Home My Page Projects Code Snippets Project Openings SML/NJ
Summary Activity Forums Tracker Lists Tasks Docs Surveys News SCM Files

SCM Repository

[smlnj] Diff of /sml/trunk/HISTORY
ViewVC logotype

Diff of /sml/trunk/HISTORY

Parent Directory Parent Directory | Revision Log Revision Log | View Patch Patch

revision 1066, Thu Feb 14 16:50:02 2002 UTC revision 1645, Mon Oct 11 21:37:17 2004 UTC
# Line 13  Line 13 
13  Description:  Description:
14    
15  ----------------------------------------------------------------------  ----------------------------------------------------------------------
16    Name: Matthias Blume (blume (at) tti - c (dot) org)
17    Date: 2004/10/11 16:37:00 CDT
18    Tag: blume-20041011-regions
19    Description:
20    
21    Made some minor modifications to elabcore.sml to have source regions
22    be propagated more tightly -- resulting in better (i.e., smaller)
23    regions being reported in error- and debug messages.
24    
25    ----------------------------------------------------------------------
26    Name: Matthias Blume (blume (at) tti - c (dot) org)
27    Date: 2004/10/08 22:50:00 CDT
28    Tag: blume-20041008-cmkw
29    Description:
30    
31    Fixed handling of keywords in .cm files: After seeing "is" the lexer
32    treats subsequent occurrences of "group", "library", "source", "is",
33    "*", and "-" as ordinary identifiers rather than keywords.
34    
35    Most seriously, this fixes a problem with CM's "shell" tool.  The tool
36    is supposed to accept a tool argument called "source", but this did
37    not work because of the clash with the keyword.
38    
39    ----------------------------------------------------------------------
40    Name: Matthias Blume (blume (at) tti - c (dot) org)
41    Date: 2004/10/07 16:00:00 CDT
42    Tag: blume-20041007-cleanup
43    Description:
44    
45    Assorted cleanup work:
46    
47      - got rid of intstrmap in favor of using the library's
48        hash table implementation
49      - threw out most of the pathnames stuff, as it was not used anyway
50      - simplified tokentable implementation
51      - fixed some minor spelling errors
52    
53    ----------------------------------------------------------------------
54    Name: Matthias Blume (blume (at) tti - c (dot) org)
55    Date: 2004/10/06 15:15:15 CDT
56    Tag: blume-20041006-handler
57    Description:
58    
59    Cleaned up the absyn to reflect the invariant that HANDLE always
60    carries a FNexp as part of the type definition.  This eliminates some
61    superfluous sanity checks at runtime down the road.
62    
63    Some minor cleanup of the btrace code.
64    
65    ----------------------------------------------------------------------
66    Name: Matthias Blume (blume (at) tti - c (dot) org)
67    Date: 2004/10/01 10:20:30 CDT
68    Tag: blume-20041001-slave
69    Description:
70    
71    Added hack to make slave mode work in the presence of the version
72    tool.  (Still, since the master does two passes over the code for
73    CMB.make, the release number gets bumped twice when slaves are
74    attached.  I don't know if this is worth fixing...)
75    
76    ----------------------------------------------------------------------
77    Name: Matthias Blume (blume (at) tti - c (dot) org)
78    Date: 2004/09/30 10:55:00 CDT
79    Tag: blume-20040930-version
80    Description:
81    
82    * Moved the "version" magic into its own little library under
83    src/system/smlnj/internal.  This avoids expensive reconstruction
84    of a stable src/compiler/core.cm.
85    
86    * At the same time, structure CompilerVersion is now known as structure
87    SMLNJVersion.
88    
89    * Arranged for the version tool to NOT kick in when rebuilding the system
90      (makeml -rebuild, fixpt).  Otherwise one would never reach a fixpoint.
91      Also, loading the versiontool does not work when rebuilding the system
92      because CM is not properly initialized at that time.
93    
94    ----------------------------------------------------------------------
95    Name: Matthias Blume (blume (at) tti - c (dot) org)
96    Date: 2004/09/29 14:00:00 CDT
97    Tag: blume-20040929-autoversion
98    Description:
99    
100    Implemented some CM magic to have
101        file src/compiler/TopLevel/main/version.sml
102    generated automagically.
103    The version is taken from two files: config/version and config/release.
104    The first is expected to contain a two-part version number such as 110.49.
105    The second should contain a single number, but it may be missing.
106    
107    If the environment variable VERSIONTOOL_BUMP_RELEASE is defined at the
108    time the version tool is loaded (which is the first time you say CMB.make),
109    then the tool will increment the value stored in config/release every
110    time CMB.make is invoked.
111    
112    The binfile format is now insensitive to anything beyond the first
113    two components of a version number, so bumping the release does not render
114    binfiles incompatible.  Auto-bumping can be used to keep track of versions
115    during development without invalidating existing binfiles.
116    
117    In any case, every CMB.make updates the date information in version.sml.
118    (This is the date that is printed in the banner.)
119    
120    ----------------------------------------------------------------------
121    Name: Matthias Blume (blume (at) tti - c (dot) org)
122    Date: 2004/09/28 10:53:00 CDT
123    Tag: blume-20040928-controls
124    Description:
125    
126    Some cleanup of the controls code.
127    
128    ----------------------------------------------------------------------
129    Name: Matthias Blume (blume (at) tti - c (dot) org)
130    Date: 2004/09/27 22:08:00 CDT
131    Tag: blume-20040927-controls
132    Description:
133    
134    Added two pieces of functionality to the Controls interface:
135    
136    1.   val save'restore: 'a control -> unit -> unit
137    
138       grabs the current value of the control in stage 1 and restores it
139       in stage 2.
140    
141    2.   val set' : 'a control * 'a -> unit -> unit
142    
143       stores the given value into the control in stage 2 (i.e., delayed)
144       but does all error checking in stage 1.
145       (This is for string controls that need to do parse their argument --
146        something that might fail.  In some cases, notably in CM, one
147        already knows the intended argument but wants to delay the actual
148        assignment until a time when error recovery would be more difficult.)
149    
150    Changed the handling of controls in tool arguments to classes "sml" and
151    "lazysml":
152       - use Controls.save'restore as a more robust way of restoring the
153         old value (in particular: without having to re-parse the string)
154       - use controls to handle the "overload" keyword in the init group
155         (I believe this change actually fixes a long-standing obscure bug.)
156    
157    ----------------------------------------------------------------------
158    Name: Matthias Blume (blume (at) tti - c (dot) org)
159    Date: 2004/09/27 17:00:00 CDT
160    Tag: blume-20040927-lazysml
161    Description:
162    
163    Added a new tool class called "lazysml" to CM's tool chest.  The only
164    difference to "sml" is that compilation is done with Control.lazysml
165    set to true.  A source of class "lazysml" is automatically recognized
166    by a file name suffix of ".lml".
167    
168    In addition to the above feature, the original class "sml" now also
169    supports a tool argument "lazy" which has the same effect.  As a
170    result, the following three lines are equivalent:
171    
172        foo.sml : lazysml
173        foo.sml : sml (lazy)
174        foo.sml (lazy)
175    
176    The setting goes into effect both during parsing and during
177    compilation.  The original setting is restored right after parsing and
178    after compilation, respectively.
179    
180    In addition to all the above, there is also a general mechanism to set
181    ANY of the "controls" that are available at the command line via
182    "-C..." on a per-sml-file basis.  The same rules that apply for "lazy"
183    apply as well. (In fact, "lazy" is implemented as a special case of
184    the general mechanism.)
185    
186    The .cm file syntax uses a new keyword tool argument called "with".
187    There are several ways of indicating the desired settings:
188    
189        foo.sml (with:parser.quotations=true)
190        foo.sml (with:(name:parser.quotations value:true))
191        foo.sml (with:(name:name1 value:value1 name:name2 value:value2 ...))
192        foo.sml (with:(name1=value1 name2=value2 ...))
193        foo.sml (with:(name1=value1 name:name2 value:value2 name3=value3 ...))
194    
195        etc.
196    
197    Another possible abbreviation is to leave out the =v or value:v part
198    if the name refers to a boolean control (in which case the value is
199    taken to be true).  Thus, one could get lazy sml also by saying:
200    
201        foo.sml (with:parser.lazy-keyword=true)
202        foo.sml (with:parser.lazy-keyword)
203        foo.sml (with:(name:parser.lazy-keyword value:true))
204        foo.sml (with:(name:parser.lazy-keyword))
205    
206    ----------------------------------------------------------------------
207    Name: Matthias Blume (blume (at) tti - c (dot) org)
208    Date: 2004/09/24 16:22:00 CDT
209    Tag: blume-20040924-ppc-long-branch
210    Description:
211    
212    Turned message about "emiting long form of branch" off by default.
213    Added a control flag to turn it back on when desired.
214    
215    ----------------------------------------------------------------------
216    Name: Matthias Blume (blume (at) tti - c (dot) org)
217    Date: 2004/09/24 16:05:00 CDT
218    Tag: blume-20040924-rounding
219    Description:
220    
221    Applied patch for setting rounding modes under Mac OS X.  Thanks to
222    Melissa O'Neill for providing the code!
223    
224    ----------------------------------------------------------------------
225    Name: Matthias Blume (blume (at) tti - c (dot) org)
226    Date: 2004/09/23 17:30:00 CDT
227    Tag: blume-20040923-envvars
228    Description:
229    
230    1. Changed definition of type ControlRegistry.registry_tree to
231       include control_info (i.e., the name of the controlling environment
232       variable).
233    
234    2. Added command-line flags -e and -E to print the names of environment
235       variables that can be used to control internal settings.
236       (This uses the new API mentioned in 1.)
237    
238    ----------------------------------------------------------------------
239    Name: Matthias Blume (blume (at) tti - c (dot) org)
240    Date: 2004/09/13 16:50:00 CDT
241    Tag: Release_110_49
242    Description:
243    
244    New working version (110.49).  NEW BOOTFILES!
245    
246    ----------------------------------------------------------------------
247    Name: Matthias Blume
248    Date: 2004/09/13 16:20:00 CDT
249    Tag: blume-20040913-config-mlrisc
250    Description:
251    
252    Put target "mlrisc" back into the default list.
253    (There is no harm in having it, and some users have expressed their
254    wish to have "mlrisc" included by default.)
255    
256    ----------------------------------------------------------------------
257    Name: John Reppy
258    Date: 2004/09/13
259    Tag: jhr-20040913-signals
260    Description:
261    
262    Fixed the signal masking code to properly nest mask/unmask operations
263    on a per-signal basis.
264    
265    ----------------------------------------------------------------------
266    Name: Matthias Blume (blume (at) tti - c (dot) org)
267    Date: 2004/09/08 13:20:00 CDT
268    Tag: blume-20040908-heap-magic
269    Description:
270    
271    Bumped the heap macig number to 0x09082004 to account for the changed
272    layout of the ML frame under MacOS X.
273    
274    ----------------------------------------------------------------------
275    Name: Allen Leung (leunga (at) reservoir (dot) com)
276    Date: 2004/09/03 11:26:00 EST
277    Tag: leunga-20040903-cygwin-install
278    Description:
279    
280       Added a patch to _arch-n-opsys to enable the Cygwin runtime.
281    The Cygwin runtime is turned on by setting the environment
282    variable SMLNJ_CYGWIN_RUNTIME to 1.
283    
284    ----------------------------------------------------------------------
285    Name: Matthias Blume (blume (at) tti - c (dot) org)
286    Date: 2004/08/31 17:15:00 CDT
287    Tag: blume-20040831-core
288    Description:
289    
290    Added some exports to src/compiler/core.cm upon request by J. Joemann.
291    
292    ----------------------------------------------------------------------
293    Name: Matthias Blume (blume (at) tti - c (dot) org)
294    Date: 2004/08/30 17:55:00 CDT
295    Tag: blume-20040830-installer
296    Description:
297    
298    Upon request by Johannes Joemann:
299    
300    - improved ML code of installer to fall back to coping when renaming
301      fails (i.e., when source and target are on different file systems);
302      the code compiles but has yet to be tested in anger
303    - removed mlrisc from list of default targets (config/targets)
304    
305    ----------------------------------------------------------------------
306    Name: Matthias Blume (blume (at) tti - c (dot) org)
307    Date: 2004/08/27 17:20:00 CDT
308    Tag: blume-20040827-ptreql
309    Description:
310    
311    Added ptreql primop to structure InlineT (upon request from Larry
312    Paulson).
313    
314    ----------------------------------------------------------------------
315    Name: Allen Leung (leunga (at) reservoir (dot) com)
316    Date: 2004/08/15 21:21:00 EST
317    Tag: leunga-110_48-udgraph
318    Description:
319    
320    Another bug fix from Carl Hauser:
321    
322    diff /net/niflab/smlnj48/src/MLRISC/graphs/udgraph.sml udgraph.sml
323    > 48c48
324    > <              | rmv((e as (k,_))::es,L) = rmv(es,if k = i then es else
325    > e::L)
326    > ---
327    > >              | rmv((e as (k,_))::es,L) = rmv(es,if k = i then L else e::L)
328    > Without this, any deletion of an edge in an undirected graph does severe
329    > violence to the graph.
330    
331    ----------------------------------------------------------------------
332    Name: Allen Leung (leunga (at) reservoir (dot) com)
333    Date: 2004/08/10 23:35:00 EST
334    Tag: leunga-110_48-ppc
335    Description:
336    
337       The IBM/MacOS syntax switch on PPC was incorrectly swapped.
338    
339    ----------------------------------------------------------------------
340    Name: Matthias Blume (blume (at) tti - c (dot) org)
341    Date: 2004/08/10 12:00:00 CDT
342    Tag: Release_110_48
343    Description:
344    
345    New working version (110.48).  NEW BOOTFILES!
346    
347    ----------------------------------------------------------------------
348    Name: Allen Leung (leunga (at) reservoir (dot) com)
349    Date: 2004/08/09 12:21:00 EST
350    Tag: leunga-110_47-dijsktra
351    Description:
352    
353    Bug fix from Carl Hauser:
354    
355    single_source_shortest_paths in dijkstra.sml was observed to get wrong
356    answers (by comparing to single_source_shortest_paths in bellman-ford.sml).
357    
358    The problem is that following the expression A.update(dist,s,Num.zero)
359    it is necessary to update the priority queue using Q.decreaseWeight(Q,s).
