handle (identifier) is limited (few operations), unless it is wrapped in a custom type, as in the pthreads approach.
     1116Similarly, the canonical threading paradigm is often based on routine pointers, \eg @pthreads@~\cite{Butenhof97}, \Csharp~\cite{Csharp}, Go~\cite{Go}, and Scala~\cite{Scala}.
     1117However, the generic thread-handle (identifier) is limited (few operations), unless it is wrapped in a custom type.
    12541118\begin{cfa}
    12551119void mycor( coroutine_t cid, void * arg ) {
    … …  
    12631127}
    12641128\end{cfa}
    1265 Since the custom type is simple to write in \CFA and solves several issues, added support for function/lambda-based coroutines adds very little.
     1129Since the custom type is simple to write in \CFA and solves several issues, added support for routine/lambda-based coroutines adds very little.
    12661130
    12671131Note, the type @coroutine_t@ must be an abstract handle to the coroutine, because the coroutine descriptor and its stack are non-copyable.

  • doc/papers/concurrency/examples/Fib.c

    r46438e4 r05d55ff  
    11#include <stdio.h>
    22
    3 typedef struct {
    4         int fn1, fn;
    5 } Fib;
     3#define FIB_INIT { 0, 1 }
     4typedef struct { int fn1, fn; } Fib;
     5int fib( Fib * f ) {
    66
    7 #define FibCtor { 1, 0 }
     7        int ret = f->fn1;
     8        f->fn1 = f->fn;
     9        f->fn = ret + f->fn;
    810
    9 int fib( Fib * f ) {
    10         int fn = f->fn; f->fn = f->fn1;
    11                 f->fn1 = f->fn + fn;
    12         return fn;
     11        return ret;
    1312}
    1413int main() {
    15         Fib f1 = FibCtor, f2 = FibCtor;
     14        Fib f1 = FIB_INIT, f2 = FIB_INIT;
    1615        for ( int i = 0; i < 10; i += 1 ) {
    1716                printf( "%d %d\n", fib( &f1 ), fib( &f2 ) );

  • doc/papers/concurrency/examples/Fib.cfa

    r46438e4 r05d55ff  
    1313}
    1414
    15 #define FibCtor { 0, 1 }
    16 typedef struct { int fn, fn1; } Fib;
     15#define FIB_INIT { 0, 1 }
     16typedef struct { int fn1, fn2; } Fib;
    1717int fib_state( Fib & f ) with( f ) {
    18         int fn0 = fn1 + fn2;  fn2 = fn1;  fn = fn0;
    19         return fn1;
     18        int ret = fn2;
     19        int fn = fn1 + fn2;
     20        fn2 = fn1; fn1 = fn;
     21        return ret;
    2022}
    2123
    … …  
    2830        }
    2931}
    30 int ?()( Fib1 & fib ) with( fib ) { return resume( fib ).fn; }
     32int next( Fib1 & fib ) with( fib ) { resume( fib ); return fn; }
    3133
    32 coroutine Fib2 { int fn; };                                             // used for communication
     34coroutine Fib2 { int ret; };                                    // used for communication
    3335void main( Fib2 & fib ) with( fib ) {                   // called on first resume
    34         int fn1 = 1, fn2 = 0;                                           // precompute first two states
     36        int fn, fn1 = 1, fn2 = 0;                                       // precompute first two states
    3537        for () {
     38                ret = fn2;
    3639                fn = fn1 + fn2;  fn2 = fn1;  fn1 = fn;  // general case
    3740                suspend();                                                              // restart last resume
    3841        }
    3942}
    40 int ?()( Fib2 & fib ) with( fib ) {
    41         return resume( fib ).fn;                                        // restart last suspend
    42 }
    43 int ?()( Fib2 & fib, int N ) with( fib ) {
    44         for ( N - 1 ) fib();
    45         return fib();
    46 }
    47 double ?()( Fib2 & fib ) with( fib ) {
    48         return (int)(fib()) / 3.14159;                                          // restart last suspend
     43int next( Fib2 & fib ) with( fib ) {
     44        resume( fib );                                                          // restart last suspend
     45        return ret;
    4946}
    5047
    … …  
    5451        sout | nl;
    5552
    56         Fib f1 = FibCtor, f2 = FibCtor;
     53        Fib f1 = FIB_INIT, f2 = FIB_INIT;
    5754        for ( 10 )
    5855                sout | fib_state( f1 ) | fib_state( f2 );
    … …  
    6158        Fib1 f1, f2;
    6259        for ( 10 )
    63                 sout | f1() | f2();
     60                sout | next( f1 ) | next( f2 );
    6461        sout | nl;
    6562
    66         Fib2 f12, f22;
     63        Fib2 f1, f2;
    6764        for ( 10 )
    68                 sout | (int)f12() | (double)f12() | f22( 2 );
     65                sout | next( (Fib2 &)f1 ) | next( (Fib2 &)f2 );
    6966}
    7067

  • doc/papers/concurrency/examples/Fib2.cfa

    r46438e4 r05d55ff  
    1010// Created On       : Thu Apr 26 23:20:08 2018
    1111// Last Modified By : Peter A. Buhr
    12 // Last Modified On : Sat May 18 08:55:59 2019
    13 // Update Count     : 36
     12// Last Modified On : Fri Mar 22 17:26:41 2019
     13// Update Count     : 28
    1414//
    1515
    … …  
    1717#include <coroutine.hfa>
    1818
    19 coroutine Fibonacci { int fn1, fn; };                                   // used for communication
     19coroutine Fibonacci { int fn1; };                                               // used for communication
    2020
    2121void main( Fibonacci & fib ) with( fib ) {                              // called on first resume
