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/branches/primop-branch-3/NOTES/HISTORY
ViewVC logotype

Diff of /sml/branches/primop-branch-3/NOTES/HISTORY

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

revision 890, Thu Jul 19 20:38:56 2001 UTC revision 1644, Sat Oct 9 03:50:36 2004 UTC
# Line 8  Line 8 
8  The form of an entry should be:  The form of an entry should be:
9    
10  Name:  Name:
11  Date:  Date: yyyy/mm/dd
12  Tag: <post-commit CVS tag>  Tag: <post-commit CVS tag>
13  Description:  Description:
14    
15  ----------------------------------------------------------------------  ----------------------------------------------------------------------
16    Name: Matthias Blume (blume (at) tti - c (dot) org)
17    Date: 2004/10/08 22:50:00 CDT
18    Tag: blume-20041008-cmkw
19    Description:
20    
21    Fixed handling of keywords in .cm files: After seeing "is" the lexer
22    treats subsequent occurrences of "group", "library", "source", "is",
23    "*", and "-" as ordinary identifiers rather than keywords.
24    
25    Most seriously, this fixes a problem with CM's "shell" tool.  The tool
26    is supposed to accept a tool argument called "source", but this did
27    not work because of the clash with the keyword.
28    
29    ----------------------------------------------------------------------
30    Name: Matthias Blume (blume (at) tti - c (dot) org)
31    Date: 2004/10/07 16:00:00 CDT
32    Tag: blume-20041007-cleanup
33    Description:
34    
35    Assorted cleanup work:
36    
37      - got rid of intstrmap in favor of using the library's
38        hash table implementation
39      - threw out most of the pathnames stuff, as it was not used anyway
40      - simplified tokentable implementation
41      - fixed some minor spelling errors
42    
43    ----------------------------------------------------------------------
44    Name: Matthias Blume (blume (at) tti - c (dot) org)
45    Date: 2004/10/06 15:15:15 CDT
46    Tag: blume-20041006-handler
47    Description:
48    
49    Cleaned up the absyn to reflect the invariant that HANDLE always
50    carries a FNexp as part of the type definition.  This eliminates some
51    superfluous sanity checks at runtime down the road.
52    
53    Some minor cleanup of the btrace code.
54    
55    ----------------------------------------------------------------------
56    Name: Matthias Blume (blume (at) tti - c (dot) org)
57    Date: 2004/10/01 10:20:30 CDT
58    Tag: blume-20041001-slave
59    Description:
60    
61    Added hack to make slave mode work in the presence of the version
62    tool.  (Still, since the master does two passes over the code for
63    CMB.make, the release number gets bumped twice when slaves are
64    attached.  I don't know if this is worth fixing...)
65    
66    ----------------------------------------------------------------------
67    Name: Matthias Blume (blume (at) tti - c (dot) org)
68    Date: 2004/09/30 10:55:00 CDT
69    Tag: blume-20040930-version
70    Description:
71    
72    * Moved the "version" magic into its own little library under
73    src/system/smlnj/internal.  This avoids expensive reconstruction
74    of a stable src/compiler/core.cm.
75    
76    * At the same time, structure CompilerVersion is now known as structure
77    SMLNJVersion.
78    
79    * Arranged for the version tool to NOT kick in when rebuilding the system
80      (makeml -rebuild, fixpt).  Otherwise one would never reach a fixpoint.
81      Also, loading the versiontool does not work when rebuilding the system
82      because CM is not properly initialized at that time.
83    
84    ----------------------------------------------------------------------
85    Name: Matthias Blume (blume (at) tti - c (dot) org)
86    Date: 2004/09/29 14:00:00 CDT
87    Tag: blume-20040929-autoversion
88    Description:
89    
90    Implemented some CM magic to have
91        file src/compiler/TopLevel/main/version.sml
92    generated automagically.
93    The version is taken from two files: config/version and config/release.
94    The first is expected to contain a two-part version number such as 110.49.
95    The second should contain a single number, but it may be missing.
96    
97    If the environment variable VERSIONTOOL_BUMP_RELEASE is defined at the
98    time the version tool is loaded (which is the first time you say CMB.make),
99    then the tool will increment the value stored in config/release every
100    time CMB.make is invoked.
101    
102    The binfile format is now insensitive to anything beyond the first
103    two components of a version number, so bumping the release does not render
104    binfiles incompatible.  Auto-bumping can be used to keep track of versions
105    during development without invalidating existing binfiles.
106    
107    In any case, every CMB.make updates the date information in version.sml.
108    (This is the date that is printed in the banner.)
109    
110    ----------------------------------------------------------------------
111    Name: Matthias Blume (blume (at) tti - c (dot) org)
112    Date: 2004/09/28 10:53:00 CDT
113    Tag: blume-20040928-controls
114    Description:
115    
116    Some cleanup of the controls code.
117    
118    ----------------------------------------------------------------------
119    Name: Matthias Blume (blume (at) tti - c (dot) org)
120    Date: 2004/09/27 22:08:00 CDT
121    Tag: blume-20040927-controls
122    Description:
123    
124    Added two pieces of functionality to the Controls interface:
125    
126    1.   val save'restore: 'a control -> unit -> unit
127    
128       grabs the current value of the control in stage 1 and restores it
129       in stage 2.
130    
131    2.   val set' : 'a control * 'a -> unit -> unit
132    
133       stores the given value into the control in stage 2 (i.e., delayed)
134       but does all error checking in stage 1.
135       (This is for string controls that need to do parse their argument --
136        something that might fail.  In some cases, notably in CM, one
137        already knows the intended argument but wants to delay the actual
138        assignment until a time when error recovery would be more difficult.)
139    
140    Changed the handling of controls in tool arguments to classes "sml" and
141    "lazysml":
142       - use Controls.save'restore as a more robust way of restoring the
143         old value (in particular: without having to re-parse the string)
144       - use controls to handle the "overload" keyword in the init group
145         (I believe this change actually fixes a long-standing obscure bug.)
146    
147    ----------------------------------------------------------------------
148    Name: Matthias Blume (blume (at) tti - c (dot) org)
149    Date: 2004/09/27 17:00:00 CDT
150    Tag: blume-20040927-lazysml
151    Description:
152    
153    Added a new tool class called "lazysml" to CM's tool chest.  The only
154    difference to "sml" is that compilation is done with Control.lazysml
155    set to true.  A source of class "lazysml" is automatically recognized
156    by a file name suffix of ".lml".
157    
158    In addition to the above feature, the original class "sml" now also
159    supports a tool argument "lazy" which has the same effect.  As a
160    result, the following three lines are equivalent:
161    
162        foo.sml : lazysml
163        foo.sml : sml (lazy)
164        foo.sml (lazy)
165    
166    The setting goes into effect both during parsing and during
167    compilation.  The original setting is restored right after parsing and
168    after compilation, respectively.
169    
170    In addition to all the above, there is also a general mechanism to set
171    ANY of the "controls" that are available at the command line via
172    "-C..." on a per-sml-file basis.  The same rules that apply for "lazy"
173    apply as well. (In fact, "lazy" is implemented as a special case of
174    the general mechanism.)
175    
176    The .cm file syntax uses a new keyword tool argument called "with".
177    There are several ways of indicating the desired settings:
178    
179        foo.sml (with:parser.quotations=true)
180        foo.sml (with:(name:parser.quotations value:true))
181        foo.sml (with:(name:name1 value:value1 name:name2 value:value2 ...))
182        foo.sml (with:(name1=value1 name2=value2 ...))
183        foo.sml (with:(name1=value1 name:name2 value:value2 name3=value3 ...))
184    
185        etc.
186    
187    Another possible abbreviation is to leave out the =v or value:v part
188    if the name refers to a boolean control (in which case the value is
189    taken to be true).  Thus, one could get lazy sml also by saying:
190    
191        foo.sml (with:parser.lazy-keyword=true)
192        foo.sml (with:parser.lazy-keyword)
193        foo.sml (with:(name:parser.lazy-keyword value:true))
194        foo.sml (with:(name:parser.lazy-keyword))
195    
196    ----------------------------------------------------------------------
197    Name: Matthias Blume (blume (at) tti - c (dot) org)
198    Date: 2004/09/24 16:22:00 CDT
199    Tag: blume-20040924-ppc-long-branch
200    Description:
201    
202    Turned message about "emiting long form of branch" off by default.
203    Added a control flag to turn it back on when desired.
204    
205    ----------------------------------------------------------------------
206    Name: Matthias Blume (blume (at) tti - c (dot) org)
207    Date: 2004/09/24 16:05:00 CDT
208    Tag: blume-20040924-rounding
209    Description:
210    
211    Applied patch for setting rounding modes under Mac OS X.  Thanks to
212    Melissa O'Neill for providing the code!
213    
214    ----------------------------------------------------------------------
215    Name: Matthias Blume (blume (at) tti - c (dot) org)
216    Date: 2004/09/23 17:30:00 CDT
217    Tag: blume-20040923-envvars
218    Description:
219    
220    1. Changed definition of type ControlRegistry.registry_tree to
221       include control_info (i.e., the name of the controlling environment
222       variable).
223    
224    2. Added command-line flags -e and -E to print the names of environment
225       variables that can be used to control internal settings.
226       (This uses the new API mentioned in 1.)
227    
228    ----------------------------------------------------------------------
229    Name: Matthias Blume (blume (at) tti - c (dot) org)
230    Date: 2004/09/13 16:50:00 CDT
231    Tag: Release_110_49
232    Description:
233    
234    New working version (110.49).  NEW BOOTFILES!
235    
236    ----------------------------------------------------------------------
237    Name: Matthias Blume
238    Date: 2004/09/13 16:20:00 CDT
239    Tag: blume-20040913-config-mlrisc
240    Description:
241    
242    Put target "mlrisc" back into the default list.
243    (There is no harm in having it, and some users have expressed their
244    wish to have "mlrisc" included by default.)
245    
246    ----------------------------------------------------------------------
247    Name: John Reppy
248    Date: 2004/09/13
249    Tag: jhr-20040913-signals
250    Description:
251    
252    Fixed the signal masking code to properly nest mask/unmask operations
253    on a per-signal basis.
254    
255    ----------------------------------------------------------------------
256    Name: Matthias Blume (blume (at) tti - c (dot) org)
257    Date: 2004/09/08 13:20:00 CDT
258    Tag: blume-20040908-heap-magic
259    Description:
260    
261    Bumped the heap macig number to 0x09082004 to account for the changed
262    layout of the ML frame under MacOS X.
263    
264    ----------------------------------------------------------------------
265    Name: Allen Leung (leunga (at) reservoir (dot) com)
266    Date: 2004/09/03 11:26:00 EST
267    Tag: leunga-20040903-cygwin-install
268    Description:
269    
270       Added a patch to _arch-n-opsys to enable the Cygwin runtime.
271    The Cygwin runtime is turned on by setting the environment
272    variable SMLNJ_CYGWIN_RUNTIME to 1.
273    
274    ----------------------------------------------------------------------
275    Name: Matthias Blume (blume (at) tti - c (dot) org)
276    Date: 2004/08/31 17:15:00 CDT
277    Tag: blume-20040831-core
278    Description:
279    
280    Added some exports to src/compiler/core.cm upon request by J. Joemann.
281    
282    ----------------------------------------------------------------------
283    Name: Matthias Blume (blume (at) tti - c (dot) org)
284    Date: 2004/08/30 17:55:00 CDT
285    Tag: blume-20040830-installer
286    Description:
287    
288    Upon request by Johannes Joemann:
289    
290    - improved ML code of installer to fall back to coping when renaming
291      fails (i.e., when source and target are on different file systems);
292      the code compiles but has yet to be tested in anger
293    - removed mlrisc from list of default targets (config/targets)
294    
295    ----------------------------------------------------------------------
296    Name: Matthias Blume (blume (at) tti - c (dot) org)
297    Date: 2004/08/27 17:20:00 CDT
298    Tag: blume-20040827-ptreql
299    Description:
300    
301    Added ptreql primop to structure InlineT (upon request from Larry
302    Paulson).
303    
304    ----------------------------------------------------------------------
305    Name: Allen Leung (leunga (at) reservoir (dot) com)
306    Date: 2004/08/15 21:21:00 EST
307    Tag: leunga-110_48-udgraph
308    Description:
309    
310    Another bug fix from Carl Hauser:
311    
312    diff /net/niflab/smlnj48/src/MLRISC/graphs/udgraph.sml udgraph.sml
313    > 48c48
314    > <              | rmv((e as (k,_))::es,L) = rmv(es,if k = i then es else
315    > e::L)
316    > ---
317    > >              | rmv((e as (k,_))::es,L) = rmv(es,if k = i then L else e::L)
318    > Without this, any deletion of an edge in an undirected graph does severe
319    > violence to the graph.
320    
321    ----------------------------------------------------------------------
322    Name: Allen Leung (leunga (at) reservoir (dot) com)
323    Date: 2004/08/10 23:35:00 EST
324    Tag: leunga-110_48-ppc
325    Description:
326    
327       The IBM/MacOS syntax switch on PPC was incorrectly swapped.
328    
329    ----------------------------------------------------------------------
330    Name: Matthias Blume (blume (at) tti - c (dot) org)
331    Date: 2004/08/10 12:00:00 CDT
332    Tag: Release_110_48
333    Description:
334    
335    New working version (110.48).  NEW BOOTFILES!
336    
337    ----------------------------------------------------------------------
338    Name: Allen Leung (leunga (at) reservoir (dot) com)
339    Date: 2004/08/09 12:21:00 EST
340    Tag: leunga-110_47-dijsktra
341    Description:
342    
343    Bug fix from Carl Hauser:
344    
345    single_source_shortest_paths in dijkstra.sml was observed to get wrong
346    answers (by comparing to single_source_shortest_paths in bellman-ford.sml).
347    
348    The problem is that following the expression A.update(dist,s,Num.zero)
349    it is necessary to update the priority queue using Q.decreaseWeight(Q,s).
350    
351    ----------------------------------------------------------------------
352    Name: Matthias Blume (blume (at) tti - c (dot) org)
353    Date: 2004/08/06 18:10:00 CDT
354    Tag: blume-20040806-cmdline
355    Description:
356    
357    Fiddled with handling of command-line options:
358    
359      * sml now quits after processing the command line
360        if -H, -S, -h<n>, or -s<n> appears as the last
361        command-line argument
362      * a new option -q terminates the session when encountered on
363        the command line; subsequent arguments will be ignored
364      * bug fixes: short (erroneous) arguments are no longer ignored
365        completely
366    
367    ----------------------------------------------------------------------
368    Name: Allen Leung (leunga (at) reservoir (dot) com)
369    Date: 2004/08/04 18:17:00 EST
370    Tag: leunga-110_47-ppc-ibm-asm
371    Description:
372    
373      - Added minimal IBM assembly syntax support for PowerPC.
374    
375      - Cygwin: manually changed the file cygwin.def.  Some exported symbols have
376        been altered in the runtime.  We need an automatic way to keep the file
377        in sync.
378    
379    ----------------------------------------------------------------------
380    Name: Matthias Blume (blume (at) tti - c (dot) org)
381    Date: 2004/08/04 14:00:00 CDT
382    Tag: Release_110_47
383    Description:
384    
385    New working version (110.47).  NEW BOOTFILES!
386    
387    ----------------------------------------------------------------------
388    Name: Matthias Blume (blume (at) tti - c (dot) org)
389    Date: 2004/08/03 14:25:00 CDT
390    Tag: blume-20040803-callingconv
391    Description:
392    
393    Added low-level support for choosing C calling conventions by
394    twiddling the type of rawccall.  (See
395    src/compiler/Semant/types/cproto.sml for details.)
396    
397    ----------------------------------------------------------------------
398    Name: Matthias Blume (blume (at) tti - c (dot) org)
399    Date: 2004/08/02 15:55:00 CDT
400    Tag: blume-20040802-backout
401    Description:
402    
403    Backed out of change to win32-filesys.c.  The earlier patch to
404    get_file_time caused CM to produce files with the wrong time stamp.