360    
361    ----------------------------------------------------------------------
362    Name: Matthias Blume (blume (at) tti - c (dot) org)
363    Date: 2004/08/06 18:10:00 CDT
364    Tag: blume-20040806-cmdline
365    Description:
366    
367    Fiddled with handling of command-line options:
368    
369      * sml now quits after processing the command line
370        if -H, -S, -h<n>, or -s<n> appears as the last
371        command-line argument
372      * a new option -q terminates the session when encountered on
373        the command line; subsequent arguments will be ignored
374      * bug fixes: short (erroneous) arguments are no longer ignored
375        completely
376    
377    ----------------------------------------------------------------------
378    Name: Allen Leung (leunga (at) reservoir (dot) com)
379    Date: 2004/08/04 18:17:00 EST
380    Tag: leunga-110_47-ppc-ibm-asm
381    Description:
382    
383      - Added minimal IBM assembly syntax support for PowerPC.
384    
385      - Cygwin: manually changed the file cygwin.def.  Some exported symbols have
386        been altered in the runtime.  We need an automatic way to keep the file
387        in sync.
388    
389    ----------------------------------------------------------------------
390    Name: Matthias Blume (blume (at) tti - c (dot) org)
391    Date: 2004/08/04 14:00:00 CDT
392    Tag: Release_110_47
393    Description:
394    
395    New working version (110.47).  NEW BOOTFILES!
396    
397    ----------------------------------------------------------------------
398    Name: Matthias Blume (blume (at) tti - c (dot) org)
399    Date: 2004/08/03 14:25:00 CDT
400    Tag: blume-20040803-callingconv
401    Description:
402    
403    Added low-level support for choosing C calling conventions by
404    twiddling the type of rawccall.  (See
405    src/compiler/Semant/types/cproto.sml for details.)
406    
407    ----------------------------------------------------------------------
408    Name: Matthias Blume (blume (at) tti - c (dot) org)
409    Date: 2004/08/02 15:55:00 CDT
410    Tag: blume-20040802-backout
411    Description:
412    
413    Backed out of change to win32-filesys.c.  The earlier patch to
414    get_file_time caused CM to produce files with the wrong time stamp.
415    
416    ----------------------------------------------------------------------
417    Name: Matthias Blume (blume (at) tti - c (dot) org)
418    Date: 2004/08/02 14:45:00 CDT
419    Tag: blume-20040802-nlffi-win32
420    Description:
421    
422    Added NLFFI support for Win32, adapted from a patch provided by David
423    Hansel.  This is currently completely untested.  Also, the issue
424    concerning stdcall vs. ccall is still unresolved.
425    
426    ----------------------------------------------------------------------
427    Name: Matthias Blume (blume (at) tti - c (dot) org)
428    Date: 2004/07/30 17:55:00 CDT
429    Tag: blume-20040730-various
430    Description:
431    
432    Gearing up towards 110.47...
433    
434    - various minor bugfixes to ml-nlffigen
435    - a beginning of a manual for nlffi
436    
437    - eliminated 'export name=value' in config/install.sh as this does
438      not work with certain versions of /bin/sh
439      (Thanks to David King at Motorola for catching this.)
440    
441    - several bugfixes provided or suggested by David Hansel at Reactive Systems:
442      - added a test for tm==NULL to gmtime.c and localtime.c
443      - applied patch for incorrect GetFileTime under win32
444      - toSeconds -> toMilliseconds in Win32/win32-process.sml
445    
446    ----------------------------------------------------------------------
447    Name: Matthias Blume (blume (at) tti - c (dot) org)
448    Date: 2004/07/21 18:20:00 CDT
449    Tag: blume-20040721-nlffigen
450    Description:
451    
452    - Fixed minor issue in ml-nlffigen: Now generate structure T_foo
453      for a typedef to an incomplete type, but leave out the "typ" member.
454      (This is just for consistency.)
455    - Started to produce what is supposed to become better (i.e., comprehensive)
456      documentation of what ml-nlffigen does and produces.
457    
458    ----------------------------------------------------------------------
459    Name: Matthias Blume (blume (at) tti - c (dot) org)
460    Date: 2004/07/14 16:25:00 CDT
461    Tag: blume-20040714-union
462    Description:
463    
464    Added C_UNION to c-calls/c-types.sml and updated the machinery
465    (ml-nlffigen, cproto.sml) that conveys C function interface
466    information to the code generator.
467    
468    However, the actual architecture-specific implementation of function
469    arguments and results that are C unions is still not implemented.
470    
471    ----------------------------------------------------------------------
472    Name: Allen Leung (leunga (at) reservoir (dot) com)
473    Date: 2004/07/14 14:38:00 EST
474    Tag: leunga-110_46_1-ppc-lwzu
475    Description:
476    
477        Added these instructions to the PowerPC architecture:
478        LBZU(X), LHZU(X), LWZU(X),
479        STWU(X), STFDU, STFSU
480    
481        etc...
482    
483        Note: I haven't added their instruction encoding into the description.
484    
485    ----------------------------------------------------------------------
486    Name: Allen Leung (leunga (at) reservoir (dot) com)
487    Date: 2004/07/13 15:04:00 EST
488    Tag: leunga-110_46_1-ppc-lwarx
489    Description:
490    
491       Added the two instructions LWARX and STWCX to the PowerPC
492    instruction set.
493    
494       A (untested) rewrite of loop-structure.sml.  The old version
495    is completely broken.
496    
497    ----------------------------------------------------------------------
498    Name: Matthias Blume (blume (at) tti - c (dot) org)
499    Date: 2004/07/13 13:50:00 CDT
500    Tag: blume-20040713-nlffi
501    Description:
502    
503    - use paramAlloc to report c-calls with too many arguments
504      (for PPC version where parameter area is pre-allocated)
505    - added ccall_maxargspace to machspec (to implement the above)
506    - made "make" commend in CM's "make" tool configurable
507    - added option (default: on) for passing the name of the SML/NJ's "bin"
508      directory to "make"; the call looks like this:
509    
510         make <options> SMLNJ_BINDIR=<dir> <target>
511    
512      This can be used by the Makefile to, e.g., pick the "right" version
513      of ml-nlffigen.
514    - minor code tweaks
515    
516    ----------------------------------------------------------------------
517    Name: Matthias Blume (blume (at) tti - c (dot) org)
518    Date: 2004/07/12 22:50:00 CDT
519    Tag: blume-110_46_1-macosx-nlffi
520    Description:
521    
522    NLFFI under Mac OS X now working (sort of).  This is largely untested,
523    though.
524    
525    Note:  1. You have to make a new, clean build of the runtime system.
526           2. There are new BOOTFILES, you have to use them!
527              (Doing the bootstrap process yourself would be *very* painful!
528               If you absolutely have to do it, build the system under
529               a different architecture and then cross-compile.)
530    
531    Version bumped to 110.46.1 to account for runtime data format changes.
532    
533    ----------------------------------------------------------------------
534    Name: Matthias Blume (blume (at) tti - c (dot) org)
535    Date: 2004/06/18 14:30:00 CDT
536    Tag: blume-20040618-unix
537    Description:
538    
539    Changed the implementation of structure Unix so that the same stream
540    is returned every time one of the {text,bin}{In,Out}streamOf functions
541    is invoked on the same proc.  This is not what the spec currently
542    says -- although IMO it arguably should.  (See discussion below.)
543    
544    ----------------------------------------------------------------------
545    Name: Matthias Blume (blume (at) tti - c (dot) org)
546    Date: 2004/06/17 18:15:00 CDT
547    Tag: Release_110_46
548    Description:
549    
550    New working version (110.46).  NEW BOOTFILES!
551    
552    ----------------------------------------------------------------------
553    Name: Matthias Blume (blume (at) tti - c (dot) org)
554    Date: 2004/06/17 17:20:00 CDT
555    Tag: blume-20040617-timer-unix
556    Description:
557    
558    Changed the interface of structures Timer and Unix to match the most
559    recent Basis spec.
560    
561    In the case of Unix there still seems to be an open/weird issue:
562    
563       The {text,bin}{In,Out}streamOf functions are supposed to create
564       fresh streams whenever they are called -- as opposed to have them
565       return the same stream every time.  This design is supposed to
566       prevent space leaks caused by proc values hanging on to streams.
567    
568       The reap function, on the other hand, is supposed to close the
569       streams.  This cannot be done without having a handle on the
570       stream in proc after all...
571    
572    I took the liberty to implement the following stopgap solution:
573    
574      The proc value hangs on to the most recently created stream(s).
575      Reap closes those.  If either or both of the two streams hadn't
576      been created at all yet, then reap will close the corresponding
577      file descriptors directly.
578    
579    PS: I don't understand the original space leak argument anymore.  If
580    a proc hangs on to the imperative stream, then I/O operations on those
581    will advance the state of the cached stream and avoid the space leak.
582    
583    ----------------------------------------------------------------------
584    Name: Matthias Blume (blume (at) tti - c (dot) org)
585    Date: 2004/05/28 16:45:00 CDT
586    Tag: blume-20040528-basis
587    Description:
588    
589    Added signature PACK_REAL and exported functor PrimIO.
590    
591    ----------------------------------------------------------------------
592    Name: Matthias Blume (blume (at) tti - c (dot) org)
593    Date: 2004/05/25 16:00:00 CDT
594    Tag: blume-20040525-group-owner
595    Description:
596    
597    CM now ignores (but still accepts) the "owner" information in group
598    descriptions.  The owner of a group is its next enclosing
599    library. Each group must have a unique owner.  (There is a virtual
600    "toplevel" library that owns groups which are not nested within a real
601    library.)  Previously, each group had to explicitly declare its owner,
602    and CM would check that such a declaration is correct.  The new scheme
603    is to have CM check that for each group there is precisely one owning
604    library.
605    
606    The advantage of the new scheme is that the programmer no longer needs
607    to maintain the somewhat annoying owner information.  The downside is
608    that CM cannot enforce the ownership rule across multiple runs of
609    CM.make.  Fortunately, enclosing the same group in two different
610    libraries A and B which are not part of the same program does not
611    cause real problems.
612    
613    ----------------------------------------------------------------------
614    Name: Matthias Blume (blume (at) tti - c (dot) org)
615    Date: 2004/05/20 16:00:00 CDT
616    Tag: blume-20040520-win32
617    Description:
618    
619    Made the win32 version work again.  (Strangely, a misplaced comma had
620    slipped into win32-process.c which prevented the runtime from being
621    compiled correctly.)
622    
623    Also, included a minor addition to ml-build.bat analogous to what was
624    done in blume-20040519-ml-build.
625    
626    ----------------------------------------------------------------------
627    Name: Matthias Blume (blume (at) tti - c (dot) org)
628    Date: 2004/05/19 22:10:00 CDT
629    Tag: blume-20040519-ml-build
630    Description:
631    
632    Arranged for ml-build to clean up after itself a little bit better.
633    The script generates a temporary SML source file and compiles it using
634    CM, so CM generates metadata (GUID, SKEL, objectfile) for it.  It now
635    gets rid of those at the end, so they don't accumulate under .cm.
636    
637    This required a minor change to install.sh because the name of the
638    metadata directory (default: .cm) is actually configurable at
639    installation time.
640    
641    ----------------------------------------------------------------------
642    Name: Matthias Blume (blume (at) tti - c (dot) org)
643    Date: 2004/05/18 15:50:00 CDT
644    Tag: blume-20040518-mkreader
645    Description:
646    
647    Added Posix.IO.mk{Bin,Text}{Reader,Writer} by lifting their respective
648    implementations from internal modules PosixBinPrimIO and PosixTextPrimIO.
649    
650    ----------------------------------------------------------------------
651    Name: Matthias Blume (blume (at) tti - c (dot) org)
652    Date: 2004/05/11 14:35:00 CDT
653    Tag: blume-20040511-win32sock
654    Description:
655    
656    Added previously missing support for many socket-related functions
657    under win32.  Thanks to David Hansel <hansel@reactive-systems.com>
658    for the voluminous patch!
659    
660    (I have not tested this patch under win32 yet.)
661    
662    Here is David's e-mail:
663    
664    Hi,
665    
666    Attached to this email you find a diff against sml/nj 110.45
667    that will enable socket support under Windows.
668    
669    To apply the patch (using unix or cygwin)
670    1) gunzip runtime.diff.gz
671    2) "cd" into "src/runtime" in the source tree of a fresh
672       110.45 installation.
673    3) patch -p 1 < [your/path/to]runtime.diff
674    
675    The code compiles fine but has NOT yet been extensively tested.
676    I only ran a few tests for basic socket client functionality
677    (which worked fine).  Especially the functions that use ioctl
678    are not tested at all and might not work (see below).
679    
680    I implemented this since we want to move to a newer version of sml/nj
681    but need socket support in order to use it.  This is the first time I
682    even had a look at the sml/nj source,  so please review my changes
683    before making this part of the distribution!  Here are a few issues
684    that I think might be better for someone to solve who is more
685    familiar with the sml/nj source (and socket programming):
686    
687    - getnetbyaddr.c and getnetbyname.c will raise a "not implemented"
688      exception since I could not figure out what the windows equivalent
689      of these functions is
690    
691    - In sockets-osdep.h there are a some #include statements that are
692      only used in a few files that include sockets-osdep.h
693    
694    - In smlnj-sock-lib.c, function init_fn() calls WSAStartup() but
695      does not process its return value since I don't know how to
696      report an error upwards.
697    
698    - It would probably be good to have a call to WSACleanup() when
699      the library is unloaded (if there is such a possibility).
700      Otherwise I think Windows will take care of this automatically
701      when the process finishes.
702    
703    - I used ioctlsocket() as a replacement for ioctl() but I have
704      no idea if that is actually the proper replacement on Windows.
705    
706    - All these issues are marked in the code by "FIXME" comments.
707    
708    We use sml/nj extensively in our products and are quite happy
709    with it.  I hope this contribution will help you.
710    
711    Keep up the good work!
712    
713    David
714    
715    ----------------------------------------------------------------------
716    Name: Matthias Blume (blume (at) tti - c (dot) org)
717    Date: 2004/05/11 14:20:00 CDT
718    Tag: blume-20040511-installml
719    Description:
720    
721    Fixed two bugs in installml script.  (Thanks to Vesa A. Norrman for
722    the patch.)
723    
724    ----------------------------------------------------------------------
725    Name: Matthias Blume (blume (at) tti - c (dot) org)
726    Date: 2004/05/11 14:05:00 CDT
727    Tag: blume-20040511-nlffi-netbsd
728    Description:
729    
730    Added support for nlffi under netbsd.  (Thanks to Vesa A. Norrman for
731    the patch.)
732    
733    ----------------------------------------------------------------------
734    Name: Matthias Blume (blume (at) tti - c (dot) org)
735    Date: 2004/05/11 12:05:00 CDT
736    Tag: blume-20040511-exports
737    Description:
738    
739    As per request by Adam Chlipala <adam@hcoop.net>, extended various
740    export lists in compiler-related .cm-files.