    22         [fn1, fn] = [1, 0];                                                                     // precompute first two states
     22        int fn;
     23        [fn1, fn] = [0, 1];                                                                     // precompute first two states
    2324        for () {
    2425                suspend();                                                                              // restart last resume
    … …  
    2829
    2930int ?()( Fibonacci & fib ) with( fib ) {                                // function call operator
    30         return resume( fib ).fn1;                                                       // restart last suspend
     31        resume( fib );                                                                          // restart last suspend
     32        return fn1;
    3133}
    3234
    … …  
    3436        Fibonacci f1, f2;
    3537        for ( 10 ) {                                                                            // print N Fibonacci values
    36                 sout | resume( f1 ).fn | resume( f2 ).fn;
     38                sout | f1() | f2();
    3739        } // for
    3840}
    … …  
    4042// Local Variables: //
    4143// tab-width: 4 //
    42 // compile-command: "cfa Fib2.cfa" //
     44// compile-command: "cfa fibonacci_1.cfa" //
    4345// End: //

  • libcfa/src/iostream.cfa

    r46438e4 r05d55ff  
    1010// Created On       : Wed May 27 17:56:53 2015
    1111// Last Modified By : Peter A. Buhr
    12 // Last Modified On : Tue Jun  4 17:32:34 2019
    13 // Update Count     : 761
     12// Last Modified On : Tue May 21 13:01:26 2019
     13// Update Count     : 674
    1414//
    1515
    … …  
    2020#include <stdbool.h>                                                                    // true/false
    2121//#include <string.h>                                                                   // strlen, strcmp
     22extern int strcmp (const char *__s1, const char *__s2) __attribute__ ((__nothrow__ , __leaf__)) __attribute__ ((__pure__)) __attribute__ ((__nonnull__ (1, 2)));
    2223extern size_t strlen (const char *__s) __attribute__ ((__nothrow__ , __leaf__)) __attribute__ ((__pure__)) __attribute__ ((__nonnull__ (1)));
    23 extern int strcmp (const char *__s1, const char *__s2) __attribute__ ((__nothrow__ , __leaf__)) __attribute__ ((__pure__)) __attribute__ ((__nonnull__ (1, 2)));
    24 extern char *strcpy (char *__restrict __dest, const char *__restrict __src) __attribute__ ((__nothrow__ , __leaf__)) __attribute__ ((__nonnull__ (1, 2)));
    25 extern void *memcpy (void *__restrict __dest, const void *__restrict __src, size_t __n) __attribute__ ((__nothrow__ , __leaf__)) __attribute__ ((__nonnull__ (1, 2)));
    2624#include <float.h>                                                                              // DBL_DIG, LDBL_DIG
    2725#include <math.h>                                                                               // isfinite
    2826#include <complex.h>                                                                    // creal, cimag
    29 } // extern "C"
     27}
    3028
    3129forall( dtype ostype | ostream( ostype ) ) {
    … …  
    200198                if ( sepPrt( os ) ) fmt( os, "%s", sepGetCur( os ) );
    201199//              os | crealf( fc ) | nonl;
    202                 PrintWithDP( os, "%g", crealf( fc ) );
    203                 PrintWithDP( os, "%+g", cimagf( fc ) );
     200                float f = crealf( fc );
     201                PrintWithDP( os, "%g", f );
     202                f = cimagf( fc );
     203                PrintWithDP( os, "%+g", f );
    204204                fmt( os, "i" );
    205205                return os;
    … …  
    212212                if ( sepPrt( os ) ) fmt( os, "%s", sepGetCur( os ) );
    213213//              os | creal( dc ) | nonl;
    214                 PrintWithDP( os, "%.*lg", creal( dc ), DBL_DIG );
    215                 PrintWithDP( os, "%+.*lg", cimag( dc ), DBL_DIG );
     214                double d = creal( dc );
     215                PrintWithDP( os, "%.*lg", d, DBL_DIG );
     216                d = cimag( dc );
     217                PrintWithDP( os, "%+.*lg", d, DBL_DIG );
    216218                fmt( os, "i" );
    217219                return os;
    … …  
    224226                if ( sepPrt( os ) ) fmt( os, "%s", sepGetCur( os ) );
    225227//              os | creall( ldc ) || nonl;
    226                 PrintWithDP( os, "%.*Lg", creall( ldc ), LDBL_DIG );
    227                 PrintWithDP( os, "%+.*Lg", cimagl( ldc ), LDBL_DIG );
     228                long double ld = creall( ldc );
     229                PrintWithDP( os, "%.*Lg", ld, LDBL_DIG );
     230                ld = cimagl( ldc );
     231                PrintWithDP( os, "%+.*Lg", ld, LDBL_DIG );
    228232                fmt( os, "i" );
    229233                return os;
    … …  
    391395} // distribution
    392396
    393 
    394 //*********************************** Integral ***********************************
    395 
    396 static const char * shortbin[] = { "0", "1", "10", "11", "100", "101", "110", "111", "1000", "1001", "1010", "1011", "1100", "1101", "1110", "1111" };
    397 static const char * longbin[]  = { "0000", "0001", "0010", "0011", "0100", "0101", "0110", "0111", "1000", "1001", "1010", "1011", "1100", "1101", "1110", "1111" };
    398 
    399 // Default prefix for non-decimal prints is 0b, 0, 0x.