405    
406    ----------------------------------------------------------------------
407    Name: Matthias Blume (blume (at) tti - c (dot) org)
408    Date: 2004/08/02 14:45:00 CDT
409    Tag: blume-20040802-nlffi-win32
410    Description:
411    
412    Added NLFFI support for Win32, adapted from a patch provided by David
413    Hansel.  This is currently completely untested.  Also, the issue
414    concerning stdcall vs. ccall is still unresolved.
415    
416    ----------------------------------------------------------------------
417    Name: Matthias Blume (blume (at) tti - c (dot) org)
418    Date: 2004/07/30 17:55:00 CDT
419    Tag: blume-20040730-various
420    Description:
421    
422    Gearing up towards 110.47...
423    
424    - various minor bugfixes to ml-nlffigen
425    - a beginning of a manual for nlffi
426    
427    - eliminated 'export name=value' in config/install.sh as this does
428      not work with certain versions of /bin/sh
429      (Thanks to David King at Motorola for catching this.)
430    
431    - several bugfixes provided or suggested by David Hansel at Reactive Systems:
432      - added a test for tm==NULL to gmtime.c and localtime.c
433      - applied patch for incorrect GetFileTime under win32
434      - toSeconds -> toMilliseconds in Win32/win32-process.sml
435    
436    ----------------------------------------------------------------------
437    Name: Matthias Blume (blume (at) tti - c (dot) org)
438    Date: 2004/07/21 18:20:00 CDT
439    Tag: blume-20040721-nlffigen
440    Description:
441    
442    - Fixed minor issue in ml-nlffigen: Now generate structure T_foo
443      for a typedef to an incomplete type, but leave out the "typ" member.
444      (This is just for consistency.)
445    - Started to produce what is supposed to become better (i.e., comprehensive)
446      documentation of what ml-nlffigen does and produces.
447    
448    ----------------------------------------------------------------------
449    Name: Matthias Blume (blume (at) tti - c (dot) org)
450    Date: 2004/07/14 16:25:00 CDT
451    Tag: blume-20040714-union
452    Description:
453    
454    Added C_UNION to c-calls/c-types.sml and updated the machinery
455    (ml-nlffigen, cproto.sml) that conveys C function interface
456    information to the code generator.
457    
458    However, the actual architecture-specific implementation of function
459    arguments and results that are C unions is still not implemented.
460    
461    ----------------------------------------------------------------------
462    Name: Allen Leung (leunga (at) reservoir (dot) com)
463    Date: 2004/07/14 14:38:00 EST
464    Tag: leunga-110_46_1-ppc-lwzu
465    Description:
466    
467        Added these instructions to the PowerPC architecture:
468        LBZU(X), LHZU(X), LWZU(X),
469        STWU(X), STFDU, STFSU
470    
471        etc...
472    
473        Note: I haven't added their instruction encoding into the description.
474    
475    ----------------------------------------------------------------------
476    Name: Allen Leung (leunga (at) reservoir (dot) com)
477    Date: 2004/07/13 15:04:00 EST
478    Tag: leunga-110_46_1-ppc-lwarx
479    Description:
480    
481       Added the two instructions LWARX and STWCX to the PowerPC
482    instruction set.
483    
484       A (untested) rewrite of loop-structure.sml.  The old version
485    is completely broken.
486    
487    ----------------------------------------------------------------------
488    Name: Matthias Blume (blume (at) tti - c (dot) org)
489    Date: 2004/07/13 13:50:00 CDT
490    Tag: blume-20040713-nlffi
491    Description:
492    
493    - use paramAlloc to report c-calls with too many arguments
494      (for PPC version where parameter area is pre-allocated)
495    - added ccall_maxargspace to machspec (to implement the above)
496    - made "make" commend in CM's "make" tool configurable
497    - added option (default: on) for passing the name of the SML/NJ's "bin"
498      directory to "make"; the call looks like this:
499    
500         make <options> SMLNJ_BINDIR=<dir> <target>
501    
502      This can be used by the Makefile to, e.g., pick the "right" version
503      of ml-nlffigen.
504    - minor code tweaks
505    
506    ----------------------------------------------------------------------
507    Name: Matthias Blume (blume (at) tti - c (dot) org)
508    Date: 2004/07/12 22:50:00 CDT
509    Tag: blume-110_46_1-macosx-nlffi
510    Description:
511    
512    NLFFI under Mac OS X now working (sort of).  This is largely untested,
513    though.
514    
515    Note:  1. You have to make a new, clean build of the runtime system.
516           2. There are new BOOTFILES, you have to use them!
517              (Doing the bootstrap process yourself would be *very* painful!
518               If you absolutely have to do it, build the system under
519               a different architecture and then cross-compile.)
520    
521    Version bumped to 110.46.1 to account for runtime data format changes.
522    
523    ----------------------------------------------------------------------
524    Name: Matthias Blume (blume (at) tti - c (dot) org)
525    Date: 2004/06/18 14:30:00 CDT
526    Tag: blume-20040618-unix
527    Description:
528    
529    Changed the implementation of structure Unix so that the same stream
530    is returned every time one of the {text,bin}{In,Out}streamOf functions
531    is invoked on the same proc.  This is not what the spec currently
532    says -- although IMO it arguably should.  (See discussion below.)
533    
534    ----------------------------------------------------------------------
535    Name: Matthias Blume (blume (at) tti - c (dot) org)
536    Date: 2004/06/17 18:15:00 CDT
537    Tag: Release_110_46
538    Description:
539    
540    New working version (110.46).  NEW BOOTFILES!
541    
542    ----------------------------------------------------------------------
543    Name: Matthias Blume (blume (at) tti - c (dot) org)
544    Date: 2004/06/17 17:20:00 CDT
545    Tag: blume-20040617-timer-unix
546    Description:
547    
548    Changed the interface of structures Timer and Unix to match the most
549    recent Basis spec.
550    
551    In the case of Unix there still seems to be an open/weird issue:
552    
553       The {text,bin}{In,Out}streamOf functions are supposed to create
554       fresh streams whenever they are called -- as opposed to have them
555       return the same stream every time.  This design is supposed to
556       prevent space leaks caused by proc values hanging on to streams.
557    
558       The reap function, on the other hand, is supposed to close the
559       streams.  This cannot be done without having a handle on the
560       stream in proc after all...
561    
562    I took the liberty to implement the following stopgap solution:
563    
564      The proc value hangs on to the most recently created stream(s).
565      Reap closes those.  If either or both of the two streams hadn't
566      been created at all yet, then reap will close the corresponding
567      file descriptors directly.
568    
569    PS: I don't understand the original space leak argument anymore.  If
570    a proc hangs on to the imperative stream, then I/O operations on those
571    will advance the state of the cached stream and avoid the space leak.
572    
573    ----------------------------------------------------------------------
574    Name: Matthias Blume (blume (at) tti - c (dot) org)
575    Date: 2004/05/28 16:45:00 CDT
576    Tag: blume-20040528-basis
577    Description:
578    
579    Added signature PACK_REAL and exported functor PrimIO.
580    
581    ----------------------------------------------------------------------
582    Name: Matthias Blume (blume (at) tti - c (dot) org)
583    Date: 2004/05/25 16:00:00 CDT
584    Tag: blume-20040525-group-owner
585    Description:
586    
587    CM now ignores (but still accepts) the "owner" information in group
588    descriptions.  The owner of a group is its next enclosing
589    library. Each group must have a unique owner.  (There is a virtual
590    "toplevel" library that owns groups which are not nested within a real
591    library.)  Previously, each group had to explicitly declare its owner,
592    and CM would check that such a declaration is correct.  The new scheme
593    is to have CM check that for each group there is precisely one owning
594    library.
595    
596    The advantage of the new scheme is that the programmer no longer needs
597    to maintain the somewhat annoying owner information.  The downside is
598    that CM cannot enforce the ownership rule across multiple runs of
599    CM.make.  Fortunately, enclosing the same group in two different
600    libraries A and B which are not part of the same program does not
601    cause real problems.
602    
603    ----------------------------------------------------------------------
604    Name: Matthias Blume (blume (at) tti - c (dot) org)
605    Date: 2004/05/20 16:00:00 CDT
606    Tag: blume-20040520-win32
607    Description:
608    
609    Made the win32 version work again.  (Strangely, a misplaced comma had
610    slipped into win32-process.c which prevented the runtime from being
611    compiled correctly.)
612    
613    Also, included a minor addition to ml-build.bat analogous to what was
614    done in blume-20040519-ml-build.
615    
616    ----------------------------------------------------------------------
617    Name: Matthias Blume (blume (at) tti - c (dot) org)
618    Date: 2004/05/19 22:10:00 CDT
619    Tag: blume-20040519-ml-build
620    Description:
621    
622    Arranged for ml-build to clean up after itself a little bit better.
623    The script generates a temporary SML source file and compiles it using
624    CM, so CM generates metadata (GUID, SKEL, objectfile) for it.  It now
625    gets rid of those at the end, so they don't accumulate under .cm.
626    
627    This required a minor change to install.sh because the name of the
628    metadata directory (default: .cm) is actually configurable at
629    installation time.
630    
631    ----------------------------------------------------------------------
632    Name: Matthias Blume (blume (at) tti - c (dot) org)
633    Date: 2004/05/18 15:50:00 CDT
634    Tag: blume-20040518-mkreader
635    Description:
636    
637    Added Posix.IO.mk{Bin,Text}{Reader,Writer} by lifting their respective
638    implementations from internal modules PosixBinPrimIO and PosixTextPrimIO.
639    
640    ----------------------------------------------------------------------
641    Name: Matthias Blume (blume (at) tti - c (dot) org)
642    Date: 2004/05/11 14:35:00 CDT
643    Tag: blume-20040511-win32sock
644    Description:
645    
646    Added previously missing support for many socket-related functions
647    under win32.  Thanks to David Hansel <hansel@reactive-systems.com>
648    for the voluminous patch!
649    
650    (I have not tested this patch under win32 yet.)
651    
652    Here is David's e-mail:
653    
654    Hi,
655    
656    Attached to this email you find a diff against sml/nj 110.45
657    that will enable socket support under Windows.
658    
659    To apply the patch (using unix or cygwin)
660    1) gunzip runtime.diff.gz
661    2) "cd" into "src/runtime" in the source tree of a fresh
662       110.45 installation.
663    3) patch -p 1 < [your/path/to]runtime.diff
664    
665    The code compiles fine but has NOT yet been extensively tested.
666    I only ran a few tests for basic socket client functionality
667    (which worked fine).  Especially the functions that use ioctl
668    are not tested at all and might not work (see below).
669    
670    I implemented this since we want to move to a newer version of sml/nj
671    but need socket support in order to use it.  This is the first time I
672    even had a look at the sml/nj source,  so please review my changes
673    before making this part of the distribution!  Here are a few issues
674    that I think might be better for someone to solve who is more
675    familiar with the sml/nj source (and socket programming):
676    
677    - getnetbyaddr.c and getnetbyname.c will raise a "not implemented"
678      exception since I could not figure out what the windows equivalent
679      of these functions is
680    
681    - In sockets-osdep.h there are a some #include statements that are
682      only used in a few files that include sockets-osdep.h
683    
684    - In smlnj-sock-lib.c, function init_fn() calls WSAStartup() but
685      does not process its return value since I don't know how to
686      report an error upwards.
687    
688    - It would probably be good to have a call to WSACleanup() when
689      the library is unloaded (if there is such a possibility).
690      Otherwise I think Windows will take care of this automatically
691      when the process finishes.
692    
693    - I used ioctlsocket() as a replacement for ioctl() but I have
694      no idea if that is actually the proper replacement on Windows.
695    
696    - All these issues are marked in the code by "FIXME" comments.
697    
698    We use sml/nj extensively in our products and are quite happy
699    with it.  I hope this contribution will help you.
700    
701    Keep up the good work!
702    
703    David
704    
705    ----------------------------------------------------------------------
706    Name: Matthias Blume (blume (at) tti - c (dot) org)
707    Date: 2004/05/11 14:20:00 CDT
708    Tag: blume-20040511-installml
709    Description:
710    
711    Fixed two bugs in installml script.  (Thanks to Vesa A. Norrman for
712    the patch.)
713    
714    ----------------------------------------------------------------------
715    Name: Matthias Blume (blume (at) tti - c (dot) org)
716    Date: 2004/05/11 14:05:00 CDT
717    Tag: blume-20040511-nlffi-netbsd
718    Description:
719    
720    Added support for nlffi under netbsd.  (Thanks to Vesa A. Norrman for
721    the patch.)
722    
723    ----------------------------------------------------------------------
724    Name: Matthias Blume (blume (at) tti - c (dot) org)
725    Date: 2004/05/11 12:05:00 CDT
726    Tag: blume-20040511-exports
727    Description:
728    
729    As per request by Adam Chlipala <adam@hcoop.net>, extended various
730    export lists in compiler-related .cm-files.
731    
732    ----------------------------------------------------------------------
733    Name: Matthias Blume (blume (at) tti - c (dot) org)
734    Date: 2004/05/11 11:35:00 CDT
735    Tag: blume-20040511-allsource
736    Description:
737    
738    The installer now honors the "src-smlnj" target again, although its meaning
739    has changed from "all sources required for the compiler" to "all sources
740    the installer knows about".  In other words, if you enable "src-smlnj"
741    in the "targets" file, then the installer will pull in sources for
742    everything.  (Notice that this refers to source code only.  Compiled
743    code is still only installed for modules that were requested explicitly
744    or which are required for other modules that were requested explicitly.)
745    
746    ----------------------------------------------------------------------
747    Name: Matthias Blume (blume (at) tti - c (dot) org)
748    Date: 2004/04/23 17:40:00 CDT
749    Tag: blume-20040423-ieee-scan
750    Description:
751    
752    Fixed IEEEReal.scan (and .fromString) so that if there is an overflow
753    in the exponent calculation we get INF or ZERO (depending on the mantissa
754    and the sign of the exponent).
755    
756    ----------------------------------------------------------------------
757    Name: Matthias Blume (blume (at) tti - c (dot) org)
758    Date: 2004/04/23 10:40:00 CDT
759    Tag: blume-20040423-ml-build
760    Description:
761    
762    The ml-build script now terminates with a non-0 status when something
763    goes wrong.
764    
765    ----------------------------------------------------------------------
766    Name: Matthias Blume (blume (at) tti - c (dot) org)
767    Date: 2004/04/22 16:35:00 CDT
768    Tag: blume-20040422-Option
769    Description:
770    
771    Made exception Option to be the same as exception Option.Option
772    (as it should be).
773    
774    ----------------------------------------------------------------------
775    Name: Allen Leung (leunga (at) reservoir (dot) com)
776    Date: 2004/03/19 14:40:00 EST
777    Tag: leunga-20040319-cygwin-nlffi
778    Description:
779    
780    Fixed the runtime so that ml-nlffi-lib runs on the cygwin version
781    of SML/NJ.  The problem is that
782    
783       lib = dlopen(NULL, ...)
784       f   = dlsym(lib, "malloc");
785    
786    does not work on Windows unless we explicitly export symbols
787    such as 'malloc' during linking.  We fixed this by explicitly
788    exporting the required symbols with the magic gcc incantation:
789    
790        -Wl,--export-all cygwin.def
791    
792    where cygwin.def is a file containing all the symbols that
793    we wish to export.
794    
795    I suspect this is a Windows problem and we'll have to
796    do the same (somehow with windows compilers) when
797    we build the native win32 version with the system
798    calls LoadLibrary/GetProcAddress.
799    
800    ----------------------------------------------------------------------
801    Name: Matthias Blume (blume (at) tti - c (dot) org)
802    Date: 2004/03/04 16:35:00 CST
803    Tag: blume-20040304-intinf-fmt
804    Description:
805    
806    Fixed problem with IntInf.fmt (sign would show up on the right instead
807    of on the left for BIN, OCT, and HEX).