741    
742    ----------------------------------------------------------------------
743    Name: Matthias Blume (blume (at) tti - c (dot) org)
744    Date: 2004/05/11 11:35:00 CDT
745    Tag: blume-20040511-allsource
746    Description:
747    
748    The installer now honors the "src-smlnj" target again, although its meaning
749    has changed from "all sources required for the compiler" to "all sources
750    the installer knows about".  In other words, if you enable "src-smlnj"
751    in the "targets" file, then the installer will pull in sources for
752    everything.  (Notice that this refers to source code only.  Compiled
753    code is still only installed for modules that were requested explicitly
754    or which are required for other modules that were requested explicitly.)
755    
756    ----------------------------------------------------------------------
757    Name: Matthias Blume (blume (at) tti - c (dot) org)
758    Date: 2004/04/23 17:40:00 CDT
759    Tag: blume-20040423-ieee-scan
760    Description:
761    
762    Fixed IEEEReal.scan (and .fromString) so that if there is an overflow
763    in the exponent calculation we get INF or ZERO (depending on the mantissa
764    and the sign of the exponent).
765    
766    ----------------------------------------------------------------------
767    Name: Matthias Blume (blume (at) tti - c (dot) org)
768    Date: 2004/04/23 10:40:00 CDT
769    Tag: blume-20040423-ml-build
770    Description:
771    
772    The ml-build script now terminates with a non-0 status when something
773    goes wrong.
774    
775    ----------------------------------------------------------------------
776    Name: Matthias Blume (blume (at) tti - c (dot) org)
777    Date: 2004/04/22 16:35:00 CDT
778    Tag: blume-20040422-Option
779    Description:
780    
781    Made exception Option to be the same as exception Option.Option
782    (as it should be).
783    
784    ----------------------------------------------------------------------
785    Name: Allen Leung (leunga (at) reservoir (dot) com)
786    Date: 2004/03/19 14:40:00 EST
787    Tag: leunga-20040319-cygwin-nlffi
788    Description:
789    
790    Fixed the runtime so that ml-nlffi-lib runs on the cygwin version
791    of SML/NJ.  The problem is that
792    
793       lib = dlopen(NULL, ...)
794       f   = dlsym(lib, "malloc");
795    
796    does not work on Windows unless we explicitly export symbols
797    such as 'malloc' during linking.  We fixed this by explicitly
798    exporting the required symbols with the magic gcc incantation:
799    
800        -Wl,--export-all cygwin.def
801    
802    where cygwin.def is a file containing all the symbols that
803    we wish to export.
804    
805    I suspect this is a Windows problem and we'll have to
806    do the same (somehow with windows compilers) when
807    we build the native win32 version with the system
808    calls LoadLibrary/GetProcAddress.
809    
810    ----------------------------------------------------------------------
811    Name: Matthias Blume (blume (at) tti - c (dot) org)
812    Date: 2004/03/04 16:35:00 CST
813    Tag: blume-20040304-intinf-fmt
814    Description:
815    
816    Fixed problem with IntInf.fmt (sign would show up on the right instead
817    of on the left for BIN, OCT, and HEX).
818    
819    ----------------------------------------------------------------------
820    Name: Matthias Blume (blume (at) tti - c (dot) org)
821    Date: 2004/03/04 11:25:00 CST
822    Tag: blume-20040304-symlinks
823    Description:
824    
825    Fixed problem with installer script (unix only) where bin/ml-yacc and
826    friends pointed (via symlinks) to absolute locations instead of just
827    .run-sml.  This was reported by Vesa A Norrman.
828    
829    ----------------------------------------------------------------------
830    Name: Matthias Blume (blume (at) tti - c (dot) org)
831    Date: 2004/02/13 14:50:00 CST
832    Tag: Release_110_45
833    Description:
834    
835    New working version (110.45). New bootfiles.
836    
837    ----------------------------------------------------------------------
838    Name: Matthias Blume (blume (at) tti - c (dot) org)
839    Date: 2004/01/26 15:15:15 CST
840    Tag: blume-20040126-toplevel
841    Description:
842    
843    Improved handling of exceptions at the interactive toplevel.
844    
845    ----------------------------------------------------------------------
846    Name: Matthias Blume (blume (at) tti - c (dot) org)
847    Date: 2004/01/26 11:25:00
848    Tag: blume-20040126-app
849    Description:
850    
851    Type of top-level "app" corrected.
852    Added code for setting vp_limitPtrMask to Win32-specific runtime.
853    
854    ----------------------------------------------------------------------
855    Name: Matthias Blume (blume (at) tti - c (dot) org)
856    Date: 2003/11/18 17:10 CST
857    Tag: blume-20031118-basis-fiddle
858    Description:
859    
860    - changed Timer interface to what might become the spec
861    - POSIX_FLAGS -> BIT_FLAGS according to spec
862    - some other minor discrepancies wrt. spec eliminated
863    
864    ----------------------------------------------------------------------
865    Name: Matthias Blume (blume (at) tti - c (dot) org)
866    Date: 2003/11/06 12:00:00 CST
867    Tag: Release_110_44
868    Description:
869    
870    New working version (110.44). New bootfiles.
871    
872    ----------------------------------------------------------------------
873    Name: Matthias Blume (blume (at) tti - c (dot) org)
874    Date: 2003/11/04 11:50:00 CST
875    Tag: blume-20031104-move-libraries
876    Description:
877    
878    Eliminated the "dont_move_libraries" directive in config/targets.
879    (The mechanism was broken and could not be fixed easily.  Moreover,
880    there does not seem to be any reason not to move all libraries into
881    lib during installation.  I originally implemented this directive as a
882    backward-compatibility feature when I first introduced the new CM.
883    Now that things have been stable for a long time and going back to the
884    old CM is not an option, there is no reason to keep it around.)
885    
886    ----------------------------------------------------------------------
887    Name: Matthias Blume (blume (at) tti - c (dot) org)
888    Date: 2003/11/03 16:00:00 CST
889    Tag: blume-20031103-installdir
890    Description:
891    
892    Made installer honor INSTALLDIR variable again.  (Thanks to Chris
893    Richards for pointing out the problem and providing the solution.)
894    
895    ----------------------------------------------------------------------
896    Name: Matthias Blume (blume (at) tti - c (dot) org)
897    Date: 2003/10/01 17:05:00 CDT
898    Tag: blume-20031001-lal-mlrisc
899    Description:
900    
901    MLRISC bug fix from Lal.
902    
903    ----------------------------------------------------------------------
904    Name: Matthias Blume (blume (at) tti - c (dot) org)
905    Date: 2003/09/30 16:10:00 CDT
906    Tag: blume-20030930-primio-bat
907    Description:
908    
909    1. Added openVector, nullRd, and nullWr to PRIM_IO.
910    2. Improved .bat files (for Win32 port) to make things work under Win95.
911       (thanks to Aaron S. Hawley for this one)
912    
913    ----------------------------------------------------------------------
914    Name: Matthias Blume (blume (at) tti - c (dot) org)
915    Date: 2003/09/26 16:05:00 CDT
916    Tag: blume-20030926-wrappriv
917    Description:
918    
919    Added missing wrapper for privilege "primitive" in $smlnj/viscomp/core.cm.
920    
921    ----------------------------------------------------------------------
922    Name: Matthias Blume (blume (at) tti - c (dot) org)
923    Date: 2003/09/26 15:00:00 CDT
924    Tag: blume-20030926-110_43_3
925    Description:
926    
927    - additional cleanup
928    - version number bump, NEW BOOTFILES
929    
930    ----------------------------------------------------------------------
931    Name: Matthias Blume (blume (at) tti - c (dot) org)
932    Date: 2003/09/26 12:00:00 CDT
933    Tag: blume-20030926-ppautoload
934    Description:
935    
936    I modified the read-eval-print loop so that the autoloader gets
937    invoked whenever the prettyprinter tries to look up a symbol that
938    is not currently defined in the toplevel environment but which
939    appears in CM's autoload registry.  As a result, we see far fewer of
940    those ?.Foo.Bar.xxx names in the prettyprinter's output.
941    
942    In addition to this I tried to clean up some pieces of the Basis
943    implementation (e.g., Socket, Word8Array) in order to prevent other
944    instances of these ?.Foo.Bar.xxx names from being printed.
945    
946    The mechanism that picks names for types still needs some work, though.
947    (Right now it seems that if there is a type A.t which is defined to
948    be B.u, but B is unavailable at toplevel, then A.t gets printed as
949    "?.B.u" although the perhaps more sensible solution would be to use
950    "A.t" in this case.  In other words, the prettyprinter should follow
951    a chain of DEFtycs not farther than there are corresponding toplevel
952    names in the current environment.)
953    
954    ----------------------------------------------------------------------
955    Name: Matthias Blume (blume (at) tti - c (dot) org)
956    Date: 2003/09/24 16:31:00 CDT
957    Tag: blume-20030924-installer
958    Description:
959    
960    Another installer tweak:  All the ML code for the installer is now
961    compiled during CMB.make and put into a little library called
962    $smlnj/installer.cm.  The installation then simply invokes
963    
964       sml -m $smlnj/installer.cm
965    
966    and everything happens automagically.
967    
968    Win32:  ML code senses value of environment variable SMLNJ_HOME.
969    Unix:   ML code senses values of environment variables ROOT, CONFIGDIR,
970            and BINDIR.
971    
972    The new scheme guarantees that the ML code responsible for the installation
973    is in sync with the APIs of the main system.  Also, the installer is
974    somewhat faster because the installer script is precompiled.
975    
976    ----------------------------------------------------------------------
977    Name: Matthias Blume (blume (at) tti - c (dot) org)
978    Date: 2003/09/24 15:35:00 CDT
979    Tag: blume-20030924-synsock
980    Description:
981    
982    Added a signature SYNCHRONOUS_SOCKET to basis.cm.  This is like SOCKET
983    but excludes all non-blocking operations.  Defined SOCKET (in Basis)
984    and CML_SOCKET in terms of SYNCHRONOUS_SOCKET.  Removed superfluous
985    implementations of non-blocking operations from CML's Socket
986    structure.
987    
988    ----------------------------------------------------------------------
989    Name: Matthias Blume (blume (at) tti - c (dot) org)
990    Date: 2003/09/24 15:10:05 CDT
991    Tag: blume-20030924-sockets
992    Description:
993    
994    1. Fixed SOCKET API and implementation to match Basis spec.
995       This required changing the internal representation of sockets to one
996       that remembers (for each socket file descriptor) whether it is currently
997       blocking or non-blocking.  This state is maintained lazily (i.e., a system
998       call is made only if the state actually needs to change).
999    
1000    2. OS-specific details of sockets were moved into separate files, thus
1001       making it possible to unify the bulk of the socket implementations
1002       between Unix and Win32.
1003    
1004    3. CML's socket API changed accordingly.
1005       (Note that we need to remove non-blocking functions from this API
1006       since they are redundant in the case of CML!)
1007    
1008    4. CML's socket implementation now makes use of non-blocking functions
1009       provided by Basis, thus removing all OS-dependent code from this part
1010       of CML.
1011    
1012    5. Changed Real64.precision from 52 to 53.  Minor cleanup in Real64 code.
1013    
1014    ----------------------------------------------------------------------
1015    Name: Matthias Blume (blume (at) tti - c (dot) org)
1016    Date: 2003/09/22 12:10:00 CDT
1017    Tag: blume-20030922-110_43_2
1018    Description:
1019    
1020    Made a new interim version and bootfiles for developer's bootstrapping
1021    convenience.
1022    
1023       110.43.2 -- NEW BOOTFILES
1024    
1025    ----------------------------------------------------------------------
1026    Name: Matthias Blume (blume (at) tti - c (dot) org)
1027    Date: 2003/09/19 15:55:00 CDT
1028    Tag: blume-20030919-cmdir
1029    Description:
1030    
1031    1. new-install.sh -> install.sh
1032    2. changed default CM "metadata" directory name to ".cm" (instead of "CM")
1033    3. tweaked installer so that another name instead of .cm can be chosen
1034       at install time (by setting the CM_DIR_ARC environment variable
1035       during installation); once installation is complete, the name is
1036       fixed
1037    
1038    ----------------------------------------------------------------------
1039    Name: Matthias Blume (blume (at) tti - c (dot) org)
1040    Date: 2003/09/18 16:00:00 CDT
1041    Tag: blume-20030918-110_43_1
1042    Description:
1043    
1044    Made a new interim version and bootfiles for developer's bootstrapping
1045    convenience.
1046    
1047       110.43.1 -- NEW BOOTFILES
1048    
1049    ----------------------------------------------------------------------
1050    Name: Matthias Blume (blume (at) tti - c (dot) org)
1051    Date: 2003/09/18 15:20:00 CDT
1052    Tag: blume-20030918-misc
1053    Description:
1054    
1055    1. Exported fractionsPerSecond etc. from TimeImp (but not from Time as
1056       this seems to be controversial at the moment) and used those in
1057       Posix.ProcEnv.times.
1058    
1059    2. Added Time.{from,to}Nanoseconds to Time.
1060    
1061    3. Improved Real.{from,to}LargeInt by avoiding needless calculations.
1062       For example, fromLargeInt never needs to look at more than 3 "big
1063       digits" to get its 53 bits of precision.
1064    
1065    ----------------------------------------------------------------------
1066    Name: Matthias Blume (blume (at) tti - c (dot) org)
1067    Date: 2003/09/17 16:30:00 CDT
1068    Tag: blume-20030917-real32-slices
1069    Description:
1070    
1071    Added an entry to the primitive environment
1072    (compiler/Semant/statenv/prim.sml) for int32->real64 conversion and
1073    added code to compiler/CodeGen/main/mlriscGen.sml to implement it.
1074    
1075    Removed some of the "magic" constants in real64.sml and replaced them
1076    with code that generates these values from their corresponding
1077    integer counterparts.
1078    
1079    Made all(?) the slice-related changes to the Basis and made everything
1080    compile again...
1081    
1082    ----------------------------------------------------------------------
1083    Name: Matthias Blume (blume (at) tti - c (dot) org)
1084    Date: 2003/09/15 17:45:00 CDT
1085    Tag: blume-20030915-rbase
1086    Description:
1087    
1088    Fixed bug in Real.fromLargeInt.
1089    
1090    ----------------------------------------------------------------------
1091    Name: Matthias Blume (blume (at) tti - c (dot) org)
1092    Date: 2003/09/13 18:11:00 CDT
1093    Tag: blume-20030913-libinstall
1094    Description:
1095    
1096    Minor bugfix in config/libinstall (set anchor with path to
1097    standalone tool after installing it, otherwise libraries that
1098    need ml-lex or ml-yacc won't compile the first time the installer
1099    runs).