    400 #define IntegralFMTImpl( T, CODE, IFMTNP, IFMTP ) \
    401 forall( dtype ostype | ostream( ostype ) ) { \
    402         ostype & ?|?( ostype & os, _Ostream_Manip(T) f ) { \
    403                 if ( sepPrt( os ) ) fmt( os, "%s", sepGetCur( os ) ); \
    404 \
    405                 if ( f.base == 'b' || f.base == 'B' ) {                 /* bespoke binary format */ \
    406                         int bits;                                                                                                       \
    407                         if ( f.val == (T){0} ) bits = 1;                        /* force at least one bit to print */ \
    408                         else bits = sizeof(long long int) * 8 - __builtin_clzll( f.val ); /* position of most significant bit */ \
    409                         bits = bits > sizeof(f.val) * 8 ? sizeof(f.val) * 8 : bits; \
    410                         int spaces = f.wd - bits;                                       /* can be negative */ \
    411                         if ( ! f.flags.nobsdp ) { spaces -= 2; }        /* base prefix takes space */ \
    412                         /* printf( "%d %d\n", bits, spaces ); */ \
    413                         if ( ! f.flags.left ) {                                         /* right justified ? */ \
    414                                 /* Note, base prefix then zero padding or spacing then prefix. */ \
    415                                 if ( f.flags.pad0 || f.flags.pc ) { \
    416                                         if ( ! f.flags.nobsdp ) { fmt( os, "0%c", f.base ); } \
    417                                         if ( f.flags.pc ) spaces = f.pc - bits; \
    418                                         if ( spaces > 0 ) fmt( os, "%0*d", spaces, 0 ); /* zero pad */ \
    419                                 } else { \
    420                                         if ( spaces > 0 ) fmt( os, "%*s", spaces, " " ); /* space pad */ \
    421                                         if ( ! f.flags.nobsdp ) { fmt( os, "0%c", f.base ); } \
    422                                 } /* if */ \
    423                         } else if ( ! f.flags.nobsdp ) { \
    424                                 fmt( os, "0%c", f.base ); \
    425                         } /* if */ \
    426                         int shift = (bits - 1) / 4 * 4; /* floor( bits - 1, 4 ) */ \
    427                         typeof( f.val ) temp = f.val; \
    428                         fmt( os, "%s", shortbin[(temp >> shift) & 0xf] ); \
    429                         for () { \
    430                                 shift -= 4; \
    431                           if ( shift < 0 ) break; \
    432                                 temp = f.val; \
    433                                 fmt( os, "%s", longbin[(temp >> shift) & 0xf] ); \
    434                         } /* for */ \
    435                         if ( f.flags.left && spaces > 0 ) fmt( os, "%*s", spaces, " " ); \
    436                         return os; \
    437                 } /* if  */ \
    438 \
    439                 char fmtstr[sizeof(IFMTP)];                                             /* sizeof includes '\0' */ \
    440                 if ( ! f.flags.pc ) memcpy( &fmtstr, IFMTNP, sizeof(IFMTNP) ); \
    441                 else memcpy( &fmtstr, IFMTP, sizeof(IFMTP) ); \
    442                 int star = 4;                                                                   /* position before first '*' */ \
    443 \
    444                 /* Insert flags into spaces before '*', from right to left. */ \
    445                 if ( ! f.flags.nobsdp ) { fmtstr[star] = '#'; star -= 1; } \
    446                 if ( f.flags.left ) { fmtstr[star] = '-'; star -= 1; } \
    447                 if ( f.flags.sign && f.base == CODE ) { fmtstr[star] = '+'; star -= 1; } \
    448                 if ( f.flags.pad0 && ! f.flags.pc ) { fmtstr[star] = '0'; star -= 1; } \
    449                 fmtstr[star] = '%'; \
    450 \
    451                 if ( ! f.flags.pc ) {                                                   /* no precision */ \
    452                         /* printf( "%s\n", &fmtstr[star] ); */ \
    453                         fmtstr[sizeof(IFMTNP)-2] = f.base;                      /* sizeof includes '\0' */ \
    454                         fmt( os, &fmtstr[star], f.wd, f.val ); \
    455                 } else {                                                                                /* precision */ \
    456                         fmtstr[sizeof(IFMTP)-2] = f.base;                       /* sizeof includes '\0' */ \
    457                         /* printf( "%s\n", &fmtstr[star] ); */ \
    458                         fmt( os, &fmtstr[star], f.wd, f.pc, f.val ); \
    459                 } /* if */ \
    460                 return os; \
    461         } /* ?|? */ \
    462         void ?|?( ostype & os, _Ostream_Manip(T) f ) { (ostype &)(os | f); nl( os ); } \
    463 } // distribution
    464 
    465 IntegralFMTImpl( signed char, 'd', "%    *hh ", "%    *.*hh " )
    466 IntegralFMTImpl( unsigned char, 'u', "%    *hh ", "%    *.*hh " )
    467 IntegralFMTImpl( signed short int, 'd', "%    *h ", "%    *.*h " )
    468 IntegralFMTImpl( unsigned short int, 'u', "%    *h ", "%    *.*h " )
    469 IntegralFMTImpl( signed int, 'd', "%    * ", "%    *.* " )
    470 IntegralFMTImpl( unsigned int, 'u', "%    * ", "%    *.* " )
    471 IntegralFMTImpl( signed long int, 'd', "%    *l ", "%    *.*l " )
    472 IntegralFMTImpl( unsigned long int, 'u', "%    *l ", "%    *.*l " )
    473 IntegralFMTImpl( signed long long int, 'd', "%    *ll ", "%    *.*ll " )
    474 IntegralFMTImpl( unsigned long long int, 'u', "%    *ll ", "%    *.*ll " )
    475 
    476 //*********************************** Floating Point ***********************************
    477 
    478 #define PrintWithDP2( os, format, val, ... ) \
    479         { \
    480                 enum { size = 48 }; \
    481                 char buf[size]; \
    482                 int bufbeg = 0, i, len = snprintf( buf, size, format, ##__VA_ARGS__, val ); \
    483                 if ( isfinite( val ) && (f.base != 'g' || f.pc != 0) ) { /* if number, print decimal point */ \
    484                         for ( i = 0; i < len && buf[i] != '.' && buf[i] != 'e' && buf[i] != 'E'; i += 1 ); /* decimal point or scientific ? */ \
    485                         if ( i == len && ! f.flags.nobsdp ) { \
    486                                 if ( ! f.flags.left ) { \
    487                                         buf[i] = '.'; buf[i + 1] = '\0'; \
    488                                         if ( buf[0] == ' ' ) bufbeg = 1; /* decimal point within width */ \
    489                                 } else { \
    490                                         for ( i = 0; i < len && buf[i] != ' '; i += 1 ); /* trailing blank ? */ \
    491                                         buf[i] = '.'; \
    492                                         if ( i == len ) buf[i + 1] = '\0'; \
    493                                 } /* if */ \
    494                         } /* if */ \
    495                 } /* if */ \
    496                 fmt( os, "%s", &buf[bufbeg] ); \
    497         }
    498 
    499 #define FloatingPointFMTImpl( T, DFMTNP, DFMTP ) \
    500 forall( dtype ostype | ostream( ostype ) ) { \
    501         ostype & ?|?( ostype & os, _Ostream_Manip(T) f ) { \