808    
809    ----------------------------------------------------------------------
810    Name: Matthias Blume (blume (at) tti - c (dot) org)
811    Date: 2004/03/04 11:25:00 CST
812    Tag: blume-20040304-symlinks
813    Description:
814    
815    Fixed problem with installer script (unix only) where bin/ml-yacc and
816    friends pointed (via symlinks) to absolute locations instead of just
817    .run-sml.  This was reported by Vesa A Norrman.
818    
819    ----------------------------------------------------------------------
820    Name: Matthias Blume (blume (at) tti - c (dot) org)
821    Date: 2004/02/13 14:50:00 CST
822    Tag: Release_110_45
823    Description:
824    
825    New working version (110.45). New bootfiles.
826    
827    ----------------------------------------------------------------------
828    Name: Matthias Blume (blume (at) tti - c (dot) org)
829    Date: 2004/01/26 15:15:15 CST
830    Tag: blume-20040126-toplevel
831    Description:
832    
833    Improved handling of exceptions at the interactive toplevel.
834    
835    ----------------------------------------------------------------------
836    Name: Matthias Blume (blume (at) tti - c (dot) org)
837    Date: 2004/01/26 11:25:00
838    Tag: blume-20040126-app
839    Description:
840    
841    Type of top-level "app" corrected.
842    Added code for setting vp_limitPtrMask to Win32-specific runtime.
843    
844    ----------------------------------------------------------------------
845    Name: Matthias Blume (blume (at) tti - c (dot) org)
846    Date: 2003/11/18 17:10 CST
847    Tag: blume-20031118-basis-fiddle
848    Description:
849    
850    - changed Timer interface to what might become the spec
851    - POSIX_FLAGS -> BIT_FLAGS according to spec
852    - some other minor discrepancies wrt. spec eliminated
853    
854    ----------------------------------------------------------------------
855    Name: Matthias Blume (blume (at) tti - c (dot) org)
856    Date: 2003/11/06 12:00:00 CST
857    Tag: Release_110_44
858    Description:
859    
860    New working version (110.44). New bootfiles.
861    
862    ----------------------------------------------------------------------
863    Name: Matthias Blume (blume (at) tti - c (dot) org)
864    Date: 2003/11/04 11:50:00 CST
865    Tag: blume-20031104-move-libraries
866    Description:
867    
868    Eliminated the "dont_move_libraries" directive in config/targets.
869    (The mechanism was broken and could not be fixed easily.  Moreover,
870    there does not seem to be any reason not to move all libraries into
871    lib during installation.  I originally implemented this directive as a
872    backward-compatibility feature when I first introduced the new CM.
873    Now that things have been stable for a long time and going back to the
874    old CM is not an option, there is no reason to keep it around.)
875    
876    ----------------------------------------------------------------------
877    Name: Matthias Blume (blume (at) tti - c (dot) org)
878    Date: 2003/11/03 16:00:00 CST
879    Tag: blume-20031103-installdir
880    Description:
881    
882    Made installer honor INSTALLDIR variable again.  (Thanks to Chris
883    Richards for pointing out the problem and providing the solution.)
884    
885    ----------------------------------------------------------------------
886    Name: Matthias Blume (blume (at) tti - c (dot) org)
887    Date: 2003/10/01 17:05:00 CDT
888    Tag: blume-20031001-lal-mlrisc
889    Description:
890    
891    MLRISC bug fix from Lal.
892    
893    ----------------------------------------------------------------------
894    Name: Matthias Blume (blume (at) tti - c (dot) org)
895    Date: 2003/09/30 16:10:00 CDT
896    Tag: blume-20030930-primio-bat
897    Description:
898    
899    1. Added openVector, nullRd, and nullWr to PRIM_IO.
900    2. Improved .bat files (for Win32 port) to make things work under Win95.
901       (thanks to Aaron S. Hawley for this one)
902    
903    ----------------------------------------------------------------------
904    Name: Matthias Blume (blume (at) tti - c (dot) org)
905    Date: 2003/09/26 16:05:00 CDT
906    Tag: blume-20030926-wrappriv
907    Description:
908    
909    Added missing wrapper for privilege "primitive" in $smlnj/viscomp/core.cm.
910    
911    ----------------------------------------------------------------------
912    Name: Matthias Blume (blume (at) tti - c (dot) org)
913    Date: 2003/09/26 15:00:00 CDT
914    Tag: blume-20030926-110_43_3
915    Description:
916    
917    - additional cleanup
918    - version number bump, NEW BOOTFILES
919    
920    ----------------------------------------------------------------------
921    Name: Matthias Blume (blume (at) tti - c (dot) org)
922    Date: 2003/09/26 12:00:00 CDT
923    Tag: blume-20030926-ppautoload
924    Description:
925    
926    I modified the read-eval-print loop so that the autoloader gets
927    invoked whenever the prettyprinter tries to look up a symbol that
928    is not currently defined in the toplevel environment but which
929    appears in CM's autoload registry.  As a result, we see far fewer of
930    those ?.Foo.Bar.xxx names in the prettyprinter's output.
931    
932    In addition to this I tried to clean up some pieces of the Basis
933    implementation (e.g., Socket, Word8Array) in order to prevent other
934    instances of these ?.Foo.Bar.xxx names from being printed.
935    
936    The mechanism that picks names for types still needs some work, though.
937    (Right now it seems that if there is a type A.t which is defined to
938    be B.u, but B is unavailable at toplevel, then A.t gets printed as
939    "?.B.u" although the perhaps more sensible solution would be to use
940    "A.t" in this case.  In other words, the prettyprinter should follow
941    a chain of DEFtycs not farther than there are corresponding toplevel
942    names in the current environment.)
943    
944    ----------------------------------------------------------------------
945    Name: Matthias Blume (blume (at) tti - c (dot) org)
946    Date: 2003/09/24 16:31:00 CDT
947    Tag: blume-20030924-installer
948    Description:
949    
950    Another installer tweak:  All the ML code for the installer is now
951    compiled during CMB.make and put into a little library called
952    $smlnj/installer.cm.  The installation then simply invokes
953    
954       sml -m $smlnj/installer.cm
955    
956    and everything happens automagically.
957    
958    Win32:  ML code senses value of environment variable SMLNJ_HOME.
959    Unix:   ML code senses values of environment variables ROOT, CONFIGDIR,
960            and BINDIR.
961    
962    The new scheme guarantees that the ML code responsible for the installation
963    is in sync with the APIs of the main system.  Also, the installer is
964    somewhat faster because the installer script is precompiled.
965    
966    ----------------------------------------------------------------------
967    Name: Matthias Blume (blume (at) tti - c (dot) org)
968    Date: 2003/09/24 15:35:00 CDT
969    Tag: blume-20030924-synsock
970    Description:
971    
972    Added a signature SYNCHRONOUS_SOCKET to basis.cm.  This is like SOCKET
973    but excludes all non-blocking operations.  Defined SOCKET (in Basis)
974    and CML_SOCKET in terms of SYNCHRONOUS_SOCKET.  Removed superfluous
975    implementations of non-blocking operations from CML's Socket
976    structure.
977    
978    ----------------------------------------------------------------------
979    Name: Matthias Blume (blume (at) tti - c (dot) org)
980    Date: 2003/09/24 15:10:05 CDT
981    Tag: blume-20030924-sockets
982    Description:
983    
984    1. Fixed SOCKET API and implementation to match Basis spec.
985       This required changing the internal representation of sockets to one
986       that remembers (for each socket file descriptor) whether it is currently
987       blocking or non-blocking.  This state is maintained lazily (i.e., a system
988       call is made only if the state actually needs to change).
989    
990    2. OS-specific details of sockets were moved into separate files, thus
991       making it possible to unify the bulk of the socket implementations
992       between Unix and Win32.
993    
994    3. CML's socket API changed accordingly.
995       (Note that we need to remove non-blocking functions from this API
996       since they are redundant in the case of CML!)
997    
998    4. CML's socket implementation now makes use of non-blocking functions
999       provided by Basis, thus removing all OS-dependent code from this part
1000       of CML.
1001    
1002    5. Changed Real64.precision from 52 to 53.  Minor cleanup in Real64 code.
1003    
1004    ----------------------------------------------------------------------
1005    Name: Matthias Blume (blume (at) tti - c (dot) org)
1006    Date: 2003/09/22 12:10:00 CDT
1007    Tag: blume-20030922-110_43_2
1008    Description:
1009    
1010    Made a new interim version and bootfiles for developer's bootstrapping
1011    convenience.
1012    
1013       110.43.2 -- NEW BOOTFILES
1014    
1015    ----------------------------------------------------------------------
1016    Name: Matthias Blume (blume (at) tti - c (dot) org)
1017    Date: 2003/09/19 15:55:00 CDT
1018    Tag: blume-20030919-cmdir
1019    Description:
1020    
1021    1. new-install.sh -> install.sh
1022    2. changed default CM "metadata" directory name to ".cm" (instead of "CM")
1023    3. tweaked installer so that another name instead of .cm can be chosen
1024       at install time (by setting the CM_DIR_ARC environment variable
1025       during installation); once installation is complete, the name is
1026       fixed
1027    
1028    ----------------------------------------------------------------------
1029    Name: Matthias Blume (blume (at) tti - c (dot) org)
1030    Date: 2003/09/18 16:00:00 CDT
1031    Tag: blume-20030918-110_43_1
1032    Description:
1033    
1034    Made a new interim version and bootfiles for developer's bootstrapping
1035    convenience.
1036    
1037       110.43.1 -- NEW BOOTFILES
1038    
1039    ----------------------------------------------------------------------
1040    Name: Matthias Blume (blume (at) tti - c (dot) org)
1041    Date: 2003/09/18 15:20:00 CDT
1042    Tag: blume-20030918-misc
1043    Description:
1044    
1045    1. Exported fractionsPerSecond etc. from TimeImp (but not from Time as
1046       this seems to be controversial at the moment) and used those in
1047       Posix.ProcEnv.times.
1048    
1049    2. Added Time.{from,to}Nanoseconds to Time.
1050    
1051    3. Improved Real.{from,to}LargeInt by avoiding needless calculations.
1052       For example, fromLargeInt never needs to look at more than 3 "big
1053       digits" to get its 53 bits of precision.
1054    
1055    ----------------------------------------------------------------------
1056    Name: Matthias Blume (blume (at) tti - c (dot) org)
1057    Date: 2003/09/17 16:30:00 CDT
1058    Tag: blume-20030917-real32-slices
1059    Description:
1060    
1061    Added an entry to the primitive environment
1062    (compiler/Semant/statenv/prim.sml) for int32->real64 conversion and
1063    added code to compiler/CodeGen/main/mlriscGen.sml to implement it.
1064    
1065    Removed some of the "magic" constants in real64.sml and replaced them
1066    with code that generates these values from their corresponding
1067    integer counterparts.
1068    
1069    Made all(?) the slice-related changes to the Basis and made everything
1070    compile again...
1071    
1072    ----------------------------------------------------------------------
1073    Name: Matthias Blume (blume (at) tti - c (dot) org)
1074    Date: 2003/09/15 17:45:00 CDT
1075    Tag: blume-20030915-rbase
1076    Description:
1077    
1078    Fixed bug in Real.fromLargeInt.
1079    
1080    ----------------------------------------------------------------------
1081    Name: Matthias Blume (blume (at) tti - c (dot) org)
1082    Date: 2003/09/13 18:11:00 CDT
1083    Tag: blume-20030913-libinstall
1084    Description:
1085    
1086    Minor bugfix in config/libinstall (set anchor with path to
1087    standalone tool after installing it, otherwise libraries that
1088    need ml-lex or ml-yacc won't compile the first time the installer
1089    runs).
1090    
1091    ----------------------------------------------------------------------
1092    Name: Matthias Blume (blume (at) tti - c (dot) org)
1093    Date: 2003/09/12 11:45:00 CDT
1094    Tag: blume-20030912-various
1095    Description:
1096    
1097     - fixed bug in Real.toLargeInt
1098     - fixed bug in Posix.ProcEnv.times
1099     - changed inputLine functions to return an option
1100     - minor installer improvements / bugfixes
1101     - changed default @SMLalloc parameter for x86/celeron to 64k
1102    
1103    ----------------------------------------------------------------------
1104    Name: Matthias Blume (blume (at) tti - c (dot) org)
1105    Date: 2003/09/09 22:00:00 CDT
1106    Tag: Release_110_43
1107    Description:
1108    
1109        New working release 110.43.  New bootfiles.
1110    
1111    ----------------------------------------------------------------------
1112    Name: Matthias Blume (blume (at) tti - c (dot) org)
1113    Date: 2003/09/09 19:20:00 CDT
1114    Tag: blume-20030909-installer
1115    Description:
1116    
1117    Rewrote large parts of config/install.sh in SML (config/libinstall.sml).
1118    Modified config/install.bat to take advantage of it.  Also modified
1119    config/install.sh (and called it config/new-install.sh) to take advantage
1120    of it on Unix systems.  (The SML code is (supposed to be) platform-
1121    independent.)
1122    
1123    The installer can now install everything under Win32
1124    as well as under *nix as long as it compiles.
1125    
1126    Other changes:
1127    
1128      - made CML compile again under Win32
1129      - made eXene compile under Win32 (by providing a fake structure UnixSock
1130        and by using OS.Process.getEnv instead of Posix.ProcEnv.getenv)
1131      - fixed a bug in nowhere: it assumed that type OS.Process.status is the
1132        same as type int;  under Win32 it isn't
1133      - fixed some slice-related problems in the win32-specific parts of CML
1134      - added a functor argument "sameVol" to os-path-fn.sml in the Basis
1135        (under Win32, the volume name is case-insensitive, and the
1136        OS.Path code compares volume names for equality)
1137    
1138    ----------------------------------------------------------------------
1139    Name: Matthias Blume (blume (at) tti - c (dot) org)
1140    Date: 2003/09/08 11:55:00 CDT
1141    Tag: blume-20030908-fullpath
1142    Description:
1143    
1144    Made Win32 version of OS.FileSys.fullPath return current directory
1145    when given an empty string.  This is what the spec says, and incidentally,
1146    CM depends on it.  (CM otherwise goes into an infinite loop in certain
1147    cases when presented with the name of a non-existing .cm file.)
1148    
1149    ----------------------------------------------------------------------
1150    Name: Matthias Blume (blume (at) tti - c (dot) org)
1151    Date: 2003/09/04 16:30:00 CDT
1152    Tag: blume-20030905-slices-etc
1153    Description:
1154    
1155      1. Changed interface to vectors and arrays in Basis to match
1156         (draft) Basis spec.
1157      2. Added signatures and implementations of slices according to
1158         Basis spec.
1159      3. Edited source code throughout the system to make it compile again
1160         under 1. and 2.  (In some cases code had to be added to have it
1161         match the new signatures.)
1162      4. MLRISC should be backward-compatible: the copies of the originals
1163         of files that needed to change under 3. were retained, the .cm files
1164         check the compiler version number and use old versions when
1165         appropriate.
1166      5. Changed type of OS.FileSys.readDir and Posix.FileSys.readdir to
1167         dirstream -> string option (in accordance with Basis spec).
1168      6. When generating code that counts lines, ml-lex used function
1169         CharVector.foldli, taking advantage of its old interface.
1170         This has been replaced with the corresponding code from
1171         CharVectorSlice.  (html-lex must be re-lexed!)
1172      7. BitArray in smlnj-lib/Util has been extended/modified to match the
1173         new MONO_ARRAY signature.  (Do we need BitArraySlice?)
1174      8. Removed temporary additions (fromInternal, toInternal) from the
1175         (now obsolete) IntInf in smlnj-lib/Util.
1176      9. Cleaned up structure Byte.
1177     10. Added localOffset, scan, and fromString to Date (according to spec).
1178         Cleaned/corrected implementation of Date.
1179         (Still need to check for correctness; implement better canonicalizeDate.)
1180     11. Added "scan" to signature IEEE_REAL.
1181     12. Some improvements to IntInf [in particular: efficiency-hack for
1182         mod and rem when second operand is 2 (for parity checks).]
1183     13. Changed representation of type Time.time, using a single IntInf.int
1184         value counting microseconds.  This considerably simplified the
1185         implementation of structure Time.  We now support negative time
1186         values; scan and fromString handle signs.