1100    
1101    ----------------------------------------------------------------------
1102    Name: Matthias Blume (blume (at) tti - c (dot) org)
1103    Date: 2003/09/12 11:45:00 CDT
1104    Tag: blume-20030912-various
1105    Description:
1106    
1107     - fixed bug in Real.toLargeInt
1108     - fixed bug in Posix.ProcEnv.times
1109     - changed inputLine functions to return an option
1110     - minor installer improvements / bugfixes
1111     - changed default @SMLalloc parameter for x86/celeron to 64k
1112    
1113    ----------------------------------------------------------------------
1114    Name: Matthias Blume (blume (at) tti - c (dot) org)
1115    Date: 2003/09/09 22:00:00 CDT
1116    Tag: Release_110_43
1117    Description:
1118    
1119        New working release 110.43.  New bootfiles.
1120    
1121    ----------------------------------------------------------------------
1122    Name: Matthias Blume (blume (at) tti - c (dot) org)
1123    Date: 2003/09/09 19:20:00 CDT
1124    Tag: blume-20030909-installer
1125    Description:
1126    
1127    Rewrote large parts of config/install.sh in SML (config/libinstall.sml).
1128    Modified config/install.bat to take advantage of it.  Also modified
1129    config/install.sh (and called it config/new-install.sh) to take advantage
1130    of it on Unix systems.  (The SML code is (supposed to be) platform-
1131    independent.)
1132    
1133    The installer can now install everything under Win32
1134    as well as under *nix as long as it compiles.
1135    
1136    Other changes:
1137    
1138      - made CML compile again under Win32
1139      - made eXene compile under Win32 (by providing a fake structure UnixSock
1140        and by using OS.Process.getEnv instead of Posix.ProcEnv.getenv)
1141      - fixed a bug in nowhere: it assumed that type OS.Process.status is the
1142        same as type int;  under Win32 it isn't
1143      - fixed some slice-related problems in the win32-specific parts of CML
1144      - added a functor argument "sameVol" to os-path-fn.sml in the Basis
1145        (under Win32, the volume name is case-insensitive, and the
1146        OS.Path code compares volume names for equality)
1147    
1148    ----------------------------------------------------------------------
1149    Name: Matthias Blume (blume (at) tti - c (dot) org)
1150    Date: 2003/09/08 11:55:00 CDT
1151    Tag: blume-20030908-fullpath
1152    Description:
1153    
1154    Made Win32 version of OS.FileSys.fullPath return current directory
1155    when given an empty string.  This is what the spec says, and incidentally,
1156    CM depends on it.  (CM otherwise goes into an infinite loop in certain
1157    cases when presented with the name of a non-existing .cm file.)
1158    
1159    ----------------------------------------------------------------------
1160    Name: Matthias Blume (blume (at) tti - c (dot) org)
1161    Date: 2003/09/04 16:30:00 CDT
1162    Tag: blume-20030905-slices-etc
1163    Description:
1164    
1165      1. Changed interface to vectors and arrays in Basis to match
1166         (draft) Basis spec.
1167      2. Added signatures and implementations of slices according to
1168         Basis spec.
1169      3. Edited source code throughout the system to make it compile again
1170         under 1. and 2.  (In some cases code had to be added to have it
1171         match the new signatures.)
1172      4. MLRISC should be backward-compatible: the copies of the originals
1173         of files that needed to change under 3. were retained, the .cm files
1174         check the compiler version number and use old versions when
1175         appropriate.
1176      5. Changed type of OS.FileSys.readDir and Posix.FileSys.readdir to
1177         dirstream -> string option (in accordance with Basis spec).
1178      6. When generating code that counts lines, ml-lex used function
1179         CharVector.foldli, taking advantage of its old interface.
1180         This has been replaced with the corresponding code from
1181         CharVectorSlice.  (html-lex must be re-lexed!)
1182      7. BitArray in smlnj-lib/Util has been extended/modified to match the
1183         new MONO_ARRAY signature.  (Do we need BitArraySlice?)
1184      8. Removed temporary additions (fromInternal, toInternal) from the
1185         (now obsolete) IntInf in smlnj-lib/Util.
1186      9. Cleaned up structure Byte.
1187     10. Added localOffset, scan, and fromString to Date (according to spec).
1188         Cleaned/corrected implementation of Date.
1189         (Still need to check for correctness; implement better canonicalizeDate.)
1190     11. Added "scan" to signature IEEE_REAL.
1191     12. Some improvements to IntInf [in particular: efficiency-hack for
1192         mod and rem when second operand is 2 (for parity checks).]
1193     13. Changed representation of type Time.time, using a single IntInf.int
1194         value counting microseconds.  This considerably simplified the
1195         implementation of structure Time.  We now support negative time
1196         values; scan and fromString handle signs.
1197     14. Functor PrimIO now takes two additional arguments (VectorSlice and
1198         ArraySlice).
1199    
1200    ----------------------------------------------------------------------
1201    Name: Matthias Blume (blume (at) tti - c (dot) org)
1202    Date: 2003/08/28 17:00:00 CDT
1203    Tag: blume-20030828-intinf
1204    Description:
1205    
1206    This is a major update which comes with a version number bump
1207    (110.42.99 -- yes, we are really close to 110.43 :-), NEW BOOTFILES,
1208    and an implementation of IntInf in the Basis.
1209    
1210    There are a fairly large number of related changes and updates throughout
1211    the system:
1212    
1213    Basis:
1214            - Implemented IntInf.
1215            - Made LargeInt a projection of IntInf (by filtering through INTEGER).
1216            - Added some missing Real64 operations, most notably Real.toLargeInt.
1217            - Added FixedInt as a synonym for Int32.
1218    
1219    compiler:
1220            * Added support for a built-in intinf type.
1221              - literals
1222              - pattern matching
1223              - conversion shortcuts (Int32.fromLarge o Int.toLarge etc.)
1224              - overloading on literals and operations
1225    
1226            This required adding a primitive type intinf, some additional
1227            primops, and implementations for several non-trivial intinf
1228            operations in Core.  (The intinf type is completely abstract
1229            to the compiler; all operations get delegated back to the Core.)
1230    
1231            * Intinf equality is handled by polyequal.  However, the compiler
1232            does not print its usual warning in this case (since polyequal
1233            is the right thing to do there).
1234    
1235            * Improved the organization of structure InlineT.
1236    
1237            * A word about conversion primops:
1238            If conversions involving intinf do not cancel out during
1239            CPS contract, then the compiler must insert calls to Core functions.
1240            Since all core access must be resolved already during the FLINT
1241            translate phase, it would be too late a the time of CPS contract
1242            to add new Core calls.  For this reason, conversion primops
1243            for intinf carry two arguments:  1. the numeric argument that
1244            they are supposed to convert, and 2. the Core function that
1245            can help with this conversion if necessary.  If CPS contract
1246            eliminates a primop, then the associated Core function becomes
1247            dead and goes away.  Intinf conversion primops that do not get
1248            eliminated by CPS contract get rewritten into calls of their
1249            core functions by a separate, new phase.
1250    
1251    interactive system:
1252            - Control.Print.intinfDepth controls max length of intinf constants
1253            being printed. (Analogous to Control.Print.stringDepth.)
1254            - Cleanup in printutil and pputil: got rid of unused stuff and
1255            duplicates; replaced some of the code with code that makes better
1256            use of library functionality.
1257    
1258    CM:
1259            Bugfix: parse-errors in init group (system/smlnj/init/init.cmi)
1260            are no longer silent.
1261    
1262    CKIT:
1263            Fixed mismatched uses of Int32 and LargeInt.  I always decided
1264            in favor of LargeInt -- which is now the same as IntInf.
1265            CKIT-knowledgable people should check whether this is what's
1266            intended and otherwise change things back to using Int32 or
1267            FixedInt.
1268    
1269    Throughout the code:
1270            Started using IntInf.int literals and built-in operations
1271            (e.g., comparison with 0) where this seems appropriate.
1272    
1273    
1274    ----------------------------------------------------------------------
1275    Name: Dave MacQueen (dbm@cs.uchicago.edu)
1276    Date: 2003/08/13 11:36:00 CDT
1277    Tag: dbm-20030813-mcz-merge1
1278    Description:
1279    
1280    Merging changes from the mcz-branch development branch into trunk.
1281    These changes involve replacement of the emulated old prettyprinter
1282    interface with direct use of the SML/NJ Lib PP library, and fixing
1283    of a couple of bugs (895, 1186) relating to error messages.  A new
1284    prettyprinter for ast datatypes (Elaborator/print/ppast.{sig,sml})
1285    has been added.
1286    
1287    ----------------------------------------------------------------------
1288    Name: Matthias Blume (blume (at) tti - c (dot) org)
1289    Date: 2003/08/11 15:45:00 CDT
1290    Tag: blume-20030811-windows
1291    Description:
1292    
1293    Version number bumped to 110.42.9.  NEW BOOTFILES!!!
1294    
1295        http://smlnj.cs.uchicago.edu/dist/working/110.42.9/
1296    
1297    This patch restores SML/NJ's ability to run under win32.  There are a
1298    number of changes, including fixes for several bugs that had gone
1299    unnoticed until now:
1300    
1301      - uname "CYGWIN_NT*" is recognized as win32 (This is relevant only when
1302        trying to run the win32 version from within cygwin.)
1303    
1304      - There are a number of simple .bat scripts that substitute for their
1305        corresponding Unix shell-scripts.  (See below.)
1306    
1307      - The internals of ml-build have been modified slightly.  The main
1308        difference is that instead of calling ".link-sml" (or link-sml.bat)
1309        using OS.Process.system, the ML process delegates this task back
1310        to the script.  Otherwise problems arise in mixed environments such
1311        as Cygwin where scripts look and work like Unix scripts, but
1312        where OS.Process.system cannot run them.
1313    
1314      - In CM, the srcpath pickler used native pathname syntax -- which
1315        is incorrect in the case of cross-compilation.  The new pickle format
1316        is independent of platform-specific naming conventions.
1317    
1318      - Path configuration files (such as lib/pathconfig) can now choose
1319        between native and standard syntax.  Placing a line of the form
1320    
1321            standard!
1322    
1323        into the file causes all subsequent paths to be interpreted using
1324        CM standard pathname syntax (= Unix conventions); a line
1325    
1326            native!
1327    
1328        switches back to native style.  This was needed so that
1329        path config files can be written portably, see src/system/pathconfig.
1330    
1331      - Runtime system:
1332    
1333          - win32-filesys.c:  get_file_time and set_file_time now
1334            access modification time, not creation time.
1335    
1336          - I/O code made aware of new array representation.
1337    
1338          - Bug fixes in X86.prim.masm.
1339    
1340       - src/system/makeml made aware of win32. (For use under cygwin
1341            and other Unix-environments for windows.)
1342    
1343       - In Basis, fixed off-by-one error in win32-io.sml (function vecF)
1344         which caused BinIO.inputAll to fail consistently.
1345    
1346    .bat scripts:
1347    
1348       Windows .bat scripts assume that SMLNJ_HOME is defined.
1349    
1350       - sml.bat, ml-yacc.bat, ml-lex.bat:  Driver scripts for standalone
1351         applications (sml, ml-yacc, ml-lex).
1352       - ml-build.bat: analogous to ml-build.
1353       - config\install.bat:  Analogous to config/install.sh.  This requires
1354         that SMLNJ_HOME is set and that Microsoft Visual C is ready to use.
1355         (nmake etc. must be on the path, and vcvars32 must have been run.)
1356         Moreover, sources for ml-lex and ml-yacc need to exist under src,
1357         and the bootfile hierarchy must have been unpacked under
1358         sml.boot.x86-win32.
1359         The script is very primitive and does a poor job at error checking.
1360         It only installs the base system, ml-lex, and ml-yacc.  No other
1361         libraries are being installed (i.e., you get only those that
1362         are part of the compiler.)
1363       - link-sml.bat:  analogous to .link-sml, but not currently used
1364    
1365    Unrelated bug fixes:
1366    
1367      - ml-nlffigen now exports structures ST_* corresponding to incomplete
1368        types.
1369      - Added getDevice to PP/src/pp-debug-fn.sml.  (Would not compile
1370        otherwise.)
1371    
1372    ----------------------------------------------------------------------
1373    Name: Dave MacQueen (macqueen@cs.uchicago.edu)
1374    Date: 2003/06/17
1375    Tag: macqueen-20030617-bug895
1376    Description:
1377    
1378    Modified compiler/Elaborator/print/pptype.sml to fix bug 895.
1379    Tag will be used for new development branch (mcz-branch) for
1380    use by MacQueen, (Lucasz) Zairek, and (George) Cao at uchicago.
1381    
1382    ----------------------------------------------------------------------
1383    Name: Matthias Blume (blume (at) tti - c (dot) org)
1384    Date: 2003/05/27 16:55:00 CDT
1385    Tag: blume-20030527-polyeq
1386    Description:
1387    
1388    Tried to eliminated most cases of polymorphic equality.
1389    
1390    ----------------------------------------------------------------------
1391    Name: Matthias Blume (blume (at) tti - c (dot) org)
1392    Date: 2003/05/21 17:45:00 CDT
1393    Tag: blume-20030517-complete
1394    Description:
1395    
1396    Two changes:
1397    
1398      1. Added a flag for controlling whether non-exhaustive bindings will
1399         be treated as errors (default is false).
1400      2. Cleaned up the *entire* source tree so that CMB.make goes through
1401         without a single non-exhaustive match- or bind warning.
1402    
1403    ----------------------------------------------------------------------
1404    Name: Matthias Blume (blume (at) tti - c (dot) org)
1405    Date: 2003/05/17 10:20:00 CDT
1406    Tag: blume-20030517-absyn
1407    Description:
1408    
1409    1. Added cases for IF, WHILE, ANDALSO, and ORELSE to Absyn.
1410    
1411       This mainly affects the quality of error messages.  However, some
1412       of the code is now more straightforward than before. (Treatment of
1413       the above four constructs in translate.sml is much simpler than
1414       the "macro-expansion" that was going on before.  Plus, the mach-
1415       compiler no longer gets invoked just to be able to compile an
1416       if-expression.)
1417    
1418    2. The ErrorMsg.Error exception is now caught and absorbed by the
1419       interactive loop.
1420    
1421    ----------------------------------------------------------------------
1422    Name: Allen Leung
1423    Date: 2003/05/16 13:05:00 CDT
1424    Tag: leunga-20030516-cygwin-runtime
1425    Description:
1426    
1427        Ported the runtime system to cygwin, which uses the unix
1428    x86-unix bin files.  Missing/buggy features:
1429    
1430       o getnetbyname, getnetbyaddr: these functions seem to be missing in
1431         the Cygwin library.