    502                 if ( sepPrt( os ) ) fmt( os, "%s", sepGetCur( os ) ); \
    503                 char fmtstr[sizeof(DFMTP)];                                             /* sizeof includes '\0' */ \
    504                 if ( ! f.flags.pc ) memcpy( &fmtstr, DFMTNP, sizeof(DFMTNP) ); \
    505                 else memcpy( &fmtstr, DFMTP, sizeof(DFMTP) ); \
    506                 int star = 4;                                                                   /* position before first '*' */ \
    507 \
    508                 /* Insert flags into spaces before '*', from right to left. */ \
    509                 if ( f.flags.left ) { fmtstr[star] = '-'; star -= 1; } \
    510                 if ( f.flags.sign ) { fmtstr[star] = '+'; star -= 1; } \
    511                 if ( f.flags.pad0 ) { fmtstr[star] = '0'; star -= 1; } \
    512                 fmtstr[star] = '%'; \
    513 \
    514                 if ( ! f.flags.pc ) {                                                   /* no precision */ \
    515                         fmtstr[sizeof(DFMTNP)-2] = f.base;                      /* sizeof includes '\0' */ \
    516                         /* printf( "%g %d %s\n", f.val, f.wd, &fmtstr[star]); */ \
    517                         PrintWithDP2( os, &fmtstr[star], f.val, f.wd ) \
    518                 } else {                                                                                /* precision */ \
    519                         fmtstr[sizeof(DFMTP)-2] = f.base;                       /* sizeof includes '\0' */ \
    520                         /* printf( "%g %d %d %s\n", f.val, f.wd, f.pc, &fmtstr[star] ); */ \
    521                         PrintWithDP2( os, &fmtstr[star], f.val, f.wd, f.pc ) \
    522                 } /* if */ \
    523                 return os; \
    524         } /* ?|? */ \
    525         void ?|?( ostype & os, _Ostream_Manip(T) f ) { (ostype &)(os | f); nl( os ); } \
    526 } // distribution
    527 
    528 FloatingPointFMTImpl( double, "%    * ", "%    *.* " )
    529 FloatingPointFMTImpl( long double, "%    *L ", "%    *.*L " )
    530 
    531 //*********************************** Character ***********************************
    532 
    533 forall( dtype ostype | ostream( ostype ) ) {
    534         ostype & ?|?( ostype & os, _Ostream_Manip(char) f ) {
    535                 if ( f.base != 'c' ) {                                                          // bespoke binary/octal/hex format
    536                         _Ostream_Manip(unsigned char) fmtuc @= { f.val, f.wd, f.pc, f.base, {'\0'} };
    537                         fmtuc.flags.pc = f.flags.pc;
    538                         fmtuc.flags.nobsdp = f.flags.nobsdp;
    539 //                      os | fmtuc | nonl;
    540                         (ostype &)(os | fmtuc);
    541                         return os;
    542                 } // if
    543 
    544                 if ( sepPrt( os ) ) fmt( os, "%s", sepGetCur( os ) );
    545 
    546                 #define CFMTNP "% * "
    547                 char fmtstr[sizeof(CFMTNP)];                                            // sizeof includes '\0'
    548                 memcpy( &fmtstr, CFMTNP, sizeof(CFMTNP) );
    549                 int star = 1;                                                                           // position before first '*'
    550 
    551                 // Insert flags into spaces before '*', from right to left.
    552                 if ( f.flags.left ) { fmtstr[star] = '-'; star -= 1; }
    553                 fmtstr[star] = '%';
    554 
    555                 fmtstr[sizeof(CFMTNP)-2] = f.base;                                      // sizeof includes '\0'
    556                 // printf( "%d %s\n", f.wd, &fmtstr[star] );
    557                 fmt( os, &fmtstr[star], f.wd, f.val );
    558                 return os;
    559         } // ?|?
    560         void ?|?( ostype & os, _Ostream_Manip(char) f ) { (ostype &)(os | f); nl( os ); }
    561 } // distribution
    562 
    563 //*********************************** C String ***********************************
    564 
    565 forall( dtype ostype | ostream( ostype ) ) {
    566         ostype & ?|?( ostype & os, _Ostream_Manip(const char *) f ) {
    567                 if ( ! f.val ) return os;                                               // null pointer ?
    568 
    569                 if ( f.base != 's' ) {                                                  // bespoke binary/octal/hex format
    570                         _Ostream_Manip(unsigned char) fmtuc @= { 0, f.wd, f.pc, f.base, {'\0'} };
    571                         fmtuc.flags.pc = f.flags.pc;
    572                         fmtuc.flags.nobsdp = f.flags.nobsdp;
    573                         for ( unsigned int i = 0; f.val[i] != '\0'; i += 1 ) {
    574                                 fmtuc.val = f.val[i];
    575 //                              os | fmtuc | nonl;
    576                                 (ostype &)(os | fmtuc);
    577                         } // for
    578                         return os;
    579                 } // if
    580 
    581                 if ( sepPrt( os ) ) fmt( os, "%s", sepGetCur( os ) );
    582 
    583                 #define SFMTNP "% * "
    584                 #define SFMTP "% *.* "
    585                 char fmtstr[sizeof(SFMTP)];                                             // sizeof includes '\0'
    586                 if ( ! f.flags.pc ) memcpy( &fmtstr, SFMTNP, sizeof(SFMTNP) );
    587                 else memcpy( &fmtstr, SFMTP, sizeof(SFMTP) );
    588                 int star = 1;                                                                   // position before first '*'
    589 
    590                 // Insert flags into spaces before '*', from right to left.
    591                 if ( f.flags.left ) { fmtstr[star] = '-'; star -= 1; }
    592                 fmtstr[star] = '%';
    593 
    594                 if ( ! f.flags.pc ) {                                                   // no precision
    595                         // printf( "%d %s\n", f.wd, &fmtstr[star] );
    596                         fmtstr[sizeof(SFMTNP)-2] = f.base;                      // sizeof includes '\0'
    597                         fmt( os, &fmtstr[star], f.wd, f.val );
    598                 } else {                                                                                // precision
    599                         fmtstr[sizeof(SFMTP)-2] = f.base;                       // sizeof includes '\0'
    600                         // printf( "%d %d %s\n", f.wd, f.pc, &fmtstr[star] );
    601                         fmt( os, &fmtstr[star], f.wd, f.pc, f.val );
    602                 } // if
    603                 return os;
    604         } // ?|?
    605         void ?|?( ostype & os, _Ostream_Manip(const char *) f ) { (ostype &)(os | f); nl( os ); }
    606 } // distribution
    607 
    608397//---------------------------------------
    609398
    … …  
    621410} // distribution
    622411
    623 
    624 //*********************************** Istream ***********************************
    625 
     412//---------------------------------------
    626413
    627414forall( dtype istype | istream( istype ) ) {
    … …  
    650437
    651438        istype & ?|?( istype & is, signed char & sc ) {
    652                 fmt( is, "%hhi", &sc );
     439                fmt( is, "%hhd", &sc );
    653440                return is;
    654441        } // ?|?