1187     14. Functor PrimIO now takes two additional arguments (VectorSlice and
1188         ArraySlice).
1189    
1190    ----------------------------------------------------------------------
1191    Name: Matthias Blume (blume (at) tti - c (dot) org)
1192    Date: 2003/08/28 17:00:00 CDT
1193    Tag: blume-20030828-intinf
1194    Description:
1195    
1196    This is a major update which comes with a version number bump
1197    (110.42.99 -- yes, we are really close to 110.43 :-), NEW BOOTFILES,
1198    and an implementation of IntInf in the Basis.
1199    
1200    There are a fairly large number of related changes and updates throughout
1201    the system:
1202    
1203    Basis:
1204            - Implemented IntInf.
1205            - Made LargeInt a projection of IntInf (by filtering through INTEGER).
1206            - Added some missing Real64 operations, most notably Real.toLargeInt.
1207            - Added FixedInt as a synonym for Int32.
1208    
1209    compiler:
1210            * Added support for a built-in intinf type.
1211              - literals
1212              - pattern matching
1213              - conversion shortcuts (Int32.fromLarge o Int.toLarge etc.)
1214              - overloading on literals and operations
1215    
1216            This required adding a primitive type intinf, some additional
1217            primops, and implementations for several non-trivial intinf
1218            operations in Core.  (The intinf type is completely abstract
1219            to the compiler; all operations get delegated back to the Core.)
1220    
1221            * Intinf equality is handled by polyequal.  However, the compiler
1222            does not print its usual warning in this case (since polyequal
1223            is the right thing to do there).
1224    
1225            * Improved the organization of structure InlineT.
1226    
1227            * A word about conversion primops:
1228            If conversions involving intinf do not cancel out during
1229            CPS contract, then the compiler must insert calls to Core functions.
1230            Since all core access must be resolved already during the FLINT
1231            translate phase, it would be too late a the time of CPS contract
1232            to add new Core calls.  For this reason, conversion primops
1233            for intinf carry two arguments:  1. the numeric argument that
1234            they are supposed to convert, and 2. the Core function that
1235            can help with this conversion if necessary.  If CPS contract
1236            eliminates a primop, then the associated Core function becomes
1237            dead and goes away.  Intinf conversion primops that do not get
1238            eliminated by CPS contract get rewritten into calls of their
1239            core functions by a separate, new phase.
1240    
1241    interactive system:
1242            - Control.Print.intinfDepth controls max length of intinf constants
1243            being printed. (Analogous to Control.Print.stringDepth.)
1244            - Cleanup in printutil and pputil: got rid of unused stuff and
1245            duplicates; replaced some of the code with code that makes better
1246            use of library functionality.
1247    
1248    CM:
1249            Bugfix: parse-errors in init group (system/smlnj/init/init.cmi)
1250            are no longer silent.
1251    
1252    CKIT:
1253            Fixed mismatched uses of Int32 and LargeInt.  I always decided
1254            in favor of LargeInt -- which is now the same as IntInf.
1255            CKIT-knowledgable people should check whether this is what's
1256            intended and otherwise change things back to using Int32 or
1257            FixedInt.
1258    
1259    Throughout the code:
1260            Started using IntInf.int literals and built-in operations
1261            (e.g., comparison with 0) where this seems appropriate.
1262    
1263    
1264    ----------------------------------------------------------------------
1265    Name: Dave MacQueen (dbm@cs.uchicago.edu)
1266    Date: 2003/08/13 11:36:00 CDT
1267    Tag: dbm-20030813-mcz-merge1
1268    Description:
1269    
1270    Merging changes from the mcz-branch development branch into trunk.
1271    These changes involve replacement of the emulated old prettyprinter
1272    interface with direct use of the SML/NJ Lib PP library, and fixing
1273    of a couple of bugs (895, 1186) relating to error messages.  A new
1274    prettyprinter for ast datatypes (Elaborator/print/ppast.{sig,sml})
1275    has been added.
1276    
1277    ----------------------------------------------------------------------
1278    Name: Matthias Blume (blume (at) tti - c (dot) org)
1279    Date: 2003/08/11 15:45:00 CDT
1280    Tag: blume-20030811-windows
1281    Description:
1282    
1283    Version number bumped to 110.42.9.  NEW BOOTFILES!!!
1284    
1285        http://smlnj.cs.uchicago.edu/dist/working/110.42.9/
1286    
1287    This patch restores SML/NJ's ability to run under win32.  There are a
1288    number of changes, including fixes for several bugs that had gone
1289    unnoticed until now:
1290    
1291      - uname "CYGWIN_NT*" is recognized as win32 (This is relevant only when
1292        trying to run the win32 version from within cygwin.)
1293    
1294      - There are a number of simple .bat scripts that substitute for their
1295        corresponding Unix shell-scripts.  (See below.)
1296    
1297      - The internals of ml-build have been modified slightly.  The main
1298        difference is that instead of calling ".link-sml" (or link-sml.bat)
1299        using OS.Process.system, the ML process delegates this task back
1300        to the script.  Otherwise problems arise in mixed environments such
1301        as Cygwin where scripts look and work like Unix scripts, but
1302        where OS.Process.system cannot run them.
1303    
1304      - In CM, the srcpath pickler used native pathname syntax -- which
1305        is incorrect in the case of cross-compilation.  The new pickle format
1306        is independent of platform-specific naming conventions.
1307    
1308      - Path configuration files (such as lib/pathconfig) can now choose
1309        between native and standard syntax.  Placing a line of the form
1310    
1311            standard!
1312    
1313        into the file causes all subsequent paths to be interpreted using
1314        CM standard pathname syntax (= Unix conventions); a line
1315    
1316            native!
1317    
1318        switches back to native style.  This was needed so that
1319        path config files can be written portably, see src/system/pathconfig.
1320    
1321      - Runtime system:
1322    
1323          - win32-filesys.c:  get_file_time and set_file_time now
1324            access modification time, not creation time.
1325    
1326          - I/O code made aware of new array representation.
1327    
1328          - Bug fixes in X86.prim.masm.
1329    
1330       - src/system/makeml made aware of win32. (For use under cygwin
1331            and other Unix-environments for windows.)
1332    
1333       - In Basis, fixed off-by-one error in win32-io.sml (function vecF)
1334         which caused BinIO.inputAll to fail consistently.
1335    
1336    .bat scripts:
1337    
1338       Windows .bat scripts assume that SMLNJ_HOME is defined.
1339    
1340       - sml.bat, ml-yacc.bat, ml-lex.bat:  Driver scripts for standalone
1341         applications (sml, ml-yacc, ml-lex).
1342       - ml-build.bat: analogous to ml-build.
1343       - config\install.bat:  Analogous to config/install.sh.  This requires
1344         that SMLNJ_HOME is set and that Microsoft Visual C is ready to use.
1345         (nmake etc. must be on the path, and vcvars32 must have been run.)
1346         Moreover, sources for ml-lex and ml-yacc need to exist under src,
1347         and the bootfile hierarchy must have been unpacked under
1348         sml.boot.x86-win32.
1349         The script is very primitive and does a poor job at error checking.
1350         It only installs the base system, ml-lex, and ml-yacc.  No other
1351         libraries are being installed (i.e., you get only those that
1352         are part of the compiler.)
1353       - link-sml.bat:  analogous to .link-sml, but not currently used
1354    
1355    Unrelated bug fixes:
1356    
1357      - ml-nlffigen now exports structures ST_* corresponding to incomplete
1358        types.
1359      - Added getDevice to PP/src/pp-debug-fn.sml.  (Would not compile
1360        otherwise.)
1361    
1362    ----------------------------------------------------------------------
1363    Name: Dave MacQueen (macqueen@cs.uchicago.edu)
1364    Date: 2003/06/17
1365    Tag: macqueen-20030617-bug895
1366    Description:
1367    
1368    Modified compiler/Elaborator/print/pptype.sml to fix bug 895.
1369    Tag will be used for new development branch (mcz-branch) for
1370    use by MacQueen, (Lucasz) Zairek, and (George) Cao at uchicago.
1371    
1372    ----------------------------------------------------------------------
1373    Name: Matthias Blume (blume (at) tti - c (dot) org)
1374    Date: 2003/05/27 16:55:00 CDT
1375    Tag: blume-20030527-polyeq
1376    Description:
1377    
1378    Tried to eliminated most cases of polymorphic equality.
1379    
1380    ----------------------------------------------------------------------
1381    Name: Matthias Blume (blume (at) tti - c (dot) org)
1382    Date: 2003/05/21 17:45:00 CDT
1383    Tag: blume-20030517-complete
1384    Description:
1385    
1386    Two changes:
1387    
1388      1. Added a flag for controlling whether non-exhaustive bindings will
1389         be treated as errors (default is false).
1390      2. Cleaned up the *entire* source tree so that CMB.make goes through
1391         without a single non-exhaustive match- or bind warning.
1392    
1393    ----------------------------------------------------------------------
1394    Name: Matthias Blume (blume (at) tti - c (dot) org)
1395    Date: 2003/05/17 10:20:00 CDT
1396    Tag: blume-20030517-absyn
1397    Description:
1398    
1399    1. Added cases for IF, WHILE, ANDALSO, and ORELSE to Absyn.
1400    
1401       This mainly affects the quality of error messages.  However, some
1402       of the code is now more straightforward than before. (Treatment of
1403       the above four constructs in translate.sml is much simpler than
1404       the "macro-expansion" that was going on before.  Plus, the mach-
1405       compiler no longer gets invoked just to be able to compile an
1406       if-expression.)
1407    
1408    2. The ErrorMsg.Error exception is now caught and absorbed by the
1409       interactive loop.
1410    
1411    ----------------------------------------------------------------------
1412    Name: Allen Leung
1413    Date: 2003/05/16 13:05:00 CDT
1414    Tag: leunga-20030516-cygwin-runtime
1415    Description:
1416    
1417        Ported the runtime system to cygwin, which uses the unix
1418    x86-unix bin files.  Missing/buggy features:
1419    
1420       o getnetbyname, getnetbyaddr: these functions seem to be missing in
1421         the Cygwin library.
1422       o Ctrl-C handling may be flaky.
1423       o Windows system calls and Windows I/O are not supported.
1424    
1425       A new set of binfiles is located at:
1426    
1427         http://www.dorsai.org/~leunga/boot.x86-unix.tgz
1428    
1429       This is only needed for bootstrapping the cygwin version of smlnj.
1430       Other x86 versions can use the existing binfiles.
1431    
1432    ----------------------------------------------------------------------
1433    Name: Matthias Blume
1434    Date: 2003/04/08 15:42:00 CDT
1435    Tag: blume-20030408-listpair
1436    Description:
1437    
1438    1. Added a target 'mlrisc' to installer.
1439    
1440    2. Added missing elements to structure ListPair.
1441    
1442    ----------------------------------------------------------------------
1443    Name: Allen Leung
1444    Date: 2003/01/07 10:40:00 EST
1445    Tag: leunga-20030107-int-rem
1446    Description:
1447    
1448    Fixed a bug in Int.rem(x,y) where y is a power of 2 on x86.
1449    The arguments to the SUBL instruction were swapped.
1450    
1451    ----------------------------------------------------------------------
1452    Name: Matthias Blume
1453    Date: 2002/12/12 16:25:00 EST
1454    Tag: blume-20021212-risc-ra
1455    Description:
1456    
1457    Fixed a serious bug in the rewrite code for FP spilling/reloading that
1458    sent the RA into an infinite loop when floating point registers get
1459    spilled.  (Because of this bug, e.g., nucleic stopped compiling between
1460    110.37 and 110.38.)
1461    There was another set of potential problems related to the handling of
1462    MLRISC annotations (but those did not yet cause real problems, apparently).
1463    
1464    ----------------------------------------------------------------------
1465    Name: Matthias Blume
1466    Date: 2002/12/06 22:40:00 EST
1467    Tag: blume-20021206-cm-fileid
1468    Description:
1469    
1470    Added a call of SrcPath.sync at the beginning of Parse.parse (in CM).
1471    This fixes the problem of CM getting confused by files that suddenly
1472    change their identity (e.g., by getting unlinked and recreated by some
1473    text editor such as vi).  There might be a better/cheaper/cleaner way
1474    of doing this, but for now this will have to do.
1475    
1476    ----------------------------------------------------------------------
1477    Name: Matthias Blume
1478    Date: 2002/10/28 09:50:00 EST
1479    Tag: blume-20021028-typecheck
1480    Description:
1481    
1482    Exported structure Typecheck from $smlnj/viscomp/core.cm.
1483    
1484    ----------------------------------------------------------------------
1485    Name: Matthias Blume
1486    Date: 2002/10/17 09:10:00 EDT
1487    Tag: Release_110_42
1488    Description:
1489    
1490    In good old tradition, there has been a slight hiccup so that we have
1491    to patch 110.42 after the fact.   The old release tag has been replaced
1492    (see below).
1493    
1494    The change solves a problem with two competing approaches the
1495    configuration problem regarding MacOS 10.1 vs. MacOS 10.2 which got in
1496    each other's way.
1497    
1498    This change only affects the runtime system code and the installer script.
1499    (No new bootfiles.)
1500    
1501    ----------------------------------------------------------------------
1502    Name: Matthias Blume
1503    Date: 2002/10/16 12:00:00 EDT
1504    Tag: Release_110_42_removed
1505    Description:
1506    
1507    New working release.  New bootfiles.
1508    
1509    ----------------------------------------------------------------------
1510    Name: Matthias Blume
1511    Date: 2002/10/10 13:10:00 EDT
1512    Tag: blume-20021010-ppc-divs
1513    Description:
1514    
1515    The mltree operator DIVS must be implemented with an overflow check on
1516    the PPC because the hardware indicates divide-by-zero using "overflow" as
1517    well.
1518    
1519    ----------------------------------------------------------------------
1520    Name: Matthias Blume
1521    Date: 2002/07/23 11:45:00
1522    Tag: blume-20020723-smlnj-home
1523    Description:
1524    
1525    Sml now senses the SMLNJ_HOME environment variable.  If this is set,
1526    then the bin dir is assumed to be in $SMLNJ_HOME/bin and (unless
1527    CM_PATHCONFIG is also set), the path configuration file is assumed
1528    to be in $SMLNJ_HOME/lib/pathconfig.  This way one can easily move
1529    the entire tree to some other place and everything will "just work".
1530    
1531    (Companion commands such as ml-build and ml-makedepend also sense this
1532    variable.)
1533    
1534    ----------------------------------------------------------------------
1535    Name: Matthias Blume
1536    Date: 2002/07/12 21:19:00 EDT
1537    Tag: blume-20020712-liveness
1538    Description:
1539    
1540    Exported two useful "step" functions from liveness module (MLRISC).
1541    
1542    ----------------------------------------------------------------------
1543    Name: Matthias Blume
1544    Date: 2002/07/05 16:00 EDT
1545    Tag: Release_110_41
1546    Description:
1547    
1548    New working release.  New bootfiles.
1549    
1550    ----------------------------------------------------------------------
1551    Name: Matthias Blume
1552    Date: 2002/07/05 10:25:00 EDT
1553    Tag: blume-20020705-btimp
1554    Description:
1555    
1556    Exported structure BTImp from $smlnj/viscomp/debugprof.cm so that
1557    other clients can set up backtracing support.
1558    
1559    ----------------------------------------------------------------------
1560    Name: Matthias Blume
1561    Date: 2002/06/25 17:23:00 EDT
1562    Tag: blume-20020625-fpmax
1563    Description:
1564    
1565    Fixed a bug in translation of INLMAX (and INLMIN) for the floating-point
1566    case.  (The sense of the isNaN test was reversed -- which made min and
1567    max always return their first argument.)
1568    
1569    ----------------------------------------------------------------------
1570    Name: Matthias Blume
1571    Date: 2002/06/11
1572    Tag: blume-20020611-unixpath
1573    Description:
1574    
1575    Back-ported OS.Path.{from,to}UnixPath from idlbasis-devel branch.
1576    
1577    ----------------------------------------------------------------------
1578    Name: Matthias Blume
1579    Date: 2002/06/10 16:35:00 EDT
1580    Tag: blume-20020610-ieeereal
1581    Description:
1582    
1583    I back-ported my implementation of IEEEReal.fromString from the
1584    idlbasis-devel branch so that we can test it.