1432       o Ctrl-C handling may be flaky.
1433       o Windows system calls and Windows I/O are not supported.
1434    
1435       A new set of binfiles is located at:
1436    
1437         http://www.dorsai.org/~leunga/boot.x86-unix.tgz
1438    
1439       This is only needed for bootstrapping the cygwin version of smlnj.
1440       Other x86 versions can use the existing binfiles.
1441    
1442    ----------------------------------------------------------------------
1443    Name: Matthias Blume
1444    Date: 2003/04/08 15:42:00 CDT
1445    Tag: blume-20030408-listpair
1446    Description:
1447    
1448    1. Added a target 'mlrisc' to installer.
1449    
1450    2. Added missing elements to structure ListPair.
1451    
1452    ----------------------------------------------------------------------
1453    Name: Allen Leung
1454    Date: 2003/01/07 10:40:00 EST
1455    Tag: leunga-20030107-int-rem
1456    Description:
1457    
1458    Fixed a bug in Int.rem(x,y) where y is a power of 2 on x86.
1459    The arguments to the SUBL instruction were swapped.
1460    
1461    ----------------------------------------------------------------------
1462    Name: Matthias Blume
1463    Date: 2002/12/12 16:25:00 EST
1464    Tag: blume-20021212-risc-ra
1465    Description:
1466    
1467    Fixed a serious bug in the rewrite code for FP spilling/reloading that
1468    sent the RA into an infinite loop when floating point registers get
1469    spilled.  (Because of this bug, e.g., nucleic stopped compiling between
1470    110.37 and 110.38.)
1471    There was another set of potential problems related to the handling of
1472    MLRISC annotations (but those did not yet cause real problems, apparently).
1473    
1474    ----------------------------------------------------------------------
1475    Name: Matthias Blume
1476    Date: 2002/12/06 22:40:00 EST
1477    Tag: blume-20021206-cm-fileid
1478    Description:
1479    
1480    Added a call of SrcPath.sync at the beginning of Parse.parse (in CM).
1481    This fixes the problem of CM getting confused by files that suddenly
1482    change their identity (e.g., by getting unlinked and recreated by some
1483    text editor such as vi).  There might be a better/cheaper/cleaner way
1484    of doing this, but for now this will have to do.
1485    
1486    ----------------------------------------------------------------------
1487    Name: Matthias Blume
1488    Date: 2002/10/28 09:50:00 EST
1489    Tag: blume-20021028-typecheck
1490    Description:
1491    
1492    Exported structure Typecheck from $smlnj/viscomp/core.cm.
1493    
1494    ----------------------------------------------------------------------
1495    Name: Matthias Blume
1496    Date: 2002/10/17 09:10:00 EDT
1497    Tag: Release_110_42
1498    Description:
1499    
1500    In good old tradition, there has been a slight hiccup so that we have
1501    to patch 110.42 after the fact.   The old release tag has been replaced
1502    (see below).
1503    
1504    The change solves a problem with two competing approaches the
1505    configuration problem regarding MacOS 10.1 vs. MacOS 10.2 which got in
1506    each other's way.
1507    
1508    This change only affects the runtime system code and the installer script.
1509    (No new bootfiles.)
1510    
1511    ----------------------------------------------------------------------
1512    Name: Matthias Blume
1513    Date: 2002/10/16 12:00:00 EDT
1514    Tag: Release_110_42_removed
1515    Description:
1516    
1517    New working release.  New bootfiles.
1518    
1519    ----------------------------------------------------------------------
1520    Name: Matthias Blume
1521    Date: 2002/10/10 13:10:00 EDT
1522    Tag: blume-20021010-ppc-divs
1523    Description:
1524    
1525    The mltree operator DIVS must be implemented with an overflow check on
1526    the PPC because the hardware indicates divide-by-zero using "overflow" as
1527    well.
1528    
1529    ----------------------------------------------------------------------
1530    Name: Matthias Blume
1531    Date: 2002/07/23 11:45:00
1532    Tag: blume-20020723-smlnj-home
1533    Description:
1534    
1535    Sml now senses the SMLNJ_HOME environment variable.  If this is set,
1536    then the bin dir is assumed to be in $SMLNJ_HOME/bin and (unless
1537    CM_PATHCONFIG is also set), the path configuration file is assumed
1538    to be in $SMLNJ_HOME/lib/pathconfig.  This way one can easily move
1539    the entire tree to some other place and everything will "just work".
1540    
1541    (Companion commands such as ml-build and ml-makedepend also sense this
1542    variable.)
1543    
1544    ----------------------------------------------------------------------
1545    Name: Matthias Blume
1546    Date: 2002/07/12 21:19:00 EDT
1547    Tag: blume-20020712-liveness
1548    Description:
1549    
1550    Exported two useful "step" functions from liveness module (MLRISC).
1551    
1552    ----------------------------------------------------------------------
1553    Name: Matthias Blume
1554    Date: 2002/07/05 16:00 EDT
1555    Tag: Release_110_41
1556    Description:
1557    
1558    New working release.  New bootfiles.
1559    
1560    ----------------------------------------------------------------------
1561    Name: Matthias Blume
1562    Date: 2002/07/05 10:25:00 EDT
1563    Tag: blume-20020705-btimp
1564    Description:
1565    
1566    Exported structure BTImp from $smlnj/viscomp/debugprof.cm so that
1567    other clients can set up backtracing support.
1568    
1569    ----------------------------------------------------------------------
1570    Name: Matthias Blume
1571    Date: 2002/06/25 17:23:00 EDT
1572    Tag: blume-20020625-fpmax
1573    Description:
1574    
1575    Fixed a bug in translation of INLMAX (and INLMIN) for the floating-point
1576    case.  (The sense of the isNaN test was reversed -- which made min and
1577    max always return their first argument.)
1578    
1579    ----------------------------------------------------------------------
1580    Name: Matthias Blume
1581    Date: 2002/06/11
1582    Tag: blume-20020611-unixpath
1583    Description:
1584    
1585    Back-ported OS.Path.{from,to}UnixPath from idlbasis-devel branch.
1586    
1587    ----------------------------------------------------------------------
1588    Name: Matthias Blume
1589    Date: 2002/06/10 16:35:00 EDT
1590    Tag: blume-20020610-ieeereal
1591    Description:
1592    
1593    I back-ported my implementation of IEEEReal.fromString from the
1594    idlbasis-devel branch so that we can test it.
1595    
1596    Another small change is that ppDec tries to give more information
1597    than just "<sig>" in the case of functors.  However, this code is
1598    broken in some mysterious way if the functor's body's signature
1599    has not been declared by ascription but gets inferred from the
1600    implementation.  This needs fixing...
1601    
1602    ----------------------------------------------------------------------
1603    Name: Matthias Blume
1604    Date: 2002/05/31
1605    Tag: blume-20020531-btrace-mode
1606    Description:
1607    
1608    Resurrected SMLofNJ.Internals.BTrace.mode.  (It accidentally fell by
1609    the wayside when I switched over to using Controls everywhere.)
1610    
1611    ----------------------------------------------------------------------
1612    Name: Lal George
1613    Date: 2002/05/23 12:21:40 EDT
1614    Tag: george-20020523-visual-labels
1615    Description:
1616    
1617    Labels are now displayed in the graphical output to make
1618    the fall-through and target blocks obvious.
1619    
1620    ----------------------------------------------------------------------
1621    Name: Matthias Blume
1622    Date: 2002/05/22 11:03:00 EDT
1623    Tag: blume-20020522-shrink
1624    Description:
1625    
1626    John tweaked yesterday's fix for 1131 to handle an out-of-memory
1627    situation that comes up when allocating huge arrays.
1628    
1629    ----------------------------------------------------------------------
1630    Name: Matthias Blume
1631    Date: 2002/05/21 16:00:00 EDT
1632    Tag: Release_110_40
1633    Description:
1634    
1635    New working release (110.40).  New bootfiles.
1636    
1637    [Also: John Reppy fixed GC bug 1131.]
1638    
1639    ----------------------------------------------------------------------
1640    Name: Matthias Blume
1641    Date: 2002/05/21 12:35:00 EDT
1642    Tag: blume-20020521-cmdoc
1643    Description:
1644    
1645    CM documentation update.
1646    
1647    ----------------------------------------------------------------------
1648    Name: Matthias Blume
1649    Date: 2002/05/21 10:55:00 EDT
1650    Tag: blume-20020521-misc
1651    Description:
1652    
1653    - John tweaked runtime to be silent on heap export (except when
1654      GC messages are on).
1655    - I added a few more things (cross-compiling versions of CMB) to
1656      config/preloads (as suggestions).
1657    
1658    ----------------------------------------------------------------------
1659    Name: Matthias Blume
1660    Date: 2002/05/20 22:25:00 EDT
1661    Tag: blume-20020520-controls
1662    Description:
1663    
1664    - Added ControlUtil structure to control-lib.cm.
1665    - Use it throughout.
1666    - Used Controls facility to define MLRISC controls (as opposed to
1667      registering MLRISC control ref cells with Controls after the fact)
1668    - Fixed messed-up controls priorities.
1669    
1670    * Removed again all the stuff from config/preloads that one wouldn't
1671      be able to preload at the time the initial heap image is built.
1672      (Many libraries, e.g., CML, do not exist yet at this time.  The
1673       only libraries that can be preloaded via config/preloads are those
1674       that come bundled with the bootfiles.)
1675    
1676    ----------------------------------------------------------------------
1677    Name: Matthias Blume
1678    Date: 2002/05/20 10:59:00 EDT
1679    Tag: blume-20020520-preloads
1680    Description:
1681    
1682    Added a lot of commented-out suggestions for things to be included
1683    in config/preloads.
1684    
1685    ----------------------------------------------------------------------
1686    Name: Allen Leung
1687    Date: 2002/05/18 14:20:00 EDT
1688    Tag: leunga-20020518-mdl
1689    Description:
1690    
1691    o Made the mdl tool stuff compile and run again.
1692    o I've disabled all the stuff that depends on RTL specifications; they
1693      are all badly broken anyway.
1694    
1695    ----------------------------------------------------------------------
1696    Name: Matthias Blume
1697    Date: 2002/05/17 16:49:00 EDT
1698    Tag: blume-20020517-controls
1699    Description:
1700    
1701    0. John Reppy made several modifications to the SML/NJ library.
1702       In particular, there is a shiny new controls-lib.cm.
1703    
1704    1. Pushed new controls interface through compiler so that everything
1705       compiles again.
1706    
1707    2. Added FormatComb and FORMAT_COMB to the CML version of the
1708       SML/NJ library (so that CML compiles again).
1709    
1710    3. Modified init scripts because XXX_DEFAULT environment variables
1711       are no longer with us.  (Boot-time initialization is now done
1712       using the same environment variables that are also used for
1713       startup-time initialization of controls.)
1714    
1715    ----------------------------------------------------------------------
1716    Name: Lal George
1717    Date: 2002/05/15 09:20:10 EDT
1718    Tag: george-20020515-pseudo-op-decls
1719    Description:
1720    
1721            All pseudo-ops emitted before the first segment declaration
1722    such as TEXT, DATA, and BSS directives are assumed to be global
1723    declarations and are emitted first in the assembly file. This is
1724    useful in a number of situations where one has pseudo-ops that are not
1725    specific to any segment, and also works around the constraint that one
1726    cannot have client pseudo-ops in the TEXT segment.
1727    
1728            Because no segment is associated with these declarations it is
1729    an error to allocate any space or objects before the first segment
1730    directive and an exception will be raised. However, we cannot make
1731    this check for client pseudo-ops.
1732    
1733            These top level declarations are a field in the CFG graph_info.
1734    In theory you can continue to add to this field after the CFG has been
1735    built -- provided you know what you are doing;-)
1736    
1737    ----------------------------------------------------------------------
1738    Name: Matthias Blume
1739    Date: 2002/05/13 16:40:00 EDT
1740    Tag: blume-20020513-pp-etc
1741    Description:
1742    
1743    A few minor bugfixes:
1744    
1745      - Stopgap measure for bug recently reported by Elsa Gunter (ppDec).
1746        (Bogus printouts for redefined bindings still occur.  Compiler
1747        bug should no longer occur now.  We need to redo the prettyprinter
1748        from scratch.)
1749    
1750      - CM pathname printer now also adds escape sequences for ( and )
1751    
1752      - commend and docu fixes for ml-nlffi
1753    
1754    ----------------------------------------------------------------------
1755    Name: Matthias Blume
1756    Date: 2002/05/10 16:40:00 EDT
1757    Tag: blume-20020510-erg-textio
1758    Description:
1759    
1760    Applied the following bugfix provided by Emden Gansner:
1761    
1762        Output is corrupted when outputSubstr is used rather than output.
1763    
1764        The problem occurs when a substring
1765    
1766            ss = (s, dataStart, dataLen)
1767    
1768        where dataStart > 0, fills a stream buffer with avail bytes left.
1769        avail bytes of s, starting at index dataStart, are copied into the
1770        buffer, the buffer is flushed, and then the remaining dataLen-avail
1771        bytes of ss are copied into the beginning of the buffer. Instead of
1772        starting this copy at index dataStart+avail in s, the current code
1773        starts the copy at index avail.
1774    
1775        Fix:
1776        In text-io-fn.sml, change line 695 from
1777             val needsFlush = copyVec(v, avail, dataLen-avail, buf, 0)
1778        to
1779             val needsFlush = copyVec(v, dataStart+avail, dataLen-avail, buf, 0)
1780    
1781    ----------------------------------------------------------------------
1782    Name: Matthias Blume
1783    Date: 2002/04/12 13:55:00 EDT
1784    Tag: blume-20020412-assyntax
1785    Description:
1786    
1787    1. Grabbed newer assyntax.h from the XFree86 project.
1788    2. Fiddled with how to compile X86.prim.asm without warnings.
1789    3. (Very) Minor cleanup in CM.
1790    
1791    ----------------------------------------------------------------------
1792    Name: Matthias Blume
1793    Date: 2002/04/01 (no joke!) 17:07:00 EST
1794    Tag: blume-20020401-x86div
1795    Description:
1796    
1797    Added full support for div/mod/rem/quot on the x86, using the machine
1798    instruction's two results (without clumsily recomputing the remainder)
1799    directly where appropriate.
1800    
1801    Some more extensive power-of-two support was added to the x86 instruction
1802    selector (avoiding expensive divs, mods, and muls where they can be
1803    replaced with cheaper shifts and masks).  However, this sort of thing
1804    ought to be done earlier, e.g., within the CPS optimizer so that
1805    all architectures benefit from it.