    655442
    656443        istype & ?|?( istype & is, unsigned char & usc ) {
    657                 fmt( is, "%hhi", &usc );
     444                fmt( is, "%hhu", &usc );
    658445                return is;
    659446        } // ?|?
    660447
    661448        istype & ?|?( istype & is, short int & si ) {
    662                 fmt( is, "%hi", &si );
     449                fmt( is, "%hd", &si );
    663450                return is;
    664451        } // ?|?
    665452
    666453        istype & ?|?( istype & is, unsigned short int & usi ) {
    667                 fmt( is, "%hi", &usi );
     454                fmt( is, "%hu", &usi );
    668455                return is;
    669456        } // ?|?
    670457
    671458        istype & ?|?( istype & is, int & i ) {
    672                 fmt( is, "%i", &i );
     459                fmt( is, "%d", &i );
    673460                return is;
    674461        } // ?|?
    675462
    676463        istype & ?|?( istype & is, unsigned int & ui ) {
    677                 fmt( is, "%i", &ui );
     464                fmt( is, "%u", &ui );
    678465                return is;
    679466        } // ?|?
    680467
    681468        istype & ?|?( istype & is, long int & li ) {
    682                 fmt( is, "%li", &li );
     469                fmt( is, "%ld", &li );
    683470                return is;
    684471        } // ?|?
    685472
    686473        istype & ?|?( istype & is, unsigned long int & ulli ) {
    687                 fmt( is, "%li", &ulli );
     474                fmt( is, "%lu", &ulli );
    688475                return is;
    689476        } // ?|?
    690477
    691478        istype & ?|?( istype & is, long long int & lli ) {
    692                 fmt( is, "%lli", &lli );
     479                fmt( is, "%lld", &lli );
    693480                return is;
    694481        } // ?|?
    695482
    696483        istype & ?|?( istype & is, unsigned long long int & ulli ) {
    697                 fmt( is, "%lli", &ulli );
     484                fmt( is, "%llu", &ulli );
    698485                return is;
    699486        } // ?|?
    … …  
    718505        istype & ?|?( istype & is, float _Complex & fc ) {
    719506                float re, im;
    720                 fmt( is, "%f%fi", &re, &im );
     507                fmt( is, "%g%gi", &re, &im );
    721508                fc = re + im * _Complex_I;
    722509                return is;
    … …  
    774561} // cstr
    775562
    776 #if 0
    777 forall( dtype istype | istream( istype ) )
    778 istype & ?|?( istype & is, _Istream_skip skip ) {
    779         fmt( is, skip.s, "" );                                                          // no input arguments
    780         return is;
    781 } // skip
    782 
    783 forall( dtype istype | istream( istype ) )
    784 istype & ?|?( istype & is, _Istream_incl incl ) {
    785         size_t len = strlen( incl.scanset ) + 4;                        // extras: "%[]\0"
    786         char fmtstr[len];
    787         fmtstr[0] = '%'; fmtstr[1] = '[';
    788         strcpy( &fmtstr[2], incl.scanset );                                     // after '[', copy includes '\0'
    789         fmtstr[len - 2] = ']'; fmtstr[len - 1] = '\0';
    790         fmt( is, fmtstr, incl.s );
    791         return is;
    792 } // incl
    793 
    794 forall( dtype istype | istream( istype ) )
    795 istype & ?|?( istype & is, _Istream_excl excl ) {
    796         size_t len = strlen( excl.scanset );
    797         char fmtstr[len+5];
    798         fmtstr[0] = '%'; fmtstr[1] = '['; fmtstr[2] = '^';
    799         strcpy( &fmtstr[3], excl.scanset );                                     // after '^', copy includes '\0'
    800         fmtstr[len - 2] = ']'; fmtstr[len - 1] = '\0';
    801         fmt( is, fmtstr, excl.s );
    802         return is;
    803 } // excl
    804 
    805 forall( dtype istype | istream( istype ) )
    806 istype & ?|?( istype & is, _Istream_cstr cstr ) {
    807         fmt( is, "%s", cstr.s );
    808         return is;
    809 } // cstr
    810 
    811 _Istream_cstrW cstr( char * s, int size ) { return (_Istream_cstrW){ s, size }; }
    812 forall( dtype istype | istream( istype ) )
    813 istype & ?|?( istype & is, _Istream_cstrW cstr ) {
    814         enum { size = 16 };
    815         char fmtstr[size];
    816         sprintf( fmtstr, "%%%ds", cstr.size );
    817         fmt( is, fmtstr, cstr.s );
    818         return is;
    819 } // cstr
    820 
    821 forall( dtype istype | istream( istype ) ) istype & ?|?( istype &, _Istream_Cstr ) {
    822        
    823 }
    824 
    825 //*********************************** Manipulators ***********************************
    826 
    827 #define InputFMTImpl( T, CODE ) \
    828 forall( dtype istype | istream( istype ) ) \
    829 istype & ?|?( istype & is, _Istream_Manip(T) f ) { \
    830         enum { size = 16 }; \
    831         char fmtstr[size]; \
    832         if ( f.wd == -1 ) { \
    833                 snprintf( fmtstr, size, "%%%s%s", f.ignore ? "*" : "", CODE ); \
    834         } else { \
    835                 snprintf( fmtstr, size, "%%%s%d%s", f.ignore ? "*" : "", f.wd, CODE ); \
    836         } /* if */ \
    837         /* printf( "%d %s %p\n", f.wd, fmtstr, &f.val ); */ \
    838         fmt( is, fmtstr, &f.val ); \
    839         return is; \
    840 } /* ?|? */
    841 
    842 InputFMTImpl( char, "c" )
    843 InputFMTImpl( signed char, "hhi" )
    844 InputFMTImpl( unsigned char, "hhi" )
    845 InputFMTImpl( signed short int, "hi" )
    846 InputFMTImpl( unsigned short int, "hi" )
    847 InputFMTImpl( signed int, "i" )
    848 InputFMTImpl( unsigned int, "i" )
    849 InputFMTImpl( signed long int, "li" )
    850 InputFMTImpl( unsigned long int, "li" )
    851 InputFMTImpl( signed long long int, "lli" )
    852 InputFMTImpl( unsigned long long int, "lli" )
    853 
    854 InputFMTImpl( float, "f" )
    855 InputFMTImpl( double, "lf" )
    856 InputFMTImpl( long double, "Lf" )
    857 
    858 InputFMTImpl( float _Complex, "ff" )
    859 InputFMTImpl( double _Complex, "lf" )
    860 InputFMTImpl( long double _Complex, "Lf" )
    861 #endif // 0
    862 
    863563// Local Variables: //
    864564// tab-width: 4 //

  • libcfa/src/iostream.hfa

    r46438e4 r05d55ff  
    55// file "LICENCE" distributed with Cforall.