1585    
1586    Another small change is that ppDec tries to give more information
1587    than just "<sig>" in the case of functors.  However, this code is
1588    broken in some mysterious way if the functor's body's signature
1589    has not been declared by ascription but gets inferred from the
1590    implementation.  This needs fixing...
1591    
1592    ----------------------------------------------------------------------
1593    Name: Matthias Blume
1594    Date: 2002/05/31
1595    Tag: blume-20020531-btrace-mode
1596    Description:
1597    
1598    Resurrected SMLofNJ.Internals.BTrace.mode.  (It accidentally fell by
1599    the wayside when I switched over to using Controls everywhere.)
1600    
1601    ----------------------------------------------------------------------
1602    Name: Lal George
1603    Date: 2002/05/23 12:21:40 EDT
1604    Tag: george-20020523-visual-labels
1605    Description:
1606    
1607    Labels are now displayed in the graphical output to make
1608    the fall-through and target blocks obvious.
1609    
1610    ----------------------------------------------------------------------
1611    Name: Matthias Blume
1612    Date: 2002/05/22 11:03:00 EDT
1613    Tag: blume-20020522-shrink
1614    Description:
1615    
1616    John tweaked yesterday's fix for 1131 to handle an out-of-memory
1617    situation that comes up when allocating huge arrays.
1618    
1619    ----------------------------------------------------------------------
1620    Name: Matthias Blume
1621    Date: 2002/05/21 16:00:00 EDT
1622    Tag: Release_110_40
1623    Description:
1624    
1625    New working release (110.40).  New bootfiles.
1626    
1627    [Also: John Reppy fixed GC bug 1131.]
1628    
1629    ----------------------------------------------------------------------
1630    Name: Matthias Blume
1631    Date: 2002/05/21 12:35:00 EDT
1632    Tag: blume-20020521-cmdoc
1633    Description:
1634    
1635    CM documentation update.
1636    
1637    ----------------------------------------------------------------------
1638    Name: Matthias Blume
1639    Date: 2002/05/21 10:55:00 EDT
1640    Tag: blume-20020521-misc
1641    Description:
1642    
1643    - John tweaked runtime to be silent on heap export (except when
1644      GC messages are on).
1645    - I added a few more things (cross-compiling versions of CMB) to
1646      config/preloads (as suggestions).
1647    
1648    ----------------------------------------------------------------------
1649    Name: Matthias Blume
1650    Date: 2002/05/20 22:25:00 EDT
1651    Tag: blume-20020520-controls
1652    Description:
1653    
1654    - Added ControlUtil structure to control-lib.cm.
1655    - Use it throughout.
1656    - Used Controls facility to define MLRISC controls (as opposed to
1657      registering MLRISC control ref cells with Controls after the fact)
1658    - Fixed messed-up controls priorities.
1659    
1660    * Removed again all the stuff from config/preloads that one wouldn't
1661      be able to preload at the time the initial heap image is built.
1662      (Many libraries, e.g., CML, do not exist yet at this time.  The
1663       only libraries that can be preloaded via config/preloads are those
1664       that come bundled with the bootfiles.)
1665    
1666    ----------------------------------------------------------------------
1667    Name: Matthias Blume
1668    Date: 2002/05/20 10:59:00 EDT
1669    Tag: blume-20020520-preloads
1670    Description:
1671    
1672    Added a lot of commented-out suggestions for things to be included
1673    in config/preloads.
1674    
1675    ----------------------------------------------------------------------
1676    Name: Allen Leung
1677    Date: 2002/05/18 14:20:00 EDT
1678    Tag: leunga-20020518-mdl
1679    Description:
1680    
1681    o Made the mdl tool stuff compile and run again.
1682    o I've disabled all the stuff that depends on RTL specifications; they
1683      are all badly broken anyway.
1684    
1685    ----------------------------------------------------------------------
1686    Name: Matthias Blume
1687    Date: 2002/05/17 16:49:00 EDT
1688    Tag: blume-20020517-controls
1689    Description:
1690    
1691    0. John Reppy made several modifications to the SML/NJ library.
1692       In particular, there is a shiny new controls-lib.cm.
1693    
1694    1. Pushed new controls interface through compiler so that everything
1695       compiles again.
1696    
1697    2. Added FormatComb and FORMAT_COMB to the CML version of the
1698       SML/NJ library (so that CML compiles again).
1699    
1700    3. Modified init scripts because XXX_DEFAULT environment variables
1701       are no longer with us.  (Boot-time initialization is now done
1702       using the same environment variables that are also used for
1703       startup-time initialization of controls.)
1704    
1705    ----------------------------------------------------------------------
1706    Name: Lal George
1707    Date: 2002/05/15 09:20:10 EDT
1708    Tag: george-20020515-pseudo-op-decls
1709    Description:
1710    
1711            All pseudo-ops emitted before the first segment declaration
1712    such as TEXT, DATA, and BSS directives are assumed to be global
1713    declarations and are emitted first in the assembly file. This is
1714    useful in a number of situations where one has pseudo-ops that are not
1715    specific to any segment, and also works around the constraint that one
1716    cannot have client pseudo-ops in the TEXT segment.
1717    
1718            Because no segment is associated with these declarations it is
1719    an error to allocate any space or objects before the first segment
1720    directive and an exception will be raised. However, we cannot make
1721    this check for client pseudo-ops.
1722    
1723            These top level declarations are a field in the CFG graph_info.
1724    In theory you can continue to add to this field after the CFG has been
1725    built -- provided you know what you are doing;-)
1726    
1727    ----------------------------------------------------------------------
1728    Name: Matthias Blume
1729    Date: 2002/05/13 16:40:00 EDT
1730    Tag: blume-20020513-pp-etc
1731    Description:
1732    
1733    A few minor bugfixes:
1734    
1735      - Stopgap measure for bug recently reported by Elsa Gunter (ppDec).
1736        (Bogus printouts for redefined bindings still occur.  Compiler
1737        bug should no longer occur now.  We need to redo the prettyprinter
1738        from scratch.)
1739    
1740      - CM pathname printer now also adds escape sequences for ( and )
1741    
1742      - commend and docu fixes for ml-nlffi
1743    
1744    ----------------------------------------------------------------------
1745    Name: Matthias Blume
1746    Date: 2002/05/10 16:40:00 EDT
1747    Tag: blume-20020510-erg-textio
1748    Description:
1749    
1750    Applied the following bugfix provided by Emden Gansner:
1751    
1752        Output is corrupted when outputSubstr is used rather than output.
1753    
1754        The problem occurs when a substring
1755    
1756            ss = (s, dataStart, dataLen)
1757    
1758        where dataStart > 0, fills a stream buffer with avail bytes left.
1759        avail bytes of s, starting at index dataStart, are copied into the
1760        buffer, the buffer is flushed, and then the remaining dataLen-avail
1761        bytes of ss are copied into the beginning of the buffer. Instead of
1762        starting this copy at index dataStart+avail in s, the current code
1763        starts the copy at index avail.
1764    
1765        Fix:
1766        In text-io-fn.sml, change line 695 from
1767             val needsFlush = copyVec(v, avail, dataLen-avail, buf, 0)
1768        to
1769             val needsFlush = copyVec(v, dataStart+avail, dataLen-avail, buf, 0)
1770    
1771    ----------------------------------------------------------------------
1772    Name: Matthias Blume
1773    Date: 2002/04/12 13:55:00 EDT
1774    Tag: blume-20020412-assyntax
1775    Description:
1776    
1777    1. Grabbed newer assyntax.h from the XFree86 project.
1778    2. Fiddled with how to compile X86.prim.asm without warnings.
1779    3. (Very) Minor cleanup in CM.
1780    
1781    ----------------------------------------------------------------------
1782    Name: Matthias Blume
1783    Date: 2002/04/01 (no joke!) 17:07:00 EST
1784    Tag: blume-20020401-x86div
1785    Description:
1786    
1787    Added full support for div/mod/rem/quot on the x86, using the machine
1788    instruction's two results (without clumsily recomputing the remainder)
1789    directly where appropriate.
1790    
1791    Some more extensive power-of-two support was added to the x86 instruction
1792    selector (avoiding expensive divs, mods, and muls where they can be
1793    replaced with cheaper shifts and masks).  However, this sort of thing
1794    ought to be done earlier, e.g., within the CPS optimizer so that
1795    all architectures benefit from it.
1796    
1797    The compiler compiles to a fixed point, but changes might be somewhat
1798    fragile nevertheless.  Please, report any strange things that you might
1799    see wrt. div/mod/quot/rem...
1800    
1801    ----------------------------------------------------------------------
1802    Name: Matthias Blume
1803    Date: 2002/03/29 17:22:00
1804    Tag: blume-20020329-div
1805    Description:
1806    
1807    Fixed my broken div/mod logic.  Unfortunately, this means that the
1808    inline code for div/mod now has one more comparison than before.
1809    Fast paths (quotient > 0 or remainder = 0) are not affected, though.
1810    The problem was with quotient = 0, because that alone does not tell
1811    us which way the rounding went.  One then has to look at whether
1812    remainder and divisor have the same sign...  :(
1813    
1814    Anyway, I replaced the bootfiles with fresh ones...
1815    
1816    ----------------------------------------------------------------------
1817    Name: Matthias Blume
1818    Date: 2002/03/29 14:10:00 EST
1819    Tag: blume-20020329-inlprims
1820    Description:
1821    
1822    NEW BOOTFILES!!!    Version number bumped to 110.39.3.
1823    
1824    Primops have changed. This means that the bin/boot-file formats have
1825    changed as well.
1826    
1827    To make sure that there is no confusion, I made a new version.
1828    
1829    
1830    CHANGES:
1831    
1832    * removed REMT from mltree (remainder should never overflow).
1833    
1834    * added primops to deal with divisions of all flavors to the frontend
1835    
1836    * handled these primops all the way through so they map to their respective
1837      MLRISC support
1838    
1839    * used these primops in the implementation of Int, Int32, Word, Word32
1840    
1841    * removed INLDIV, INLMOD, and INLREM as they are no longer necessary
1842    
1843    * parameterized INLMIN, INLMAX, and INLABS by a numkind
1844    
1845    * translate.sml now deals with all flavors of INL{MIN,MAX,ABS}, including
1846      floating point
1847    
1848    * used INL{MIN,MAX,ABS} in the implementation of Int, Int32, Word, Word32,
1849      and Real (but Real.abs maps to a separate floating-point-only primop)
1850    
1851    
1852    TODO items:
1853    
1854    * Hacked Alpha32 instruction selection, disabling the selection of REMx
1855      instructions because the machine instruction encoder cannot handle
1856      them.  (Hppa, PPC, and Sparc instruction selection did not handle
1857      REM in the first place, and REM is supported by the x86 machine coder.)
1858    
1859    * Handle DIV and MOD with DIV_TO_NEGINF directly in the x86 instruction
1860      selection phase.  (The two can be streamlined because the hardware
1861      delivers both quotient and remainder at the same time anyway.)
1862    
1863    * Think about what to do with "valOf(Int32.minInt) div ~1" and friends.
1864      (Currently the behavior is inconsistent both across architectures and
1865      wrt. the draft Basis spec.)
1866    
1867    * Word8 should eventually be handled natively, too.
1868    
1869    * There seems to be one serious bug in mltree-gen.sml.  It appears, though,
1870      as if there currently is no execution path that could trigger it in
1871      SML/NJ.  (The assumptions underlying functions arith and promotable do not
1872      hold for things like multiplication and division.)
1873    
1874    ----------------------------------------------------------------------
1875    Name: Matthias Blume
1876    Date: 2002/03/27 16:27:00 EST
1877    Tag: blume-20020327-mlrisc-divisions
1878    Description:
1879    
1880    Added support for all four division operations (ML's div, mod, quot,
1881    and rem) to MLRISC.  In the course of doing so, I also rationalized
1882    the naming (no more annoying switch-around of DIV and QUOT), by
1883    parameterizing the operation by div_rounding_mode (which can be either
1884    DIV_TO_ZERO or DIV_TO_NEGINF).
1885    
1886    The generic MLTreeGen functor takes care of compiling all four
1887    operations down to only round-to-zero div.
1888    
1889    Missing pieces:
1890    
1891      * Doing something smarter than relying on MLTreeGen on architectures
1892        like, e.g., the x86 where hardware division delivers both quotient and
1893        remainder at the same time.  With this, the implementation of the
1894        round-to-neginf operations could be further streamlined.
1895    
1896      * Remove inlining support for div/mod/rem from the frontend and replace it
1897        with primops that get carried through to the backend.  Do this for all
1898        int and word types.
1899    
1900    ----------------------------------------------------------------------
1901    Name: Matthias Blume
1902    Date: 2002/03/25 17:25:00 EST
1903    Tag: blume-20020325-divmod
1904    Description:
1905    
1906    I improved (hopefully without breaking them) the implementation of Int.div,
1907    Int.mod, and Int.rem.   For this, the code in translate.sml now takes
1908    advantage of the following observations:
1909    
1910      Let  q = x quot y      r = x rem y
1911           d = x div  y      m = x mod y
1912    
1913    where "quot" is the round-to-zero version of integer division that
1914    hardware usually provides.  Then we have:
1915    
1916         r = x - q * y        where neither the * nor the - will overflow
1917         d = if q >= 0 orelse x = q * y then q else q - 1
1918                              where neither the * nor the - will overflow
1919         m = if q >= 0 orelse r = 0 then r else r + y
1920                              where the + will not overflow
1921    
1922    This results in substantial simplification of the generated code.
1923    The following table shows the number of CFG nodes and edges generated
1924    for
1925            fun f (x, y) = x OPER y
1926            (* with OPER \in div, mod, quot, rem *)
1927    
1928    
1929        OPER | nodes(old) | edges(old) | nodes(new) | edges(new)
1930        --------------------------------------------------------
1931         div |         24 |         39 |         12 |         16
1932         mod |         41 |         71 |         12 |         16
1933        quot |          8 |         10 |          8 |         10
1934         rem |         10 |         14 |          8 |         10
1935    
1936    
1937    ----------------------------------------------------------------------
1938    Name: Matthias Blume
1939    Date: 2002/03/25 22:06:00 EST
1940    Tag: blume-20020325-cprotobug
1941    Description:
1942    
1943    Fixed a bug in cproto (c prototype decoder).
1944    
1945    ----------------------------------------------------------------------
1946    Name: Matthias Blume
1947    Date: 2002/03/25 16:00:00 EST
1948    Tag: blume-20020325-raw-primops
1949    Description:
1950    
1951    I did some cleanup to Allen's new primop code and
1952    replaced yesterday's bootfiles with new ones.
1953    (But they are stored in the same place.)
1954    
1955    ----------------------------------------------------------------------
1956    Name: Matthias Blume
1957    Date: 2002/03/24 22:40:00 EST
1958    Tag: blume-20020324-bootfiles
1959    Description:
1960    
1961    Made the bootfiles that Allen asked for.
1962    
1963    ----------------------------------------------------------------------
1964    Name: Allen Leung
1965    Date: 2002/03/23 15:50:00 EST
1966    Tag: leunga-20020323-flint-cps-rcc-primops
1967    Description:
1968    
1969      1. Changes to FLINT primops:
1970    
1971        (* make a call to a C-function;
1972         * The primop carries C function prototype information and specifies
1973         * which of its (ML-) arguments are floating point. C prototype
1974         * information is for use by the backend, ML information is for
1975         * use by the CPS converter. *)
1976      | RAW_CCALL of { c_proto: CTypes.c_proto,
1977                       ml_args: ccall_type list,
1978                       ml_res_opt: ccall_type option,
1979                       reentrant : bool
1980                     } option
1981       (* Allocate uninitialized storage on the heap.