1806    
1807    The compiler compiles to a fixed point, but changes might be somewhat
1808    fragile nevertheless.  Please, report any strange things that you might
1809    see wrt. div/mod/quot/rem...
1810    
1811    ----------------------------------------------------------------------
1812    Name: Matthias Blume
1813    Date: 2002/03/29 17:22:00
1814    Tag: blume-20020329-div
1815    Description:
1816    
1817    Fixed my broken div/mod logic.  Unfortunately, this means that the
1818    inline code for div/mod now has one more comparison than before.
1819    Fast paths (quotient > 0 or remainder = 0) are not affected, though.
1820    The problem was with quotient = 0, because that alone does not tell
1821    us which way the rounding went.  One then has to look at whether
1822    remainder and divisor have the same sign...  :(
1823    
1824    Anyway, I replaced the bootfiles with fresh ones...
1825    
1826    ----------------------------------------------------------------------
1827    Name: Matthias Blume
1828    Date: 2002/03/29 14:10:00 EST
1829    Tag: blume-20020329-inlprims
1830    Description:
1831    
1832    NEW BOOTFILES!!!    Version number bumped to 110.39.3.
1833    
1834    Primops have changed. This means that the bin/boot-file formats have
1835    changed as well.
1836    
1837    To make sure that there is no confusion, I made a new version.
1838    
1839    
1840    CHANGES:
1841    
1842    * removed REMT from mltree (remainder should never overflow).
1843    
1844    * added primops to deal with divisions of all flavors to the frontend
1845    
1846    * handled these primops all the way through so they map to their respective
1847      MLRISC support
1848    
1849    * used these primops in the implementation of Int, Int32, Word, Word32
1850    
1851    * removed INLDIV, INLMOD, and INLREM as they are no longer necessary
1852    
1853    * parameterized INLMIN, INLMAX, and INLABS by a numkind
1854    
1855    * translate.sml now deals with all flavors of INL{MIN,MAX,ABS}, including
1856      floating point
1857    
1858    * used INL{MIN,MAX,ABS} in the implementation of Int, Int32, Word, Word32,
1859      and Real (but Real.abs maps to a separate floating-point-only primop)
1860    
1861    
1862    TODO items:
1863    
1864    * Hacked Alpha32 instruction selection, disabling the selection of REMx
1865      instructions because the machine instruction encoder cannot handle
1866      them.  (Hppa, PPC, and Sparc instruction selection did not handle
1867      REM in the first place, and REM is supported by the x86 machine coder.)
1868    
1869    * Handle DIV and MOD with DIV_TO_NEGINF directly in the x86 instruction
1870      selection phase.  (The two can be streamlined because the hardware
1871      delivers both quotient and remainder at the same time anyway.)
1872    
1873    * Think about what to do with "valOf(Int32.minInt) div ~1" and friends.
1874      (Currently the behavior is inconsistent both across architectures and
1875      wrt. the draft Basis spec.)
1876    
1877    * Word8 should eventually be handled natively, too.
1878    
1879    * There seems to be one serious bug in mltree-gen.sml.  It appears, though,
1880      as if there currently is no execution path that could trigger it in
1881      SML/NJ.  (The assumptions underlying functions arith and promotable do not
1882      hold for things like multiplication and division.)
1883    
1884    ----------------------------------------------------------------------
1885    Name: Matthias Blume
1886    Date: 2002/03/27 16:27:00 EST
1887    Tag: blume-20020327-mlrisc-divisions
1888    Description:
1889    
1890    Added support for all four division operations (ML's div, mod, quot,
1891    and rem) to MLRISC.  In the course of doing so, I also rationalized
1892    the naming (no more annoying switch-around of DIV and QUOT), by
1893    parameterizing the operation by div_rounding_mode (which can be either
1894    DIV_TO_ZERO or DIV_TO_NEGINF).
1895    
1896    The generic MLTreeGen functor takes care of compiling all four
1897    operations down to only round-to-zero div.
1898    
1899    Missing pieces:
1900    
1901      * Doing something smarter than relying on MLTreeGen on architectures
1902        like, e.g., the x86 where hardware division delivers both quotient and
1903        remainder at the same time.  With this, the implementation of the
1904        round-to-neginf operations could be further streamlined.
1905    
1906      * Remove inlining support for div/mod/rem from the frontend and replace it
1907        with primops that get carried through to the backend.  Do this for all
1908        int and word types.
1909    
1910    ----------------------------------------------------------------------
1911    Name: Matthias Blume
1912    Date: 2002/03/25 17:25:00 EST
1913    Tag: blume-20020325-divmod
1914    Description:
1915    
1916    I improved (hopefully without breaking them) the implementation of Int.div,
1917    Int.mod, and Int.rem.   For this, the code in translate.sml now takes
1918    advantage of the following observations:
1919    
1920      Let  q = x quot y      r = x rem y
1921           d = x div  y      m = x mod y
1922    
1923    where "quot" is the round-to-zero version of integer division that
1924    hardware usually provides.  Then we have:
1925    
1926         r = x - q * y        where neither the * nor the - will overflow
1927         d = if q >= 0 orelse x = q * y then q else q - 1
1928                              where neither the * nor the - will overflow
1929         m = if q >= 0 orelse r = 0 then r else r + y
1930                              where the + will not overflow
1931    
1932    This results in substantial simplification of the generated code.
1933    The following table shows the number of CFG nodes and edges generated
1934    for
1935            fun f (x, y) = x OPER y
1936            (* with OPER \in div, mod, quot, rem *)
1937    
1938    
1939        OPER | nodes(old) | edges(old) | nodes(new) | edges(new)
1940        --------------------------------------------------------
1941         div |         24 |         39 |         12 |         16
1942         mod |         41 |         71 |         12 |         16
1943        quot |          8 |         10 |          8 |         10
1944         rem |         10 |         14 |          8 |         10
1945    
1946    
1947    ----------------------------------------------------------------------
1948    Name: Matthias Blume
1949    Date: 2002/03/25 22:06:00 EST
1950    Tag: blume-20020325-cprotobug
1951    Description:
1952    
1953    Fixed a bug in cproto (c prototype decoder).
1954    
1955    ----------------------------------------------------------------------
1956    Name: Matthias Blume
1957    Date: 2002/03/25 16:00:00 EST
1958    Tag: blume-20020325-raw-primops
1959    Description:
1960    
1961    I did some cleanup to Allen's new primop code and
1962    replaced yesterday's bootfiles with new ones.
1963    (But they are stored in the same place.)
1964    
1965    ----------------------------------------------------------------------
1966    Name: Matthias Blume
1967    Date: 2002/03/24 22:40:00 EST
1968    Tag: blume-20020324-bootfiles
1969    Description:
1970    
1971    Made the bootfiles that Allen asked for.
1972    
1973    ----------------------------------------------------------------------
1974    Name: Allen Leung
1975    Date: 2002/03/23 15:50:00 EST
1976    Tag: leunga-20020323-flint-cps-rcc-primops
1977    Description:
1978    
1979      1. Changes to FLINT primops:
1980    
1981        (* make a call to a C-function;
1982         * The primop carries C function prototype information and specifies
1983         * which of its (ML-) arguments are floating point. C prototype
1984         * information is for use by the backend, ML information is for
1985         * use by the CPS converter. *)
1986      | RAW_CCALL of { c_proto: CTypes.c_proto,
1987                       ml_args: ccall_type list,
1988                       ml_res_opt: ccall_type option,
1989                       reentrant : bool
1990                     } option
1991       (* Allocate uninitialized storage on the heap.
1992        * The record is meant to hold short-lived C objects, i.e., they
1993        * are not ML pointers.  With the tag, the representation is
1994        * the same as RECORD with tag tag_raw32 (sz=4), or tag_fblock (sz=8)
1995        *)
1996      | RAW_RECORD of {tag:bool,sz:int}
1997      and ccall_type = CCALL_INT32 | CCALL_REAL64 | CCALL_ML_PTR
1998    
1999      2.  These CPS primops are now overloaded:
2000    
2001           rawload of {kind:numkind}
2002           rawstore of {kind:numkind}
2003    
2004          The one argument form is:
2005    
2006             rawload {kind} address
2007    
2008          The two argument form is:
2009    
2010             rawload {kind} [ml object, byte-offset]
2011    
2012      3. RAW_CCALL/RCC now takes two extra arguments:
2013    
2014         a. The first is whether the C call is reentrant, i.e., whether
2015            ML state should be saved and restored.
2016         b. The second argument is a string argument specifying the name of
2017            library and the C function.
2018    
2019         These things are currently not handled in the code generator, yet.
2020    
2021      4. In CProto,
2022    
2023         An encoding type of "bool" means "ml object" and is mapped into
2024         C prototype of PTR.  Note that "bool" is different than "string",
2025         even though "string" is also mapped into PTR, because "bool"
2026         is assigned an CPS type of BOGt, while "string" is assigned INT32t.
2027    
2028      5. Pickler/unpicker
2029    
2030         Changed to handle RAW_RECORD and newest RAW_CCALL
2031    
2032      6. MLRiscGen,
2033    
2034         1. Changed to handle the new rawload/rawstore/rawrecord operators.
2035         2. Code for handling C Calls has been moved to a new module CPSCCalls,
2036            in the file CodeGen/cpscompile/cps-c-calls.sml
2037    
2038      7. Added the conditional move operator
2039    
2040             condmove of branch
2041    
2042         to cps.  Generation of this is still buggy so it is currently
2043         disabled.
2044    
2045    ----------------------------------------------------------------------
2046    Name: Lal George
2047    Date: 2002/03/22 14:18:25 EST
2048    Tag: george-20020322-cps-branch-prob
2049    Description:
2050    
2051    Implemented the Ball-Larus branch prediction-heuristics, and
2052    incorporated graphical viewers for control flow graphs.
2053    
2054    Ball-Larus Heuristics:
2055    ---------------------
2056    See the file compiler/CodeGen/cpscompile/cpsBranchProb.sml.
2057    
2058    By design it uses the Dempster-Shafer theory for combining
2059    probabilities.  For example, in the function:
2060    
2061        fun f(n,acc) = if n = 0 then acc else f(n-1, n*acc)
2062    
2063    the ball-larus heuristics predicts that the n=0 is unlikely
2064    (OH-heuristic), and the 'then' branch is unlikely because of the
2065    RH-heuristic -- giving the 'then' branch an even lower combined
2066    probability using the Dempster-Shafer theory.
2067    
2068    Finally, John Reppy's loop analysis in MLRISC, further lowers the
2069    probability of the 'then' branch because of the loop in the else
2070    branch.
2071    
2072    
2073    Graphical Viewing:
2074    ------------------
2075    I merely plugged in Allen's graphical viewers into the compiler. The
2076    additional code is not much. At the top level, saying:
2077    
2078            Control.MLRISC.getFlag "cfg-graphical-view" := true;
2079    
2080    will display the graphical view of the control flow graph just before
2081    back-patching.  daVinci must be in your path for this to work. If
2082    daVinci is not available, then the default viewer can be changed
2083    using:
2084    
2085            Control.MLRISC.getString "viewer"
2086    
2087    which can be set to "dot" or "vcg" for the corresponding viewers. Of
2088    course, these viewers must be in your path.
2089    
2090    The above will display the compilation unit at the level of clusters,
2091    many of which are small, boring, and un-interesting. Also setting:
2092    
2093            Control.MLRISC.getInt "cfg-graphical-view_size"
2094    
2095    will display clusters that are larger than the value set by the above.
2096    
2097    
2098    ----------------------------------------------------------------------
2099    Name: Matthias Blume
2100    Date: 2002/03/21 22:20:00 EST
2101    Tag: blume-20020321-kmp-bugfix
2102    Description:
2103    
2104    Changed the interface to the KMP routine in PreString and fixed
2105    a minor bug in one place where it was used.
2106    
2107    ----------------------------------------------------------------------
2108    Name: Allen Leung
2109    Date: 2002/03/21 20:30:00 EST
2110    Tag: leunga-20020321-cfg
2111    Description:
2112    
2113      Fixed a potential problem in cfg edge splitting.
2114    
2115    ----------------------------------------------------------------------
2116    Name: Allen Leung
2117    Date: 2002/03/21 17:15:00 EST
2118    Tag: leunga-20020321-x86-fp-cfg
2119    Description:
2120    
2121      1. Recoded the buggy parts of x86-fp.
2122    
2123         a. All the block reordering code has been removed.
2124            We now depend on the block placement phases to do this work.
2125    
2126         b. Critical edge splitting code has been simplified and moved into the
2127            CFG modules, as where they belong.
2128    
2129         Both of these were quite buggy and complex.  The code is now much, much
2130         simpler.
2131    
2132      2. X86 backend.
2133    
2134         a. Added instructions for 64-bit support.  Instruction selection for
2135            64-bit has not been committed, however, since that
2136            requires changes to MLTREE which haven't been approved by
2137            Lal and John.
2138    
2139         b. Added support for FUCOMI and FUCOMIP when generating code for
2140            PentiumPro and above.  We only generate these instructions in
2141            the fast-fp mode.
2142    
2143         c. Added cases for JP and JNP in X86FreqProps.
2144    
2145      3. CFG
2146    
2147         CFG now has a bunch of methods for edge splitting and merging.
2148    
2149      4. Machine description.
2150    
2151         John's simplification of MLTREE_BASIS.fcond broke a few machine
2152         description things:
2153    
2154         rtl-build.{sig,sml} and hppa.mdl fixed.
2155    
2156         NOTE: the machine description stuff in the repository is still broken.
2157               Again, I can't put my fixes in because that involves
2158               changes to MLTREE.
2159    
2160    ----------------------------------------------------------------------
2161    Name: Matthias Blume
2162    Date: 2002/03/20 15:55:00 EST
2163    Tag: blume-20020320-kmp
2164    Description:
2165    
2166    Implemented Knuth-Morris-Pratt string matching in PreString and used
2167    it for String.isSubstring, Substring.isSubstring, and
2168    Substring.position.
2169    
2170    (Might need some stress-testing.  Simple examples worked fine.)
2171    
2172    ----------------------------------------------------------------------
2173    Name: Matthias Blume
2174    Date: 2002/03/19 16:37:00 EST
2175    Tag: blume-20020319-witnesses
2176    Description:
2177    
2178    Added a structure C.W and functions convert/Ptr.convert to ml-nlffi-lib.
2179    
2180    This implements a generic mechanism for changing constness qualifiers
2181    anywhere within big C types without resorting to outright "casts".