    66//
    7 // iostream.hfa --
     7// iostream --
    88//
    99// Author           : Peter A. Buhr
    1010// Created On       : Wed May 27 17:56:53 2015
    1111// Last Modified By : Peter A. Buhr
    12 // Last Modified On : Tue Jun  4 17:33:43 2019
    13 // Update Count     : 286
     12// Last Modified On : Sat May 11 10:31:27 2019
     13// Update Count     : 232
    1414//
    1515
    … …  
    1717
    1818#include "iterator.hfa"
    19 
    20 
    21 //*********************************** Ostream ***********************************
    22 
    2319
    2420trait ostream( dtype ostype ) {
    … …  
    150146} // distribution
    151147
    152 //*********************************** Manipulators ***********************************
    153 
    154 forall( otype T )
    155 struct _Ostream_Manip {
    156         T val;                                                                                          // polymorphic base-type
    157         unsigned char wd, pc;                                                           // width, precision
    158         char base;                                                                                      // numeric base / floating-point style
    159         union {
    160                 unsigned char all;
    161                 struct {
    162                         unsigned char pc:1;                                                     // precision specified
    163                         unsigned char left:1;                                           // left justify
    164                         unsigned char nobsdp:1;                                         // base prefix / decimal point
    165                         unsigned char sign:1;                                           // plus / minus sign
    166                         unsigned char pad0:1;                                           // zero pad
    167                 } flags;
    168         };
    169 }; // _Ostream_Manip
    170 
    171 //*********************************** Integral ***********************************
    172 
    173 #define IntegralFMTDecl( T, CODE ) \
    174 static inline { \
    175         _Ostream_Manip(T) bin( T v ) { return (_Ostream_Manip(T))@{ v, 1, 0, 'b', { .all : 0 } }; } \
    176         _Ostream_Manip(T) oct( T v ) { return (_Ostream_Manip(T))@{ v, 1, 0, 'o', { .all : 0 } }; } \
    177         _Ostream_Manip(T) hex( T v ) { return (_Ostream_Manip(T))@{ v, 1, 0, 'x', { .all : 0 } }; } \
    178         _Ostream_Manip(T) wd( unsigned char w, T v ) { return (_Ostream_Manip(T))@{ v, w, 0, CODE, { .all : 0 } }; } \
    179         _Ostream_Manip(T) wd( unsigned char w, unsigned char p, T v ) { return (_Ostream_Manip(T))@{ v, w, p, CODE, { .flags.pc : true } }; } \
    180         _Ostream_Manip(T) & wd( unsigned char w, _Ostream_Manip(T) & fmt ) with(fmt) { wd = w; return fmt; } \
    181         _Ostream_Manip(T) & wd( unsigned char w, unsigned char p, _Ostream_Manip(T) & fmt ) with(fmt) { wd = w; pc = p; flags.pc = true; return fmt; } \
    182         _Ostream_Manip(T) & left( _Ostream_Manip(T) & fmt ) { fmt.flags.left = true; return fmt; } \
    183         _Ostream_Manip(T) & upcase( _Ostream_Manip(T) & fmt ) { if ( fmt.base == 'x' || fmt.base == 'b' ) fmt.base -= 32; /* upper case */ return fmt; } \
    184         _Ostream_Manip(T) & nobase( _Ostream_Manip(T) & fmt ) { fmt.flags.nobsdp = true; return fmt; } \
    185         _Ostream_Manip(T) & pad0( _Ostream_Manip(T) & fmt ) { fmt.flags.pad0 = true; return fmt; } \
    186         _Ostream_Manip(T) sign( T v ) { return (_Ostream_Manip(T))@{ v, 1, 0, CODE, { .flags.sign : true } }; } \
    187         _Ostream_Manip(T) & sign( _Ostream_Manip(T) & fmt ) { fmt.flags.sign = true; return fmt; } \
    188 } \
    189 forall( dtype ostype | ostream( ostype ) ) { \
    190         ostype & ?|?( ostype & os, _Ostream_Manip(T) f ); \
    191         void ?|?( ostype & os, _Ostream_Manip(T) f ); \
    192 }
    193 
    194 IntegralFMTDecl( signed char, 'd' )
    195 IntegralFMTDecl( unsigned char, 'u' )
    196 IntegralFMTDecl( signed short int, 'd' )
    197 IntegralFMTDecl( unsigned short int, 'u' )
    198 IntegralFMTDecl( signed int, 'd' )
    199 IntegralFMTDecl( unsigned int, 'u' )
    200 IntegralFMTDecl( signed long int, 'd' )
    201 IntegralFMTDecl( unsigned long int, 'u' )
    202 IntegralFMTDecl( signed long long int, 'd' )
    203 IntegralFMTDecl( unsigned long long int, 'u' )
    204 
    205 //*********************************** Floating Point ***********************************
    206 
    207 // Default suffix for values with no fraction is "."