1982        * The record is meant to hold short-lived C objects, i.e., they
1983        * are not ML pointers.  With the tag, the representation is
1984        * the same as RECORD with tag tag_raw32 (sz=4), or tag_fblock (sz=8)
1985        *)
1986      | RAW_RECORD of {tag:bool,sz:int}
1987      and ccall_type = CCALL_INT32 | CCALL_REAL64 | CCALL_ML_PTR
1988    
1989      2.  These CPS primops are now overloaded:
1990    
1991           rawload of {kind:numkind}
1992           rawstore of {kind:numkind}
1993    
1994          The one argument form is:
1995    
1996             rawload {kind} address
1997    
1998          The two argument form is:
1999    
2000             rawload {kind} [ml object, byte-offset]
2001    
2002      3. RAW_CCALL/RCC now takes two extra arguments:
2003    
2004         a. The first is whether the C call is reentrant, i.e., whether
2005            ML state should be saved and restored.
2006         b. The second argument is a string argument specifying the name of
2007            library and the C function.
2008    
2009         These things are currently not handled in the code generator, yet.
2010    
2011      4. In CProto,
2012    
2013         An encoding type of "bool" means "ml object" and is mapped into
2014         C prototype of PTR.  Note that "bool" is different than "string",
2015         even though "string" is also mapped into PTR, because "bool"
2016         is assigned an CPS type of BOGt, while "string" is assigned INT32t.
2017    
2018      5. Pickler/unpicker
2019    
2020         Changed to handle RAW_RECORD and newest RAW_CCALL
2021    
2022      6. MLRiscGen,
2023    
2024         1. Changed to handle the new rawload/rawstore/rawrecord operators.
2025         2. Code for handling C Calls has been moved to a new module CPSCCalls,
2026            in the file CodeGen/cpscompile/cps-c-calls.sml
2027    
2028      7. Added the conditional move operator
2029    
2030             condmove of branch
2031    
2032         to cps.  Generation of this is still buggy so it is currently
2033         disabled.
2034    
2035    ----------------------------------------------------------------------
2036    Name: Lal George
2037    Date: 2002/03/22 14:18:25 EST
2038    Tag: george-20020322-cps-branch-prob
2039    Description:
2040    
2041    Implemented the Ball-Larus branch prediction-heuristics, and
2042    incorporated graphical viewers for control flow graphs.
2043    
2044    Ball-Larus Heuristics:
2045    ---------------------
2046    See the file compiler/CodeGen/cpscompile/cpsBranchProb.sml.
2047    
2048    By design it uses the Dempster-Shafer theory for combining
2049    probabilities.  For example, in the function:
2050    
2051        fun f(n,acc) = if n = 0 then acc else f(n-1, n*acc)
2052    
2053    the ball-larus heuristics predicts that the n=0 is unlikely
2054    (OH-heuristic), and the 'then' branch is unlikely because of the
2055    RH-heuristic -- giving the 'then' branch an even lower combined
2056    probability using the Dempster-Shafer theory.
2057    
2058    Finally, John Reppy's loop analysis in MLRISC, further lowers the
2059    probability of the 'then' branch because of the loop in the else
2060    branch.
2061    
2062    
2063    Graphical Viewing:
2064    ------------------
2065    I merely plugged in Allen's graphical viewers into the compiler. The
2066    additional code is not much. At the top level, saying:
2067    
2068            Control.MLRISC.getFlag "cfg-graphical-view" := true;
2069    
2070    will display the graphical view of the control flow graph just before
2071    back-patching.  daVinci must be in your path for this to work. If
2072    daVinci is not available, then the default viewer can be changed
2073    using:
2074    
2075            Control.MLRISC.getString "viewer"
2076    
2077    which can be set to "dot" or "vcg" for the corresponding viewers. Of
2078    course, these viewers must be in your path.
2079    
2080    The above will display the compilation unit at the level of clusters,
2081    many of which are small, boring, and un-interesting. Also setting:
2082    
2083            Control.MLRISC.getInt "cfg-graphical-view_size"
2084    
2085    will display clusters that are larger than the value set by the above.
2086    
2087    
2088    ----------------------------------------------------------------------
2089    Name: Matthias Blume
2090    Date: 2002/03/21 22:20:00 EST
2091    Tag: blume-20020321-kmp-bugfix
2092    Description:
2093    
2094    Changed the interface to the KMP routine in PreString and fixed
2095    a minor bug in one place where it was used.
2096    
2097    ----------------------------------------------------------------------
2098    Name: Allen Leung
2099    Date: 2002/03/21 20:30:00 EST
2100    Tag: leunga-20020321-cfg
2101    Description:
2102    
2103      Fixed a potential problem in cfg edge splitting.
2104    
2105    ----------------------------------------------------------------------
2106    Name: Allen Leung
2107    Date: 2002/03/21 17:15:00 EST
2108    Tag: leunga-20020321-x86-fp-cfg
2109    Description:
2110    
2111      1. Recoded the buggy parts of x86-fp.
2112    
2113         a. All the block reordering code has been removed.
2114            We now depend on the block placement phases to do this work.
2115    
2116         b. Critical edge splitting code has been simplified and moved into the
2117            CFG modules, as where they belong.
2118    
2119         Both of these were quite buggy and complex.  The code is now much, much
2120         simpler.
2121    
2122      2. X86 backend.
2123    
2124         a. Added instructions for 64-bit support.  Instruction selection for
2125            64-bit has not been committed, however, since that
2126            requires changes to MLTREE which haven't been approved by
2127            Lal and John.
2128    
2129         b. Added support for FUCOMI and FUCOMIP when generating code for
2130            PentiumPro and above.  We only generate these instructions in
2131            the fast-fp mode.
2132    
2133         c. Added cases for JP and JNP in X86FreqProps.
2134    
2135      3. CFG
2136    
2137         CFG now has a bunch of methods for edge splitting and merging.
2138    
2139      4. Machine description.
2140    
2141         John's simplification of MLTREE_BASIS.fcond broke a few machine
2142         description things:
2143    
2144         rtl-build.{sig,sml} and hppa.mdl fixed.
2145    
2146         NOTE: the machine description stuff in the repository is still broken.
2147               Again, I can't put my fixes in because that involves
2148               changes to MLTREE.
2149    
2150    ----------------------------------------------------------------------
2151    Name: Matthias Blume
2152    Date: 2002/03/20 15:55:00 EST
2153    Tag: blume-20020320-kmp
2154    Description:
2155    
2156    Implemented Knuth-Morris-Pratt string matching in PreString and used
2157    it for String.isSubstring, Substring.isSubstring, and
2158    Substring.position.
2159    
2160    (Might need some stress-testing.  Simple examples worked fine.)
2161    
2162    ----------------------------------------------------------------------
2163    Name: Matthias Blume
2164    Date: 2002/03/19 16:37:00 EST
2165    Tag: blume-20020319-witnesses
2166    Description:
2167    
2168    Added a structure C.W and functions convert/Ptr.convert to ml-nlffi-lib.
2169    
2170    This implements a generic mechanism for changing constness qualifiers
2171    anywhere within big C types without resorting to outright "casts".
2172    (So far, functions such as C.rw/C.ro or C.Ptr.rw/C.Ptr.ro only let you
2173    modify the constness at the outermost level.)
2174    The implementation of "convert" is based on the idea of "witness"
2175    values -- values that are not used by the operation but whose types
2176    "testify" to their applicability.  On the implementation side, "convert"
2177    is simply a projection (returning its second curried argument).  With
2178    cross-module inlining, it should not result in any machine code being
2179    generated.
2180    
2181    ----------------------------------------------------------------------
2182    Name: Matthias Blume
2183    Date: 2002/03/15 16:40:00 EST
2184    Tag: blume-20020315-basis
2185    Description:
2186    
2187    Provided (preliminary?) implementations for
2188    
2189      {String,Substring}.{concatWith,isSuffix,isSubstring}
2190    
2191    and
2192    
2193      Substring.full
2194    
2195    Those are in the Basis spec but they were missing in SML/NJ.
2196    
2197    ----------------------------------------------------------------------
2198    Name: Matthias Blume
2199    Date: 2002/03/14 21:30:00 EST
2200    Tag: blume-20020314-controls
2201    Description:
2202    
2203    Controls:
2204    ---------
2205    
2206    1. Factored out the recently-added Controls : CONTROLS stuff and put
2207       it into its own library $/controls-lib.cm.  The source tree for
2208       this is under src/smlnj-lib/Controls.
2209    
2210    2. Changed the names of types and functions in this interface, so they
2211       make a bit more "sense":
2212    
2213          module -> registry
2214          'a registry -> 'a group
2215    
2216    3. The interface now deals in ref cells only.  The getter/setter interface
2217       is (mostly) gone.
2218    
2219    4. Added a function that lets one register an already-existing ref cell.
2220    
2221    5. Made the corresponding modifications to the rest of the code so that
2222       everything compiles again.
2223    
2224    6. Changed the implementation of Controls.MLRISC back to something closer
2225       to the original.  In particular, this module (and therefore MLRISC)
2226       does not depend on Controls.  There now is some link-time code in
2227       int-sys.sml that registers the MLRISC controls with the Controls
2228       module.
2229    
2230    CM:
2231    ---
2232    
2233      * One can now specify the lambda-split aggressiveness in init.cmi.
2234    
2235    ----------------------------------------------------------------------
2236    Name: Allen Leung
2237    Date: 2002/03/13 17:30:00 EST
2238    Tag: leunga-20020313-x86-fp-unary
2239    Description:
2240    
2241    Bug fix for:
2242    
2243    > leunga@weaselbane:~/Yale/tmp/sml-dist{21} bin/sml
2244    > Standard ML of New Jersey v110.39.1 [FLINT v1.5], March 08, 2002
2245    > - fun f(x,(y,z)) = Real.~ y;
2246    > [autoloading]
2247    > [autoloading done]
2248    >       fchsl   (%eax), 184(%esp)
2249    > Error: MLRisc bug: X86MCEmitter.emitInstr
2250    >
2251    > uncaught exception Error
2252    >   raised at: ../MLRISC/control/mlriscErrormsg.sml:16.14-16.19
2253    
2254    The problem was that the code generator did not generate any fp registers
2255    in this case, and the ra didn't know that it needed to run the X86FP phase to
2256    translate the pseudo fp instruction.   This only happened with unary fp
2257    operators in certain situations.
2258    
2259    ----------------------------------------------------------------------
2260    Name: Matthias Blume
2261    Date: 2002/03/13 14:00:00 EST
2262    Tag: blume-20020313-overload-etc
2263    Description:
2264    
2265    1. Added _overload as a synonym for overload for backward compatibility.
2266       (Control.overloadKW must be true for either version to be accepted.)
2267    
2268    2. Fixed bug in install script that caused more things to be installed
2269       than what was requested in config/targets.
2270    
2271    3. Made CM aware of the (_)overload construct so that autoloading
2272       works.
2273    
2274    ----------------------------------------------------------------------
2275    Name: Matthias Blume
2276    Date: 2002/03/12 22:03:00 EST
2277    Tag: blume-20020312-url
2278    Description:
2279    
2280    Forgot to update BOOT and srcarchiveurl.
2281    
2282    ----------------------------------------------------------------------
2283    Name: Matthias Blume
2284    Date: 2002/03/12 17:30:00 EST
2285    Tag: blume-20020312-version110392
2286    Description:
2287    
2288    Yet another version number bump (because of small changes to the
2289    binfile format).  Version number is now 110.39.2.  NEW BOOTFILES!
2290    
2291    Changes:
2292    
2293      The new pid generation scheme described a few weeks ago was overly
2294      complicated.  I implemented a new mechanism that is simpler and
2295      provides a bit more "stability":  Once CM has seen a compilation
2296      unit, it keeps its identity constant (as long as you do not delete
2297      those crucial CM/GUID/* files).  This means that when you change
2298      an interface, compile, then go back to the old interface, and
2299      compile again, you arrive at the original pid.
2300    
2301      There now also is a mechanism that instructs CM to use the plain
2302      environment hash as a module's pid (effectively making its GUID
2303      the empty string).  For this, "noguid" must be specified as an
2304      option to the .sml file in question within its .cm file.
2305      This is most useful for code that is being generated by tools such
2306      as ml-nlffigen (because during development programmers tend to
2307      erase the tool's entire output directory tree including CM's cached
2308      GUIDs).  "noguid" is somewhat dangerous (since it can be used to locally
2309      revert to the old, broken behavior of SML/NJ, but in specific cases
2310      where there is no danger of interface confusion, its use is ok
2311      (I think).
2312    
2313      ml-nlffigen by default generates "noguid" annotations.  They can be
2314      turned off by specifying -guid in its command line.
2315    
2316    ----------------------------------------------------------------------
2317    Name: Lal George
2318    Date: 2002/03/12 12 14:42:36 EST
2319    Tag: george-20020312-frequency-computation
2320    Description:
2321    
2322    Integrated jump chaining and static block frequency into the
2323    compiler. More details and numbers later.
2324    
2325    ----------------------------------------------------------------------
2326    Name: Lal George
2327    Date: 2002/03/11 11 22:38:53 EST
2328    Tag: george-20020311-jump-chain-elim
2329    Description:
2330    
2331    Tested the jump chain elimination on all architectures (except the
2332    hppa).  This is on by default right now and is profitable for the
2333    alpha and x86, however, it may not be profitable for the sparc and ppc
2334    when compiling the compiler.
2335    
2336    The gc test will typically jump to a label at the end of the cluster,
2337    where there is another jump to an external cluster containing the actual
2338    code to invoke gc. This is to allow factoring of common gc invocation
2339    sequences. That is to say, we generate:
2340    
2341            f:
2342               testgc
2343               ja   L1      % jump if above to L1
2344    
2345            L1:
2346               jmp L2
2347    
2348    
2349    After jump chain elimination the 'ja L1' instructions is converted to
2350    'ja L2'. On the sparc and ppc, many of the 'ja L2' instructions may end
2351    up being implemented in their long form (if L2 is far away) using:
2352    
2353            jbe     L3      % jump if below or equal to L3
2354            jmp     L2
2355         L3:
2356            ...
2357    
2358    
2359    For large compilation units L2  may be far away.
2360    
2361    
2362    ----------------------------------------------------------------------
2363    Name: Matthias Blume
2364    Date: 2002/03/11 13:30:00 EST
2365    Tag: blume-20020311-mltreeeval
2366    Description:
2367    
2368    A functor parameter was missing.
2369    
2370    ----------------------------------------------------------------------
2371    Name: Allen Leung
2372    Date: 2002/03/11 10:30:00 EST
2373    Tag: leunga-20020311-runtime-string0
2374    Description:
2375    
2376       The representation of the empty string now points to a
2377    legal null terminated C string instead of unit.  It is now possible
2378    to convert an ML string into C string with InlineT.CharVector.getData.
2379    This compiles into one single machine instruction.
2380    
2381    ----------------------------------------------------------------------
2382    Name: Allen Leung
2383    Date: 2002/03/10 23:55:00 EST
2384    Tag: leunga-20020310-x86-call
2385    Description:
2386    
2387       Added machine generation for CALL instruction (relative displacement mode)
2388    
2389    ----------------------------------------------------------------------
2390    Name: Matthias Blume
2391    Date: 2002/03/08 16:05:00
2392    Tag: blume-20020308-entrypoints
2393    Description:
2394    
2395    Version number bumped to 110.39.1.  NEW BOOTFILES!
2396    
2397    Entrypoints: non-zero offset into a code object where execution should begin.
2398    
2399    - Added the notion of an entrypoint to CodeObj.
2400    - Added reading/writing of entrypoint info to Binfile.
2401    - Made runtime system bootloader aware of entrypoints.
2402    - Use the address of the label of the first function given to mlriscGen
2403      as the entrypoint.  This address is currently always 0, but it will
2404      not be 0 once we turn on block placement.
2405    - Removed the linkage cluster code (which was The Other Way(tm) of dealing
2406      with entry points) from mlriscGen.