2182    (So far, functions such as C.rw/C.ro or C.Ptr.rw/C.Ptr.ro only let you
2183    modify the constness at the outermost level.)
2184    The implementation of "convert" is based on the idea of "witness"
2185    values -- values that are not used by the operation but whose types
2186    "testify" to their applicability.  On the implementation side, "convert"
2187    is simply a projection (returning its second curried argument).  With
2188    cross-module inlining, it should not result in any machine code being
2189    generated.
2190    
2191    ----------------------------------------------------------------------
2192    Name: Matthias Blume
2193    Date: 2002/03/15 16:40:00 EST
2194    Tag: blume-20020315-basis
2195    Description:
2196    
2197    Provided (preliminary?) implementations for
2198    
2199      {String,Substring}.{concatWith,isSuffix,isSubstring}
2200    
2201    and
2202    
2203      Substring.full
2204    
2205    Those are in the Basis spec but they were missing in SML/NJ.
2206    
2207    ----------------------------------------------------------------------
2208    Name: Matthias Blume
2209    Date: 2002/03/14 21:30:00 EST
2210    Tag: blume-20020314-controls
2211    Description:
2212    
2213    Controls:
2214    ---------
2215    
2216    1. Factored out the recently-added Controls : CONTROLS stuff and put
2217       it into its own library $/controls-lib.cm.  The source tree for
2218       this is under src/smlnj-lib/Controls.
2219    
2220    2. Changed the names of types and functions in this interface, so they
2221       make a bit more "sense":
2222    
2223          module -> registry
2224          'a registry -> 'a group
2225    
2226    3. The interface now deals in ref cells only.  The getter/setter interface
2227       is (mostly) gone.
2228    
2229    4. Added a function that lets one register an already-existing ref cell.
2230    
2231    5. Made the corresponding modifications to the rest of the code so that
2232       everything compiles again.
2233    
2234    6. Changed the implementation of Controls.MLRISC back to something closer
2235       to the original.  In particular, this module (and therefore MLRISC)
2236       does not depend on Controls.  There now is some link-time code in
2237       int-sys.sml that registers the MLRISC controls with the Controls
2238       module.
2239    
2240    CM:
2241    ---
2242    
2243      * One can now specify the lambda-split aggressiveness in init.cmi.
2244    
2245    ----------------------------------------------------------------------
2246    Name: Allen Leung
2247    Date: 2002/03/13 17:30:00 EST
2248    Tag: leunga-20020313-x86-fp-unary
2249    Description:
2250    
2251    Bug fix for:
2252    
2253    > leunga@weaselbane:~/Yale/tmp/sml-dist{21} bin/sml
2254    > Standard ML of New Jersey v110.39.1 [FLINT v1.5], March 08, 2002
2255    > - fun f(x,(y,z)) = Real.~ y;
2256    > [autoloading]
2257    > [autoloading done]
2258    >       fchsl   (%eax), 184(%esp)
2259    > Error: MLRisc bug: X86MCEmitter.emitInstr
2260    >
2261    > uncaught exception Error
2262    >   raised at: ../MLRISC/control/mlriscErrormsg.sml:16.14-16.19
2263    
2264    The problem was that the code generator did not generate any fp registers
2265    in this case, and the ra didn't know that it needed to run the X86FP phase to
2266    translate the pseudo fp instruction.   This only happened with unary fp
2267    operators in certain situations.
2268    
2269    ----------------------------------------------------------------------
2270    Name: Matthias Blume
2271    Date: 2002/03/13 14:00:00 EST
2272    Tag: blume-20020313-overload-etc
2273    Description:
2274    
2275    1. Added _overload as a synonym for overload for backward compatibility.
2276       (Control.overloadKW must be true for either version to be accepted.)
2277    
2278    2. Fixed bug in install script that caused more things to be installed
2279       than what was requested in config/targets.
2280    
2281    3. Made CM aware of the (_)overload construct so that autoloading
2282       works.
2283    
2284    ----------------------------------------------------------------------
2285    Name: Matthias Blume
2286    Date: 2002/03/12 22:03:00 EST
2287    Tag: blume-20020312-url
2288    Description:
2289    
2290    Forgot to update BOOT and srcarchiveurl.
2291    
2292    ----------------------------------------------------------------------
2293    Name: Matthias Blume
2294    Date: 2002/03/12 17:30:00 EST
2295    Tag: blume-20020312-version110392
2296    Description:
2297    
2298    Yet another version number bump (because of small changes to the
2299    binfile format).  Version number is now 110.39.2.  NEW BOOTFILES!
2300    
2301    Changes:
2302    
2303      The new pid generation scheme described a few weeks ago was overly
2304      complicated.  I implemented a new mechanism that is simpler and
2305      provides a bit more "stability":  Once CM has seen a compilation
2306      unit, it keeps its identity constant (as long as you do not delete
2307      those crucial CM/GUID/* files).  This means that when you change
2308      an interface, compile, then go back to the old interface, and
2309      compile again, you arrive at the original pid.
2310    
2311      There now also is a mechanism that instructs CM to use the plain
2312      environment hash as a module's pid (effectively making its GUID
2313      the empty string).  For this, "noguid" must be specified as an
2314      option to the .sml file in question within its .cm file.
2315      This is most useful for code that is being generated by tools such
2316      as ml-nlffigen (because during development programmers tend to
2317      erase the tool's entire output directory tree including CM's cached
2318      GUIDs).  "noguid" is somewhat dangerous (since it can be used to locally
2319      revert to the old, broken behavior of SML/NJ, but in specific cases
2320      where there is no danger of interface confusion, its use is ok
2321      (I think).
2322    
2323      ml-nlffigen by default generates "noguid" annotations.  They can be
2324      turned off by specifying -guid in its command line.
2325    
2326    ----------------------------------------------------------------------
2327    Name: Lal George
2328    Date: 2002/03/12 12 14:42:36 EST
2329    Tag: george-20020312-frequency-computation
2330    Description:
2331    
2332    Integrated jump chaining and static block frequency into the
2333    compiler. More details and numbers later.
2334    
2335    ----------------------------------------------------------------------
2336    Name: Lal George
2337    Date: 2002/03/11 11 22:38:53 EST
2338    Tag: george-20020311-jump-chain-elim
2339    Description:
2340    
2341    Tested the jump chain elimination on all architectures (except the
2342    hppa).  This is on by default right now and is profitable for the
2343    alpha and x86, however, it may not be profitable for the sparc and ppc
2344    when compiling the compiler.
2345    
2346    The gc test will typically jump to a label at the end of the cluster,
2347    where there is another jump to an external cluster containing the actual
2348    code to invoke gc. This is to allow factoring of common gc invocation
2349    sequences. That is to say, we generate:
2350    
2351            f:
2352               testgc
2353               ja   L1      % jump if above to L1
2354    
2355            L1:
2356               jmp L2
2357    
2358    
2359    After jump chain elimination the 'ja L1' instructions is converted to
2360    'ja L2'. On the sparc and ppc, many of the 'ja L2' instructions may end
2361    up being implemented in their long form (if L2 is far away) using:
2362    
2363            jbe     L3      % jump if below or equal to L3
2364            jmp     L2
2365         L3:
2366            ...
2367    
2368    
2369    For large compilation units L2  may be far away.
2370    
2371    
2372    ----------------------------------------------------------------------
2373    Name: Matthias Blume
2374    Date: 2002/03/11 13:30:00 EST
2375    Tag: blume-20020311-mltreeeval
2376    Description:
2377    
2378    A functor parameter was missing.
2379    
2380    ----------------------------------------------------------------------
2381    Name: Allen Leung
2382    Date: 2002/03/11 10:30:00 EST
2383    Tag: leunga-20020311-runtime-string0
2384    Description:
2385    
2386       The representation of the empty string now points to a
2387    legal null terminated C string instead of unit.  It is now possible
2388    to convert an ML string into C string with InlineT.CharVector.getData.
2389    This compiles into one single machine instruction.
2390    
2391    ----------------------------------------------------------------------
2392    Name: Allen Leung
2393    Date: 2002/03/10 23:55:00 EST
2394    Tag: leunga-20020310-x86-call
2395    Description:
2396    
2397       Added machine generation for CALL instruction (relative displacement mode)
2398    
2399    ----------------------------------------------------------------------
2400    Name: Matthias Blume
2401    Date: 2002/03/08 16:05:00
2402    Tag: blume-20020308-entrypoints
2403    Description:
2404    
2405    Version number bumped to 110.39.1.  NEW BOOTFILES!
2406    
2407    Entrypoints: non-zero offset into a code object where execution should begin.
2408    
2409    - Added the notion of an entrypoint to CodeObj.
2410    - Added reading/writing of entrypoint info to Binfile.
2411    - Made runtime system bootloader aware of entrypoints.
2412    - Use the address of the label of the first function given to mlriscGen
2413      as the entrypoint.  This address is currently always 0, but it will
2414      not be 0 once we turn on block placement.
2415    - Removed the linkage cluster code (which was The Other Way(tm) of dealing
2416      with entry points) from mlriscGen.
2417    
2418    ----------------------------------------------------------------------
2419    Name: Allen Leung
2420    Date: 2002/03/07 20:45:00 EST
2421    Tag: leunga-20020307-x86-cmov
2422    Description:
2423    
2424       Bug fixes for CMOVcc on x86.
2425    
2426       1. Added machine code generation for CMOVcc
2427       2. CMOVcc is now generated in preference over SETcc on PentiumPro or above.
2428       3. CMOVcc cannot have an immediate operand as argument.
2429    
2430    ----------------------------------------------------------------------
2431    Name: Matthias Blume
2432    Date: 2002/03/07 16:15:00 EST
2433    Tag: blume-20020307-controls
2434    Description:
2435    
2436    This is a very large but mostly boring patch which makes (almost)
2437    every tuneable compiler knob (i.e., pretty much everything under
2438    Control.* plus a few other things) configurable via both the command
2439    line and environment variables in the style CM did its configuration
2440    until now.
2441    
2442    Try starting sml with '-h' (or, if you are brave, '-H')
2443    
2444    To this end, I added a structure Controls : CONTROLS to smlnj-lib.cm which
2445    implements the underlying generic mechanism.
2446    
2447    The interface to some of the existing such facilities has changed somewhat.
2448    For example, the MLRiscControl module now provides mkFoo instead of getFoo.
2449    (The getFoo interface is still there for backward-compatibility, but its
2450    use is deprecated.)
2451    
2452    The ml-build script passes -Cxxx=yyy command-line arguments through so
2453    that one can now twiddle the compiler settings when using this "batch"
2454    compiler.
2455    
2456    TODO items:
2457    
2458    We should go through and throw out all controls that are no longer
2459    connected to anything.  Moreover, we should go through and provide
2460    meaningful (and correct!) documentation strings for those controls
2461    that still are connected.
2462    
2463    Currently, multiple calls to Controls.new are accepted (only the first
2464    has any effect).  Eventually we should make sure that every control
2465    is being made (via Controls.new) exactly once.  Future access can then
2466    be done using Controls.acc.
2467    
2468    Finally, it would probably be a good idea to use the getter-setter
2469    interface to controls rather than ref cells.  For the time being, both
2470    styles are provided by the Controls module, but getter-setter pairs are
2471    better if thread-safety is of any concern because they can be wrapped.
2472    
2473    *****************************************
2474    
2475    One bug fix: The function blockPlacement in three of the MLRISC
2476    backpatch files used to be hard-wired to one of two possibilities at
2477    link time (according to the value of the placementFlag).  But (I
2478    think) it should rather sense the flag every time.
2479    
2480    *****************************************
2481    
2482    Other assorted changes (by other people who did not supply a HISTORY entry):
2483    
2484    1. the cross-module inliner now works much better (Monnier)
2485    2. representation of weights, frequencies, and probabilities in MLRISC
2486       changed in preparation of using those for weighted block placement
2487       (Reppy, George)
2488    
2489    ----------------------------------------------------------------------
2490    Name: Lal George
2491    Date: 2002/03/07 14:44:24 EST 2002
2492    Tag: george-20020307-weighted-block-placement
2493    
2494    Tested the weighted block placement optimization on all architectures
2495    (except the hppa) using AMPL to generate the block and edge frequencies.
2496    Changes were required in the machine properties to correctly
2497    categorize trap instructions. There is an MLRISC flag
2498    "weighted-block-placement" that can be used to enable weighted block
2499    placement, but this will be ineffective without block/edge
2500    frequencies (coming soon).
2501    
2502    
2503    ----------------------------------------------------------------------
2504    Name: Lal George
2505    Date: 2002/03/05 17:24:48 EST
2506    Tag: george-20020305-linkage-cluster
2507    
2508    In order to support the block placement optimization, a new cluster
2509    is generated as the very first cluster (called the linkage cluster).
2510    It contains a single jump to the 'real' entry point for the compilation
2511    unit. Block placement has no effect on the linkage cluster itself, but
2512    all the other clusters  have full freedom in the manner in which they
2513    reorder blocks or functions.
2514    
2515    On the x86 the typical linkage code that is generated is:
2516       ----------------------
2517            .align 2
2518       L0:
2519            addl    $L1-L0, 72(%esp)
2520            jmp     L1
2521    
2522    
2523            .align  2
2524       L1:
2525       ----------------------
2526    
2527    72(%esp) is the memory location for the stdlink register. This
2528    must contain the address of the CPS function being called. In the
2529    above example, it contains the address of  L0; before
2530    calling L1 (the real entry point for the compilation unit), it
2531    must contain the address for L1, and hence
2532    
2533            addl $L1-L0, 72(%esp)
2534    
2535    I have tested this on all architectures except the hppa.The increase
2536    in code size is of course negligible
2537    
2538    ----------------------------------------------------------------------
2539    Name: Allen Leung
2540    Date: 2002/03/03 13:20:00 EST
2541    Tag: leunga-20020303-mlrisc-tools
2542    
2543      Added #[ ... ] expressions to mlrisc tools
2544    
2545    ----------------------------------------------------------------------
2546    Name: Matthias Blume
2547    Date: 2002/02/27 12:29:00 EST
2548    Tag: blume-20020227-cdebug
2549    Description:
2550    
2551    - made types in structure C and C_Debug to be equal
2552    - got rid of code duplication (c-int.sml vs. c-int-debug.sml)
2553    - there no longer is a C_Int_Debug (C_Debug is directly derived from C)
2554    
2555    ----------------------------------------------------------------------
2556    Name: Matthias Blume
2557    Date: 2002/02/26 12:00:00 EST
2558    Tag: blume-20020226-ffi
2559    Description:
2560    
2561    1. Fixed a minor bug in CM's "noweb" tool:
2562       If numbering is turned off, then truly don't number (i.e., do not
2563       supply the -L option to noweb).  The previous behavior was to supply
2564       -L'' -- which caused noweb to use the "default" line numbering scheme.