    208 #define FloatingPointFMTDecl( T ) \
    209 static inline { \
    210         _Ostream_Manip(T) sci( T v ) { return (_Ostream_Manip(T))@{ v, 1, 0, 'e', { .all : 0 } }; } \
    211         _Ostream_Manip(T) hex( T v ) { return (_Ostream_Manip(T))@{ v, 1, 0, 'a', { .all : 0 } }; } \
    212         _Ostream_Manip(T) wd( unsigned char w, T v ) { return (_Ostream_Manip(T))@{ v, w, 0, 'f', { .all : 0 } }; } \
    213         _Ostream_Manip(T) wd( unsigned char w, unsigned char p, T v ) { return (_Ostream_Manip(T))@{ v, w, p, 'f', { .flags.pc : true } }; } \
    214         _Ostream_Manip(T) ws( unsigned char w, unsigned char p, T v ) { return (_Ostream_Manip(T))@{ v, w, p, 'g', { .flags.pc : true } }; } \
    215         _Ostream_Manip(T) & wd( unsigned char w, _Ostream_Manip(T) & fmt ) with(fmt) { wd = w; return fmt; } \
    216         _Ostream_Manip(T) & wd( unsigned char w, unsigned char p, _Ostream_Manip(T) & fmt ) with(fmt) { wd = w; pc = p; flags.pc = true; return fmt; } \
    217         _Ostream_Manip(T) & left( _Ostream_Manip(T) & fmt ) { fmt.flags.left = true; return fmt; } \
    218         _Ostream_Manip(T) upcase( T v ) { return (_Ostream_Manip(T))@{ v, 1, 0, 'G', { .all : 0 } }; } \
    219         _Ostream_Manip(T) & upcase( _Ostream_Manip(T) & fmt ) { fmt.base -= 32; /* upper case */ return fmt; } \
    220         _Ostream_Manip(T) & pad0( _Ostream_Manip(T) & fmt ) { fmt.flags.pad0 = true; return fmt; } \
    221         _Ostream_Manip(T) sign( T v ) { return (_Ostream_Manip(T))@{ v, 1, 0, 'g', { .flags.sign : true } }; } \
    222         _Ostream_Manip(T) & sign( _Ostream_Manip(T) & fmt ) { fmt.flags.sign = true; return fmt; } \
    223         _Ostream_Manip(T) nodp( T v ) { return (_Ostream_Manip(T))@{ v, 1, 0, 'g', { .flags.nobsdp : true } }; } \
    224         _Ostream_Manip(T) & nodp( _Ostream_Manip(T) & fmt ) { fmt.flags.nobsdp = true; return fmt; } \
    225 } \
    226 forall( dtype ostype | ostream( ostype ) ) { \
    227         ostype & ?|?( ostype & os, _Ostream_Manip(T) f ); \
    228         void ?|?( ostype & os, _Ostream_Manip(T) f ); \
    229 }
    230 
    231 FloatingPointFMTDecl( double )
    232 FloatingPointFMTDecl( long double )
    233 
    234 //*********************************** Character ***********************************
    235 
    236 static inline {
    237         _Ostream_Manip(char) bin( char v ) { return (_Ostream_Manip(char))@{ v, 1, 0, 'b', { .all : 0 } }; }
    238         _Ostream_Manip(char) oct( char v ) { return (_Ostream_Manip(char))@{ v, 1, 0, 'o', { .all : 0 } }; }
    239         _Ostream_Manip(char) hex( char v ) { return (_Ostream_Manip(char))@{ v, 1, 0, 'x', { .all : 0 } }; }
    240         _Ostream_Manip(char) wd( unsigned char w, char v ) { return (_Ostream_Manip(char))@{ v, w, 0, 'c', { .all : 0 } }; }
    241         _Ostream_Manip(char) & wd( unsigned char w, _Ostream_Manip(char) & fmt ) with(fmt) { wd = w; return fmt; }
    242         _Ostream_Manip(char) & left( _Ostream_Manip(char) & fmt ) { fmt.flags.left = true; return fmt; }
    243         _Ostream_Manip(char) & upcase( _Ostream_Manip(char) & fmt ) { if ( fmt.base == 'x' || fmt.base == 'b' ) fmt.base -= 32; /* upper case */ return fmt; }
    244         _Ostream_Manip(char) & nobase( _Ostream_Manip(char) & fmt ) { fmt.flags.nobsdp = true; return fmt; }
    245 } // distribution
    246 forall( dtype ostype | ostream( ostype ) ) {
    247         ostype & ?|?( ostype & os, _Ostream_Manip(char) f );
    248         void ?|?( ostype & os, _Ostream_Manip(char) f );
    249 }
    250 
    251 //*********************************** C String ***********************************
    252 
    253 static inline {
    254         _Ostream_Manip(const char *) bin( const char * v ) { return (_Ostream_Manip(const char *))@{ v, 1, 0, 'b', { .all : 0 } }; }
    255         _Ostream_Manip(const char *) oct( const char * v ) { return (_Ostream_Manip(const char *))@{ v, 1, 0, 'o', { .all : 0 } }; }
    256         _Ostream_Manip(const char *) hex( const char * v ) { return (_Ostream_Manip(const char *))@{ v, 1, 0, 'x', { .all : 0 } }; }
    257         _Ostream_Manip(const char *) wd( unsigned char w, const char * v ) { return (_Ostream_Manip(const char *))@{ v, w, 0, 's', { .all : 0 } }; }
    258         _Ostream_Manip(const char *) wd( unsigned char w, unsigned char p, const char * v ) { return (_Ostream_Manip(const char *))@{ v, w, p, 's', { .flags.pc : true } }; }
    259         _Ostream_Manip(const char *) & wd( unsigned char w, _Ostream_Manip(const char *) & fmt ) with(fmt) { wd = w; return fmt; }
    260         _Ostream_Manip(const char *) & wd( unsigned char w, unsigned char p, _Ostream_Manip(const char *) & fmt ) with(fmt) { wd = w; pc = p; flags.pc = true; return fmt; }
    261         _Ostream_Manip(const char *) & left( _Ostream_Manip(const char *) & fmt ) { fmt.flags.left = true; return fmt; }
    262         _Ostream_Manip(const char *) & nobase( _Ostream_Manip(const char *) & fmt ) { fmt.flags.nobsdp = true; return fmt; }
    263 } // distribution
    264 forall( dtype ostype | ostream( ostype ) ) {
    265         ostype & ?|?( ostype & os, _Ostream_Manip(const char *) f );
    266         void ?|?( ostype & os, _Ostream_Manip(const char *) f );
    267 }
    268 
    269 
    270 //*********************************** Istream ***********************************
    271 
     148//---------------------------------------
    272149
    273150trait istream( dtype istype ) {
    … …  
    327204forall( dtype istype | istream( istype ) ) istype & ?|?( istype &, _Istream_cstrC );
    328205
    329 #if 0
    330 struct _Istream_Cstr {
    331         char * s, * scanset;
    332         int wd;                                                                                         // width
    333         bool ignore;                                                                            // no input argument
    334 };
    335 static inline _Istream_Cstr skip( char * s ) { return (_Istream_Cstr){ 0p, s, -1, false }; }
    336 static inline _Istream_Cstr incl( const char * scanset, char * s ) { return (_Istream_Cstr){ s, scanset, -1, false }; }
    337 static inline _Istream_Cstr excl( const char * scanset, char * s ) { return (_Istream_Cstr){ s, scanset, -1, false }; }