2407    
2408    ----------------------------------------------------------------------
2409    Name: Allen Leung
2410    Date: 2002/03/07 20:45:00 EST
2411    Tag: leunga-20020307-x86-cmov
2412    Description:
2413    
2414       Bug fixes for CMOVcc on x86.
2415    
2416       1. Added machine code generation for CMOVcc
2417       2. CMOVcc is now generated in preference over SETcc on PentiumPro or above.
2418       3. CMOVcc cannot have an immediate operand as argument.
2419    
2420    ----------------------------------------------------------------------
2421    Name: Matthias Blume
2422    Date: 2002/03/07 16:15:00 EST
2423    Tag: blume-20020307-controls
2424    Description:
2425    
2426    This is a very large but mostly boring patch which makes (almost)
2427    every tuneable compiler knob (i.e., pretty much everything under
2428    Control.* plus a few other things) configurable via both the command
2429    line and environment variables in the style CM did its configuration
2430    until now.
2431    
2432    Try starting sml with '-h' (or, if you are brave, '-H')
2433    
2434    To this end, I added a structure Controls : CONTROLS to smlnj-lib.cm which
2435    implements the underlying generic mechanism.
2436    
2437    The interface to some of the existing such facilities has changed somewhat.
2438    For example, the MLRiscControl module now provides mkFoo instead of getFoo.
2439    (The getFoo interface is still there for backward-compatibility, but its
2440    use is deprecated.)
2441    
2442    The ml-build script passes -Cxxx=yyy command-line arguments through so
2443    that one can now twiddle the compiler settings when using this "batch"
2444    compiler.
2445    
2446    TODO items:
2447    
2448    We should go through and throw out all controls that are no longer
2449    connected to anything.  Moreover, we should go through and provide
2450    meaningful (and correct!) documentation strings for those controls
2451    that still are connected.
2452    
2453    Currently, multiple calls to Controls.new are accepted (only the first
2454    has any effect).  Eventually we should make sure that every control
2455    is being made (via Controls.new) exactly once.  Future access can then
2456    be done using Controls.acc.
2457    
2458    Finally, it would probably be a good idea to use the getter-setter
2459    interface to controls rather than ref cells.  For the time being, both
2460    styles are provided by the Controls module, but getter-setter pairs are
2461    better if thread-safety is of any concern because they can be wrapped.
2462    
2463    *****************************************
2464    
2465    One bug fix: The function blockPlacement in three of the MLRISC
2466    backpatch files used to be hard-wired to one of two possibilities at
2467    link time (according to the value of the placementFlag).  But (I
2468    think) it should rather sense the flag every time.
2469    
2470    *****************************************
2471    
2472    Other assorted changes (by other people who did not supply a HISTORY entry):
2473    
2474    1. the cross-module inliner now works much better (Monnier)
2475    2. representation of weights, frequencies, and probabilities in MLRISC
2476       changed in preparation of using those for weighted block placement
2477       (Reppy, George)
2478    
2479    ----------------------------------------------------------------------
2480    Name: Lal George
2481    Date: 2002/03/07 14:44:24 EST 2002
2482    Tag: george-20020307-weighted-block-placement
2483    
2484    Tested the weighted block placement optimization on all architectures
2485    (except the hppa) using AMPL to generate the block and edge frequencies.
2486    Changes were required in the machine properties to correctly
2487    categorize trap instructions. There is an MLRISC flag
2488    "weighted-block-placement" that can be used to enable weighted block
2489    placement, but this will be ineffective without block/edge
2490    frequencies (coming soon).
2491    
2492    
2493    ----------------------------------------------------------------------
2494    Name: Lal George
2495    Date: 2002/03/05 17:24:48 EST
2496    Tag: george-20020305-linkage-cluster
2497    
2498    In order to support the block placement optimization, a new cluster
2499    is generated as the very first cluster (called the linkage cluster).
2500    It contains a single jump to the 'real' entry point for the compilation
2501    unit. Block placement has no effect on the linkage cluster itself, but
2502    all the other clusters  have full freedom in the manner in which they
2503    reorder blocks or functions.
2504    
2505    On the x86 the typical linkage code that is generated is:
2506       ----------------------
2507            .align 2
2508       L0:
2509            addl    $L1-L0, 72(%esp)
2510            jmp     L1
2511    
2512    
2513            .align  2
2514       L1:
2515       ----------------------
2516    
2517    72(%esp) is the memory location for the stdlink register. This
2518    must contain the address of the CPS function being called. In the
2519    above example, it contains the address of  L0; before
2520    calling L1 (the real entry point for the compilation unit), it
2521    must contain the address for L1, and hence
2522    
2523            addl $L1-L0, 72(%esp)
2524    
2525    I have tested this on all architectures except the hppa.The increase
2526    in code size is of course negligible
2527    
2528    ----------------------------------------------------------------------
2529    Name: Allen Leung
2530    Date: 2002/03/03 13:20:00 EST
2531    Tag: leunga-20020303-mlrisc-tools
2532    
2533      Added #[ ... ] expressions to mlrisc tools
2534    
2535    ----------------------------------------------------------------------
2536    Name: Matthias Blume
2537    Date: 2002/02/27 12:29:00 EST
2538    Tag: blume-20020227-cdebug
2539    Description:
2540    
2541    - made types in structure C and C_Debug to be equal
2542    - got rid of code duplication (c-int.sml vs. c-int-debug.sml)
2543    - there no longer is a C_Int_Debug (C_Debug is directly derived from C)
2544    
2545    ----------------------------------------------------------------------
2546    Name: Matthias Blume
2547    Date: 2002/02/26 12:00:00 EST
2548    Tag: blume-20020226-ffi
2549    Description:
2550    
2551    1. Fixed a minor bug in CM's "noweb" tool:
2552       If numbering is turned off, then truly don't number (i.e., do not
2553       supply the -L option to noweb).  The previous behavior was to supply
2554       -L'' -- which caused noweb to use the "default" line numbering scheme.
2555       Thanks to Chris Richards for pointing this out (and supplying the fix).
2556    
2557    2. Once again, I reworked some aspects of the FFI:
2558    
2559       A. The incomplete/complete type business:
2560    
2561       - Signatures POINTER_TO_INCOMPLETE_TYPE and accompanying functors are
2562         gone!
2563       - ML types representing an incomplete type are now *equal* to
2564         ML types representing their corresponding complete types (just like
2565         in C).  This is still safe because ml-nlffigen will not generate
2566         RTTI for incomplete types, nor will it generate functions that
2567         require access to such RTTI.   But when ML code generated from both
2568         incomplete and complete versions of the C type meet, the ML types
2569         are trivially interoperable.
2570    
2571         NOTE:  These changes restore the full generality of the translation
2572         (which was previously lost when I eliminated functorization)!
2573    
2574       B. Enum types:
2575    
2576       - Structure C now has a type constructor "enum" that is similar to
2577         how the "su" constructor works.  However, "enum" is not a phantom
2578         type because each "T enum" has values (and is isomorphic to
2579         MLRep.Signed.int).
2580       - There are generic access operations for enum objects (using
2581         MLRep.Signed.int).
2582       - ml-nlffigen will generate a structure E_foo for each "enum foo".
2583         * The structure contains the definition of type "mlrep" (the ML-side
2584         representation type of the enum).  Normally, mlrep is the same
2585         as "MLRep.Signed.int", but if ml-nlffigen was invoked with "-ec",
2586         then mlrep will be defined as a datatype -- thus facilitating
2587         pattern matching on mlrep values.
2588         ("-ec" will be suppressed if there are duplicate values in an
2589          enumeration.)
2590         * Constructors ("-ec") or values (no "-ec") e_xxx of type mlrep
2591         will be generated for each C enum constant xxx.
2592         * Conversion functions m2i and i2m convert between mlrep and
2593         MLRep.Signed.int.  (Without "-ec", these functions are identities.)
2594         * Coversion functions c and ml convert between mlrep and "tag enum".
2595         * Access functions (get/set) fetch and store mlrep values.
2596       - By default (unless ml-nlffigen was invoked with "-nocollect"), unnamed
2597         enumerations are merged into one single enumeration represented by
2598         structure E_'.
2599    
2600    ----------------------------------------------------------------------
2601    Name: Allen Leung
2602    Date: 2002/02/25 04:45:00 EST
2603    Tag: leunga-20020225-cps-spill
2604    
2605    This is a new implementation of the CPS spill phase.
2606    The new phase is in the new file compiler/CodeGen/cpscompile/spill-new.sml
2607    In case of problems, replace it with the old file spill.sml
2608    
2609    The current compiler runs into some serious performance problems when
2610    constructing a large record.  This can happen when we try to compile a
2611    structure with many items.  Even a very simple structure like the following
2612    makes the compiler slow down.
2613    
2614        structure Foo = struct
2615           val x_1 = 0w1 : Word32.int
2616           val x_2 = 0w2 : Word32.int
2617           val x_3 = 0w3 : Word32.int
2618           ...
2619           val x_N = 0wN : Word32.int
2620        end
2621    
2622    The following table shows the compile time, from N=1000 to N=4000,
2623    with the old compiler:
2624    
2625    N
2626    1000   CPS 100 spill                           0.04u  0.00s  0.00g
2627           MLRISC ra                               0.06u  0.00s  0.05g
2628              (spills = 0 reloads = 0)
2629           TOTAL                                   0.63u  0.07s  0.21g
2630    
2631    1100   CPS 100 spill                           8.25u  0.32s  0.64g
2632           MLRISC ra                               5.68u  0.59s  3.93g
2633              (spills = 0 reloads = 0)
2634           TOTAL                                   14.71u  0.99s  4.81g
2635    
2636    1500   CPS 100 spill                           58.55u  2.34s  1.74g
2637           MLRISC ra                               5.54u  0.65s  3.91g
2638              (spills = 543 reloads = 1082)
2639           TOTAL                                   65.40u  3.13s  6.00g
2640    
2641    2000   CPS 100 spill                           126.69u  4.84s  3.08g
2642           MLRISC ra                               0.80u  0.10s  0.55g
2643              (spills = 42 reloads = 84)
2644           TOTAL                                   129.42u  5.10s  4.13g
2645    
2646    3000   CPS 100 spill                           675.59u  19.03s  11.64g
2647           MLRISC ra                               2.69u  0.27s  1.38g
2648              (spills = 62 reloads = 124)
2649           TOTAL                                   682.48u  19.61s  13.99g
2650    
2651    4000   CPS 100 spill                           2362.82u  56.28s  43.60g
2652           MLRISC ra                               4.96u  0.27s  2.72g
2653              (spills = 85 reloads = 170)
2654           TOTAL                                   2375.26u  57.21s  48.00g
2655    
2656    As you can see the old cps spill module suffers from some serious
2657    performance problem.  But since I cannot decipher the old code fully,
2658    instead of patching the problems up, I'm reimplementing it
2659    with a different algorithm.  The new code is more modular,
2660    smaller when compiled, and substantially faster
2661    (O(n log n) time and O(n) space).  Timing of the new spill module:
2662    
2663    4000  CPS 100 spill                           0.02u  0.00s  0.00g
2664          MLRISC ra                               0.25u  0.02s  0.15g
2665             (spills=1 reloads=3)
2666          TOTAL                                   7.74u  0.34s  1.62g
2667    
2668    Implementation details:
2669    
2670    As far as I can tell, the purpose of the CPS spill module is to make sure the
2671    number of live variables at any program point (the bandwidth)
2672    does not exceed a certain limit, which is determined by the
2673    size of the spill area.
2674    
2675    When the bandwidth is too large, we decrease the register pressure by
2676    packing live variables into spill records.  How we achieve this is
2677    completely different than what we did in the old code.
2678    
2679    First, there is something about the MLRiscGen code generator
2680    that we should be aware of:
2681    
2682    o MLRiscGen performs code motion!
2683    
2684       In particular, it will move floating point computations and
2685       address computations involving only the heap pointer to
2686       their use sites (if there is only a single use).
2687       What this means is that if we have a CPS record construction
2688       statement
2689    
2690           RECORD(k,vl,w,e)
2691    
2692       we should never count the new record address w as live if w
2693       has only one use (which is often the case).
2694    
2695       We should do something similar to floating point, but the transformation
2696       there is much more complex, so I won't deal with that.
2697    
2698    Secondly, there are now two new cps primops at our disposal:
2699    
2700     1. rawrecord of record_kind option
2701        This pure operator allocates some uninitialized storage from the heap.
2702        There are two forms:
2703    
2704         rawrecord NONE [INT n]  allocates a tagless record of length n
2705         rawrecord (SOME rk) [INT n] allocates a tagged record of length n
2706                                     and initializes the tag.
2707    
2708     2. rawupdate of cty
2709          rawupdate cty (v,i,x)
2710          Assigns to x to the ith component of record v.
2711          The storelist is not updated.
2712    
2713    We use these new primops for both spilling and increment record construction.
2714    
2715     1. Spilling.
2716    
2717        This is implemented with a linear scan algorithm (but generalized
2718        to trees).  The algorithm will create a single spill record at the
2719        beginning of the cps function and use rawupdate to spill to it,
2720        and SELECT or SELp to reload from it.  So both spills and reloads
2721        are fine-grain operations.  In contrast, in the old algorithm
2722        "spills" have to be bundled together in records.
2723    
2724        Ideally, we should sink the spill record construction to where
2725        it is needed.  We can even split the spill record into multiple ones
2726        at the places where they are needed.  But CPS is not a good
2727        representation for global code motion, so I'll keep it simple and
2728        am not attempting this.
2729    
2730     2. Incremental record construction (aka record splitting).
2731    
2732        Long records with many component values which are simulatenously live
2733        (recall that single use record addresses are not considered to
2734         be live) are constructed with rawrecord and rawupdate.
2735        We allocate space on the heap with rawrecord first, then gradually
2736        fill it in with rawupdate.  This is the technique suggested to me
2737        by Matthias.
2738    
2739        Some restrictions on when this is applicable:
2740        1. It is not a VECTOR record.  The code generator currently does not handle
2741           this case. VECTOR record uses double indirection like arrays.
2742        2. All the record component values are defined in the same "basic block"
2743           as the record constructor.  This is to prevent speculative
2744           record construction.
2745    
2746    ----------------------------------------------------------------------
2747    Name: Allen Leung
2748    Date: 2002/02/22 01:02:00 EST
2749    Tag: leunga-20020222-mlrisc-tools
2750    
2751    Minor bug fixes in the parser and rewriter
2752    
2753    ----------------------------------------------------------------------
2754    Name: Allen Leung
2755    Date: 2002/02/21 20:20:00 EST
2756    Tag: leunga-20020221-peephole
2757    
2758    Regenerated the peephole files.  Some contained typos in the specification
2759    and some didn't compile because of pretty printing bugs in the old version
2760    of 'nowhere'.
2761    
2762    ----------------------------------------------------------------------
2763    Name: Allen Leung
2764    Date: 2002/02/19 20:20:00 EST
2765    Tag: leunga-20020219-mlrisc-tools
2766    Description:
2767    
2768       Minor bug fixes to the mlrisc-tools library:
2769    
2770       1.  Fixed up parsing colon suffixed keywords
2771       2.  Added the ability to shut the error messages up
2772       3.  Reimplemented the pretty printer and fixed up/improved
2773           the pretty printing of handle and -> types.
2774       4.  Fixed up generation of literal symbols in the nowhere tool.
2775       5.  Added some SML keywords to to sml.sty
2776    
2777    ----------------------------------------------------------------------
2778    Name: Matthias Blume
2779    Date: 2002/02/19 16:20:00 EST
2780    Tag: blume-20020219-cmffi
2781    Description:
2782    
2783    A wild mix of changes, some minor, some major:
2784    
2785    * All C FFI-related libraries are now anchored under $c:
2786        $/c.cm      --> $c/c.cm
2787        $/c-int.cm  --> $c/internals/c-int.cm
2788        $/memory.cm --> $c/memory/memory.cm
2789    
2790    * "make" tool (in CM) now treats its argument pathname slightly
2791      differently:
2792        1. If the native expansion is an absolute name, then before invoking
2793           the "make" command on it, CM will apply OS.Path.mkRelative
2794           (with relativeTo = OS.FileSys.getDir()) to it.