2565       Thanks to Chris Richards for pointing this out (and supplying the fix).
2566    
2567    2. Once again, I reworked some aspects of the FFI:
2568    
2569       A. The incomplete/complete type business:
2570    
2571       - Signatures POINTER_TO_INCOMPLETE_TYPE and accompanying functors are
2572         gone!
2573       - ML types representing an incomplete type are now *equal* to
2574         ML types representing their corresponding complete types (just like
2575         in C).  This is still safe because ml-nlffigen will not generate
2576         RTTI for incomplete types, nor will it generate functions that
2577         require access to such RTTI.   But when ML code generated from both
2578         incomplete and complete versions of the C type meet, the ML types
2579         are trivially interoperable.
2580    
2581         NOTE:  These changes restore the full generality of the translation
2582         (which was previously lost when I eliminated functorization)!
2583    
2584       B. Enum types:
2585    
2586       - Structure C now has a type constructor "enum" that is similar to
2587         how the "su" constructor works.  However, "enum" is not a phantom
2588         type because each "T enum" has values (and is isomorphic to
2589         MLRep.Signed.int).
2590       - There are generic access operations for enum objects (using
2591         MLRep.Signed.int).
2592       - ml-nlffigen will generate a structure E_foo for each "enum foo".
2593         * The structure contains the definition of type "mlrep" (the ML-side
2594         representation type of the enum).  Normally, mlrep is the same
2595         as "MLRep.Signed.int", but if ml-nlffigen was invoked with "-ec",
2596         then mlrep will be defined as a datatype -- thus facilitating
2597         pattern matching on mlrep values.
2598         ("-ec" will be suppressed if there are duplicate values in an
2599          enumeration.)
2600         * Constructors ("-ec") or values (no "-ec") e_xxx of type mlrep
2601         will be generated for each C enum constant xxx.
2602         * Conversion functions m2i and i2m convert between mlrep and
2603         MLRep.Signed.int.  (Without "-ec", these functions are identities.)
2604         * Coversion functions c and ml convert between mlrep and "tag enum".
2605         * Access functions (get/set) fetch and store mlrep values.
2606       - By default (unless ml-nlffigen was invoked with "-nocollect"), unnamed
2607         enumerations are merged into one single enumeration represented by
2608         structure E_'.
2609    
2610    ----------------------------------------------------------------------
2611    Name: Allen Leung
2612    Date: 2002/02/25 04:45:00 EST
2613    Tag: leunga-20020225-cps-spill
2614    
2615    This is a new implementation of the CPS spill phase.
2616    The new phase is in the new file compiler/CodeGen/cpscompile/spill-new.sml
2617    In case of problems, replace it with the old file spill.sml
2618    
2619    The current compiler runs into some serious performance problems when
2620    constructing a large record.  This can happen when we try to compile a
2621    structure with many items.  Even a very simple structure like the following
2622    makes the compiler slow down.
2623    
2624        structure Foo = struct
2625           val x_1 = 0w1 : Word32.int
2626           val x_2 = 0w2 : Word32.int
2627           val x_3 = 0w3 : Word32.int
2628           ...
2629           val x_N = 0wN : Word32.int
2630        end
2631    
2632    The following table shows the compile time, from N=1000 to N=4000,
2633    with the old compiler:
2634    
2635    N
2636    1000   CPS 100 spill                           0.04u  0.00s  0.00g
2637           MLRISC ra                               0.06u  0.00s  0.05g
2638              (spills = 0 reloads = 0)
2639           TOTAL                                   0.63u  0.07s  0.21g
2640    
2641    1100   CPS 100 spill                           8.25u  0.32s  0.64g
2642           MLRISC ra                               5.68u  0.59s  3.93g
2643              (spills = 0 reloads = 0)
2644           TOTAL                                   14.71u  0.99s  4.81g
2645    
2646    1500   CPS 100 spill                           58.55u  2.34s  1.74g
2647           MLRISC ra                               5.54u  0.65s  3.91g
2648              (spills = 543 reloads = 1082)
2649           TOTAL                                   65.40u  3.13s  6.00g
2650    
2651    2000   CPS 100 spill                           126.69u  4.84s  3.08g
2652           MLRISC ra                               0.80u  0.10s  0.55g
2653              (spills = 42 reloads = 84)
2654           TOTAL                                   129.42u  5.10s  4.13g
2655    
2656    3000   CPS 100 spill                           675.59u  19.03s  11.64g
2657           MLRISC ra                               2.69u  0.27s  1.38g
2658              (spills = 62 reloads = 124)
2659           TOTAL                                   682.48u  19.61s  13.99g
2660    
2661    4000   CPS 100 spill                           2362.82u  56.28s  43.60g
2662           MLRISC ra                               4.96u  0.27s  2.72g
2663              (spills = 85 reloads = 170)
2664           TOTAL                                   2375.26u  57.21s  48.00g
2665    
2666    As you can see the old cps spill module suffers from some serious
2667    performance problem.  But since I cannot decipher the old code fully,
2668    instead of patching the problems up, I'm reimplementing it
2669    with a different algorithm.  The new code is more modular,
2670    smaller when compiled, and substantially faster
2671    (O(n log n) time and O(n) space).  Timing of the new spill module:
2672    
2673    4000  CPS 100 spill                           0.02u  0.00s  0.00g
2674          MLRISC ra                               0.25u  0.02s  0.15g
2675             (spills=1 reloads=3)
2676          TOTAL                                   7.74u  0.34s  1.62g
2677    
2678    Implementation details:
2679    
2680    As far as I can tell, the purpose of the CPS spill module is to make sure the
2681    number of live variables at any program point (the bandwidth)
2682    does not exceed a certain limit, which is determined by the
2683    size of the spill area.
2684    
2685    When the bandwidth is too large, we decrease the register pressure by
2686    packing live variables into spill records.  How we achieve this is
2687    completely different than what we did in the old code.
2688    
2689    First, there is something about the MLRiscGen code generator
2690    that we should be aware of:
2691    
2692    o MLRiscGen performs code motion!
2693    
2694       In particular, it will move floating point computations and
2695       address computations involving only the heap pointer to
2696       their use sites (if there is only a single use).
2697       What this means is that if we have a CPS record construction
2698       statement
2699    
2700           RECORD(k,vl,w,e)
2701    
2702       we should never count the new record address w as live if w
2703       has only one use (which is often the case).
2704    
2705       We should do something similar to floating point, but the transformation
2706       there is much more complex, so I won't deal with that.
2707    
2708    Secondly, there are now two new cps primops at our disposal:
2709    
2710     1. rawrecord of record_kind option
2711        This pure operator allocates some uninitialized storage from the heap.
2712        There are two forms:
2713    
2714         rawrecord NONE [INT n]  allocates a tagless record of length n
2715         rawrecord (SOME rk) [INT n] allocates a tagged record of length n
2716                                     and initializes the tag.
2717    
2718     2. rawupdate of cty
2719          rawupdate cty (v,i,x)
2720          Assigns to x to the ith component of record v.
2721          The storelist is not updated.
2722    
2723    We use these new primops for both spilling and increment record construction.
2724    
2725     1. Spilling.
2726    
2727        This is implemented with a linear scan algorithm (but generalized
2728        to trees).  The algorithm will create a single spill record at the
2729        beginning of the cps function and use rawupdate to spill to it,
2730        and SELECT or SELp to reload from it.  So both spills and reloads
2731        are fine-grain operations.  In contrast, in the old algorithm
2732        "spills" have to be bundled together in records.
2733    
2734        Ideally, we should sink the spill record construction to where
2735        it is needed.  We can even split the spill record into multiple ones
2736        at the places where they are needed.  But CPS is not a good
2737        representation for global code motion, so I'll keep it simple and
2738        am not attempting this.
2739    
2740     2. Incremental record construction (aka record splitting).
2741    
2742        Long records with many component values which are simulatenously live
2743        (recall that single use record addresses are not considered to
2744         be live) are constructed with rawrecord and rawupdate.
2745        We allocate space on the heap with rawrecord first, then gradually
2746        fill it in with rawupdate.  This is the technique suggested to me
2747        by Matthias.
2748    
2749        Some restrictions on when this is applicable:
2750        1. It is not a VECTOR record.  The code generator currently does not handle
2751           this case. VECTOR record uses double indirection like arrays.
2752        2. All the record component values are defined in the same "basic block"
2753           as the record constructor.  This is to prevent speculative
2754           record construction.
2755    
2756    ----------------------------------------------------------------------
2757    Name: Allen Leung
2758    Date: 2002/02/22 01:02:00 EST
2759    Tag: leunga-20020222-mlrisc-tools
2760    
2761    Minor bug fixes in the parser and rewriter
2762    
2763    ----------------------------------------------------------------------
2764    Name: Allen Leung
2765    Date: 2002/02/21 20:20:00 EST
2766    Tag: leunga-20020221-peephole
2767    
2768    Regenerated the peephole files.  Some contained typos in the specification
2769    and some didn't compile because of pretty printing bugs in the old version
2770    of 'nowhere'.
2771    
2772    ----------------------------------------------------------------------
2773    Name: Allen Leung
2774    Date: 2002/02/19 20:20:00 EST
2775    Tag: leunga-20020219-mlrisc-tools
2776    Description:
2777    
2778       Minor bug fixes to the mlrisc-tools library:
2779    
2780       1.  Fixed up parsing colon suffixed keywords
2781       2.  Added the ability to shut the error messages up
2782       3.  Reimplemented the pretty printer and fixed up/improved
2783           the pretty printing of handle and -> types.
2784       4.  Fixed up generation of literal symbols in the nowhere tool.
2785       5.  Added some SML keywords to to sml.sty
2786    
2787    ----------------------------------------------------------------------
2788    Name: Matthias Blume
2789    Date: 2002/02/19 16:20:00 EST
2790    Tag: blume-20020219-cmffi
2791    Description:
2792    
2793    A wild mix of changes, some minor, some major:
2794    
2795    * All C FFI-related libraries are now anchored under $c:
2796        $/c.cm      --> $c/c.cm
2797        $/c-int.cm  --> $c/internals/c-int.cm
2798        $/memory.cm --> $c/memory/memory.cm
2799    
2800    * "make" tool (in CM) now treats its argument pathname slightly
2801      differently:
2802        1. If the native expansion is an absolute name, then before invoking
2803           the "make" command on it, CM will apply OS.Path.mkRelative
2804           (with relativeTo = OS.FileSys.getDir()) to it.
2805        2. The argument will be passed through to subsequent phases of CM
2806           processing without "going native".  In particular, if the argument
2807           was an anchored path, then "make" will not lose track of that anchor.
2808    
2809    * Compiler backends now "know" their respective C calling conventions
2810      instead of having to be told about it by ml-nlffigen.  This relieves
2811      ml-nlffigen from one of its burdens.
2812    
2813    * The X86Backend has been split into X86CCallBackend and X86StdCallBackend.
2814    
2815    * Export C_DEBUG and C_Debug from $c/c.cm.
2816    
2817    * C type encoding in ml-nlffi-lib has been improved to model the conceptual
2818      subtyping relationship between incomplete pointers and their complete
2819      counterparts.  For this, ('t, 'c) ptr has been changed to 'o ptr --
2820      with the convention of instantiating 'o with ('t, 'c) obj whenever
2821      the pointer target type is complete.  In the incomplete case, 'o
2822      will be instantiated with some "'c iobj" -- a type obtained by
2823      using one of the functors PointerToIncompleteType or PointerToCompleteType.
2824    
2825      Operations that work on both incomplete and complete pointer types are
2826      typed as taking an 'o ptr while operations that require the target to
2827      be known are typed as taking some ('t, 'c) obj ptr.
2828    
2829      voidptr is now a bit "more concrete", namely "type voidptr = void ptr'"
2830      where void is an eqtype without any values.  This makes it possible
2831      to work on voidptr values using functions meant to operate on light
2832      incomplete pointers.
2833    
2834    * As a result of the above, signature POINTER_TO_INCOMPLETE_TYPE has
2835      been vastly simplified.
2836    
2837    ----------------------------------------------------------------------
2838    Name: Matthias Blume
2839    Date: 2002/02/19 10:48:00 EST
2840    Tag: blume-20020219-pqfix
2841    Description:
2842    
2843    Applied Chris Okasaki's bug fix for priority queues.
2844    
2845    ----------------------------------------------------------------------
2846    Name: Matthias Blume
2847    Date: 2002/02/15 17:05:00
2848    Tag: Release_110_39
2849    Description:
2850    
2851    Last-minute retagging is becoming a tradition... :-(
2852    
2853    This is the working release 110.39.
2854    
2855    ----------------------------------------------------------------------
2856    Name: Matthias Blume
2857    Date: 2002/02/15 16:00:00 EST
2858    Tag: Release_110_39-orig
2859    Description:
2860    
2861    Working release 110.39.  New bootfiles.
2862    
2863    (Update: There was a small bug in the installer so it wouldn't work
2864    with all shells.  So I retagged. -Matthias)
2865    
2866    ----------------------------------------------------------------------
2867    Name: Matthias Blume
2868    Date: 2002/02/15 14:17:00 EST
2869    Tag: blume-20020215-showbindings
2870    Description:
2871    
2872    Added EnvRef.listBoundSymbols and CM.State.showBindings.  Especially
2873    the latter can be useful for exploring what bindings are available at
2874    the interactive prompt.  (The first function returns only the list
2875    of symbols that are really bound, the second prints those but also the
2876    ones that CM's autoloading mechanism knows about.)
2877    
2878    ----------------------------------------------------------------------
2879    Name: Matthias Blume
2880    Date: 2002/02/15 12:08:00 EST
2881    Tag: blume-20020215-iptrs
2882    Description:
2883    
2884    Two improvements to ml-nlffigen:
2885    
2886      1. Write files only if they do not exist or if their current contents
2887         do not coincide with what's being written.  (That is, avoid messing
2888         with the time stamps unless absolutely necessary.)
2889    
2890      2. Implement a "repository" mechanism for generated files related
2891         to "incomplete pointer types".   See the README file for details.
2892    
2893    ----------------------------------------------------------------------
2894  Name: Matthias Blume  Name: Matthias Blume
2895  Date: 2002/02/14 11:50:00 EST  Date: 2002/02/14 11:50:00 EST
2896  Tag: blume-20020214-quote  Tag: blume-20020214-quote

Legend:
Removed from v.1066  
changed lines
  Added in v.1645

root@smlnj-gforge.cs.uchicago.edu
ViewVC Help
Powered by ViewVC 1.0.0