    338 static inline _Istream_Cstr cstr( char * s ) { return (_Istream_Cstr){ s, 0p, -1, false }; }
    339 static inline _Istream_Cstr ignore( char * s ) { return (_Istream_Cstr)@{ s, 0p, -1, true }; }
    340 static inline _Istream_Cstr ignore( _Istream_Cstr & fmt ) { fmt.ignore = true; return fmt; }
    341 static inline _Istream_Cstr wd( unsigned int w, char * s ) { return (_Istream_Cstr)@{ s, 0p, w, false }; }
    342 forall( dtype istype | istream( istype ) ) istype & ?|?( istype &, _Istream_Cstr );
    343 
    344 //*********************************** Manipulators ***********************************
    345 
    346 forall( otype T )
    347 struct _Istream_Manip {
    348         T & val;                                                                                        // polymorphic base-type
    349         int wd;                                                                                         // width
    350         bool ignore;                                                                            // no input argument
    351 }; // _Istream_Manip
    352 
    353 #define InputFMTDecl( T ) \
    354 static inline _Istream_Manip(T) ignore( T & v ) { return (_Istream_Manip(T))@{ v, -1, true }; } \
    355 static inline _Istream_Manip(T) ignore( _Istream_Manip(T) & fmt ) { fmt.ignore = true; return fmt; } \
    356 static inline _Istream_Manip(T) wd( unsigned int w, T & v ) { return (_Istream_Manip(T))@{ v, w, false }; } \
    357 forall( dtype istype | istream( istype ) ) { \
    358         istype & ?|?( istype & is, _Istream_Manip(T) f ); \
    359 }
    360 
    361 InputFMTDecl( char )
    362 InputFMTDecl( signed char )
    363 InputFMTDecl( unsigned char )
    364 InputFMTDecl( signed short int )
    365 InputFMTDecl( unsigned short int )
    366 InputFMTDecl( signed int )
    367 InputFMTDecl( unsigned int )
    368 InputFMTDecl( signed long int )
    369 InputFMTDecl( unsigned long int )
    370 InputFMTDecl( signed long long int )
    371 InputFMTDecl( unsigned long long int )
    372 
    373 InputFMTDecl( float )
    374 InputFMTDecl( double )
    375 InputFMTDecl( long double )
    376 
    377 InputFMTDecl( float _Complex )
    378 InputFMTDecl( double _Complex )
    379 InputFMTDecl( long double _Complex )
    380 #endif // 0
    381 
    382 
    383 //*********************************** Time ***********************************
    384 
    385206
    386207#include <time_t.hfa>                                                                   // Duration (constructors) / Time (constructors)

  • src/Parser/parser.yy

    r46438e4 r05d55ff  
    1010// Created On       : Sat Sep  1 20:22:55 2001
    1111// Last Modified By : Peter A. Buhr
    12 // Last Modified On : Tue May 28 17:06:37 2019
    13 // Update Count     : 4354
     12// Last Modified On : Wed May 15 21:25:27 2019
     13// Update Count     : 4296
    1414//
    1515
    … …  
    278278%token OTYPE FTYPE DTYPE TTYPE TRAIT                                    // CFA
    279279%token SIZEOF OFFSETOF
    280 // %token SUSPEND RESUME                                                                        // CFA
    281280%token ATTRIBUTE EXTENSION                                                              // GCC
    282281%token IF ELSE SWITCH CASE DEFAULT DO WHILE FOR BREAK CONTINUE GOTO RETURN
    … …  
    483482%precedence '}'
    484483%precedence '('
    485 
    486 // %precedence RESUME
    487 // %precedence '{'
    488 // %precedence ')'
    489484
    490485%locations                                                                                              // support location tracking for error messages
    … …  
    604599                        $$ = new ExpressionNode( $5 );
    605600                }
    606         // | RESUME '(' comma_expression ')'
    607         //      { SemanticError( yylloc, "Resume expression is currently unimplemented." ); $$ = nullptr; }
    608         // | RESUME '(' comma_expression ')' compound_statement
    609         //      { SemanticError( yylloc, "Resume expression is currently unimplemented." ); $$ = nullptr; }
    610601        ;
    611602
    … …  
    12721263        | RETURN comma_expression_opt ';'
    12731264                { $$ = new StatementNode( build_return( $2 ) ); }
    1274         | RETURN '{' initializer_list_opt comma_opt '}' ';'
     1265        | RETURN '{' initializer_list_opt comma_opt '}'
    12751266                { SemanticError( yylloc, "Initializer return is currently unimplemented." ); $$ = nullptr; }
    1276         // | SUSPEND ';'
    1277         //      { SemanticError( yylloc, "Suspend expression is currently unimplemented." ); $$ = nullptr; }
    1278         // | SUSPEND compound_statement ';'
    1279         //      { SemanticError( yylloc, "Suspend expression is currently unimplemented." ); $$ = nullptr; }
    12801267        | THROW assignment_expression_opt ';'                           // handles rethrow
    12811268                { $$ = new StatementNode( build_throw( $2 ) ); }
    … …  
    21712158
    21722159bit_subrange_size:
    2173         ':' assignment_expression
     2160        ':' constant_expression
    21742161                { $$ = $2; }
    21752162        ;

  • src/main.cc

    r46438e4 r05d55ff  
    1010// Created On       : Fri May 15 23:12:02 2015
    1111// Last Modified By : Peter A. Buhr
    12 // Last Modified On : Wed Jun  5 20:35:13 2019
    13 // Update Count     : 601
     12// Last Modified On : Wed Jun  5 13:48:41 2019
     13// Update Count     : 600
    1414//
    1515
    … …  
    162162        backtrace( 6 );                                                                         // skip first 6 stack frames
    163163        signal( SIGABRT, SIG_DFL);                                                      // reset default signal handler
    164         raise( SIGABRT );                                                                       // reraise SIGABRT
     164                raise( SIGABRT );                                                                       // reraise SIGABRT
    165165} // sigAbortHandler
    166166
Note: See TracChangeset for help on using the changeset viewer.