2795        2. The argument will be passed through to subsequent phases of CM
2796           processing without "going native".  In particular, if the argument
2797           was an anchored path, then "make" will not lose track of that anchor.
2798    
2799    * Compiler backends now "know" their respective C calling conventions
2800      instead of having to be told about it by ml-nlffigen.  This relieves
2801      ml-nlffigen from one of its burdens.
2802    
2803    * The X86Backend has been split into X86CCallBackend and X86StdCallBackend.
2804    
2805    * Export C_DEBUG and C_Debug from $c/c.cm.
2806    
2807    * C type encoding in ml-nlffi-lib has been improved to model the conceptual
2808      subtyping relationship between incomplete pointers and their complete
2809      counterparts.  For this, ('t, 'c) ptr has been changed to 'o ptr --
2810      with the convention of instantiating 'o with ('t, 'c) obj whenever
2811      the pointer target type is complete.  In the incomplete case, 'o
2812      will be instantiated with some "'c iobj" -- a type obtained by
2813      using one of the functors PointerToIncompleteType or PointerToCompleteType.
2814    
2815      Operations that work on both incomplete and complete pointer types are
2816      typed as taking an 'o ptr while operations that require the target to
2817      be known are typed as taking some ('t, 'c) obj ptr.
2818    
2819      voidptr is now a bit "more concrete", namely "type voidptr = void ptr'"
2820      where void is an eqtype without any values.  This makes it possible
2821      to work on voidptr values using functions meant to operate on light
2822      incomplete pointers.
2823    
2824    * As a result of the above, signature POINTER_TO_INCOMPLETE_TYPE has
2825      been vastly simplified.
2826    
2827    ----------------------------------------------------------------------
2828    Name: Matthias Blume
2829    Date: 2002/02/19 10:48:00 EST
2830    Tag: blume-20020219-pqfix
2831    Description:
2832    
2833    Applied Chris Okasaki's bug fix for priority queues.
2834    
2835    ----------------------------------------------------------------------
2836    Name: Matthias Blume
2837    Date: 2002/02/15 17:05:00
2838    Tag: Release_110_39
2839    Description:
2840    
2841    Last-minute retagging is becoming a tradition... :-(
2842    
2843    This is the working release 110.39.
2844    
2845    ----------------------------------------------------------------------
2846    Name: Matthias Blume
2847    Date: 2002/02/15 16:00:00 EST
2848    Tag: Release_110_39-orig
2849    Description:
2850    
2851    Working release 110.39.  New bootfiles.
2852    
2853    (Update: There was a small bug in the installer so it wouldn't work
2854    with all shells.  So I retagged. -Matthias)
2855    
2856    ----------------------------------------------------------------------
2857    Name: Matthias Blume
2858    Date: 2002/02/15 14:17:00 EST
2859    Tag: blume-20020215-showbindings
2860    Description:
2861    
2862    Added EnvRef.listBoundSymbols and CM.State.showBindings.  Especially
2863    the latter can be useful for exploring what bindings are available at
2864    the interactive prompt.  (The first function returns only the list
2865    of symbols that are really bound, the second prints those but also the
2866    ones that CM's autoloading mechanism knows about.)
2867    
2868    ----------------------------------------------------------------------
2869    Name: Matthias Blume
2870    Date: 2002/02/15 12:08:00 EST
2871    Tag: blume-20020215-iptrs
2872    Description:
2873    
2874    Two improvements to ml-nlffigen:
2875    
2876      1. Write files only if they do not exist or if their current contents
2877         do not coincide with what's being written.  (That is, avoid messing
2878         with the time stamps unless absolutely necessary.)
2879    
2880      2. Implement a "repository" mechanism for generated files related
2881         to "incomplete pointer types".   See the README file for details.
2882    
2883    ----------------------------------------------------------------------
2884    Name: Matthias Blume
2885    Date: 2002/02/14 11:50:00 EST
2886    Tag: blume-20020214-quote
2887    Description:
2888    
2889    Added a type 't t_' to tag.sml (in ml-nlffi-lib.cm).  This is required
2890    because of the new and improved tag generation scheme.  (Thanks to Allen
2891    Leung for pointing it out.)
2892    
2893    ----------------------------------------------------------------------
2894    Name: Lal George
2895    Date: 2002/02/14 09:55:27 EST 2002
2896    Tag: george-20020214-isabelle-bug
2897    Description:
2898    
2899    Fixed the MLRISC bug sent by Markus Wenzel regarding the compilation
2900    of Isabelle on the x86.
2901    
2902    From Allen:
2903    -----------
2904     I've found the problem:
2905    
2906         in ra-core.sml, I use the counter "blocked" to keep track of the
2907         true number of elements in the freeze queue.  When the counter goes
2908         to zero, I skip examining the queue.  But I've messed up the
2909         bookkeeping in combine():
2910    
2911             else ();
2912             case !ucol of
2913               PSEUDO => (if !cntv > 0 then
2914                     (if !cntu > 0 then blocked := !blocked - 1 else ();
2915                                        ^^^^^^^^^^^^^^^^^^^^^^^
2916                      moveu := mergeMoveList(!movev, !moveu)
2917                     )
2918                  else ();
2919    
2920         combine() is called to coalesce two nodes u and v.
2921         I think I was thinking that if the move counts of u and v are both
2922         greater than zero then after they are coalesced then one node is
2923         removed from the freeze queue.  Apparently I was thinking that
2924         both u and v are of low degree, but that's clearly not necessarily true.
2925    
2926    
2927    02/12/2002:
2928        Here's the patch.  HOL now compiles.
2929    
2930        I don't know how this impact on performance (compile
2931        time or runtime).  This bug caused the RA (especially on the x86)
2932        to go thru the potential spill phase when there are still nodes on the
2933        freeze queue.
2934    
2935    
2936    
2937    
2938    ----------------------------------------------------------------------
2939    Name: Matthias Blume
2940    Date: 2002/02/13 22:40:00 EST
2941    Tag: blume-20020213-fptr-rtti
2942    Description:
2943    
2944    Fixed a bug in ml-nlffigen that was introduced with one of the previous
2945    updates.
2946    
2947    ----------------------------------------------------------------------
2948    Name: Matthias Blume
2949    Date: 2002/02/13 16:41:00 EST
2950    Tag: blume-20020213-cmlpq
2951    Description:
2952    
2953    Added new priority queue export symbols (which have just been added to
2954    smlnj-lib.cm) to CML's version of smlnj-lib.cm.  (Otherwise CML would
2955    not compile and the installer would choke.)
2956    
2957    ----------------------------------------------------------------------
2958    Name: Matthias Blume
2959    Date: 2002/02/13 16:15:00 EST
2960    Tag: blume-20020213-various
2961    Description:
2962    
2963    1. More tweaks to ml-nlffigen:
2964    
2965       - better internal datastructures (resulting in slight speedup)
2966       - "-match" option requires exact match
2967       - "localized" gensym counters (untagged structs/unions nested within
2968         other structs/unions or within typedefs get a fresh counter; their
2969         tag will be prefixed by a concatenation of their parents' tags)
2970       - bug fixes (related to calculation of transitive closure of types
2971         to be included in the output)
2972    
2973    2. Minor Basis updates:
2974    
2975       - added implementations for List.collate and Option.app
2976    
2977    ----------------------------------------------------------------------
2978    Name: Matthias Blume
2979    Date: 2002/02/11 15:55:00 EST
2980    Tag: blume-20020211-gensym
2981    Description:
2982    
2983    Added a "-gensym" option to command line of ml-nlffigen.  This can be
2984    used to specify a "stem" -- a string that is inserted in all "gensym'd"
2985    names (ML structure names that correspond to unnamed C structs, unions,
2986    and enums), so that separate runs of ml-nlffigen do not clash.
2987    
2988    ----------------------------------------------------------------------
2989    Name: Matthias Blume
2990    Date: 2002/02/11 12:05:00 EST
2991    Tag: blume-20020211-gensml
2992    Description:
2993    
2994    A quick fix for a problem with GenSML (in the pgraph-util library):
2995    Make generation of toplevel "local" optional.  (Strictly speaking,
2996    signature definitions within "local" are not legal SML.)
2997    
2998    Other than that: updates to INSTALL and cm/TODO.
2999    
3000    ----------------------------------------------------------------------
3001    Name: Matthias Blume
3002    Date: 2002/02/08 15:00:00 EST
3003    Tag: blume-20020208-uniquepid
3004    Description:
3005    
3006    0. Version number has been bumped to 110.38.1.  NEW BOOTFILES!!!
3007    
3008    1. The installer (config/install.sh) has gotten smarter:
3009    
3010         - Configuration options are a bit easier to specify now
3011           (in config/targets).
3012         - Bug in recognizing .tar.bz2 files fixed.
3013         - Installer automatically resolves dependencies between
3014           configuration options (e.g., if you ask for eXene, you will
3015           also get cml -- regardless whether you asked for it or not).
3016         - Installer can run in "quieter mode" by setting the environment
3017           variable INSTALL_QUIETLY to "true".  "Quieter" does not mean
3018           "completely silent", though.
3019         - Build HashCons library as part of smlnj-lib.
3020    
3021    2. A new scheme for assigning persistent identifiers to compilation
3022       units (and, by extension, to types etc.) has been put into place.
3023       This fixes a long-standing bug where types and even dynamic values
3024       can get internally confused, thereby compromising type safety
3025       (abstraction) and dynamic correctness.  See
3026    
3027         http://cm.bell-labs.com/cm/cs/who/blume/pid-confusion.tgz
3028    
3029       for an example of how things could go wrong until now.
3030    
3031       The downside of the new scheme is that pids are not quite as
3032       persistent as they used to be: CM will generate a fresh pid
3033       for every compilation unit that it thinks it sees for the first
3034       time.  That means that if you compile starting from a clean, fresh
3035       source tree at two different times, you end up with different
3036       binaries.
3037    
3038       Cutoff recompilation, however, has not been compromised because
3039       CM keeps pid information in special caches between runs.
3040    
3041    ----------------------------------------------------------------------
3042    Name: Lal George
3043    Date: 2002/02/07 15:34:13 EST 2002
3044    Tag: <none>
3045    Description:
3046    
3047    Compilers that generate assembly code may produce  global labels
3048    whose value is resolved at link time. The various peephole optimization
3049    modules did not take this in account.
3050    
3051    TODO. The Labels.addrOf function should really return an option
3052    type so that clients are forced to deal with this issue, rather
3053    than an exception being raised.
3054    
3055    ----------------------------------------------------------------------
3056    Name: Lal George
3057    Date: 2002/02/06 13:55:02 EST
3058    Tag: george-20020206-ra-breakup
3059    Description:
3060    
3061    1. A bug fix from Allen.
3062    
3063        A typo causes extra fstp %st(0)'s to be generated at compensation
3064        edges, which might cause stack underflow traps at runtime.  This
3065        occurs in fft where there are extraneous fstps right before the 'into'
3066        trap instruction (in this case they are harmless since none of the
3067        integers overflow.)
3068    
3069    2. Pulled out various utility modules that were embedded in the modules
3070       of the register allocator. I need these modules for other purposes, but
3071       they are not complete enough to put into a library (just yet).
3072    ----------------------------------------------------------------------
3073    Name: Matthias Blume
3074    Date: 2002/01/31 16:05:00 EST
3075    Tag: blume-20020131-sparc-ccalls
3076    Description:
3077    
3078    1. C-calls on Sparc needlessly allocated a huge chunk (96 bytes)
3079       of extra stack space by mistake.  Fixed.
3080    
3081    2. Bug in logic of handling of command-line options in ml-nlffigen fixed.
3082    
3083    ----------------------------------------------------------------------
3084    Name: Allen Leung
3085    Date: 2002/01/30
3086    Tag: leunga-20020130-nowhere-bug-fix
3087    Description:
3088    
3089       MLRISC bug fixes:
3090       1. Fixed a bindings computation bug in the 'nowhere' program generator tool.
3091       2. MachineInt.fromString was negating its value.
3092    
3093    ----------------------------------------------------------------------
3094    Name: Matthias Blume
3095    Date: 2002/01/29
3096    Tag: blume-20020129-INSTALL
3097    Description:
3098    
3099    - Added somewhat detailed installation instructions (file INSTALL).
3100    - Fixed curl-detection bug in config/install.sh.
3101    - It is now possible to select the URL getter using the URLGETTER
3102      environment variable:
3103    
3104          not set / "unknown"      --> automatic detection (script tries wget,
3105                                       curl, and lynx)
3106          "wget" / "curl" / "lynx" --> use the specified program (script "knows"
3107                                       how to properly invoke them)
3108          other                    --> use $URLGETTER directly, it must take
3109                                       precisely two command-line arguments
3110                                       (source URL and destination file name)
3111    
3112    ----------------------------------------------------------------------
3113    Name: Matthias Blume
3114    Date: 2002/01/28
3115    Tag: blume-20020128-sparc-ccalls
3116    Description:
3117    
3118    - Fixed problem with calculation of "used" registers in sparc-c-calls.
3119    - Make use of the allocParam argument in sparc-c-calls.
3120    
3121    ----------------------------------------------------------------------
3122    Name: Matthias Blume
3123    Date: 2002/01/28
3124    Tag: blume-20020128-allocParam
3125    Description:
3126    
3127    John Reppy:  Changes c-calls API to accept client-callback for
3128    allocating extra stack space.
3129    me: Corresponding changes to mlriscGen (using a dummy argument that
3130        does not change the current behavior).
3131    
3132    ----------------------------------------------------------------------
3133    Name: Matthias Blume
3134    Date: 2002/01/28 12:00:00
3135    Tag: Release_110_38
3136    Description:
3137    
3138    This time for real!!!
3139    
3140    ----------------------------------------------------------------------
3141    Name: Matthias Blume
3142    Date: 2002/01/28 10:56:00 EST
3143    Tag: blume-20020128-retraction
3144    Description:
3145    
3146    0. Retracted earlier 110.38.  (The Release_110_38 tag has been replaced
3147       with blume-Release_110_38-retracted.)
3148    
3149    1. Fixed a problem with incorrect rounding modes in real64.sml.
3150       (Thanks to Andrew Mccreight <andrew.mccreight@yale.edu>.)
3151    
3152    2. A bug in ml-nlffigen related to the handling of unnamed structs, unions,
3153       and enums fixed.  The naming of corresponding ML identifiers should
3154       now be consistent again.
3155    
3156    ----------------------------------------------------------------------
3157    Name: Allen Leung
3158    Date: 2002/01/27
3159    Tag: leunga-20020127-nowhere
3160    Description:
3161    
3162       Added a target called nowhere in the configuration scripts.
3163       Enabling this will build the MLRISC 'nowhere' tool (for translating
3164       programs with where-clauses into legal SML code) during installation.
3165    
3166    ----------------------------------------------------------------------
3167    Name: Matthias Blume
3168    Date: 2002/01/25 21:27:00 EST
3169    Tag: blume-Release_110_38-retracted
3170    Description:
3171    
3172    Call it a (working) release!  Version is 110.38. Bootfiles are ready.
3173    
3174    README will be added later.
3175    
3176    !!! NOTE:  Re-tagged as blume-Release_110_38-retracted. Original tag
3177    (Release_110_38) removed.  Reason: Last-minute bug fixes.
3178    
3179    ----------------------------------------------------------------------
3180    Name: Matthias Blume
3181    Date: 2002/01/25
3182    Tag: blume-20020125-ffi
3183    Description:
3184    
3185    A large number of tweaks and improvements to ml-nlffi-lib and
3186    ml-nlffigen:
3187    
3188       - ML represenation types have been streamlined
3189       - getter and setter functions work with concrete values, not abstract
3190         ones where possible
3191       - ml-nlffigen command line more flexible (see README file there)
3192       - some bugs have been fixed (hopefully)
3193    
3194    ----------------------------------------------------------------------
3195    Name: Lal George
3196    Date: 2002/01/24
3197    Tag: george-20020124-risc-ra-interface
3198    Description:
3199    
3200       There is a dramatic simplification in the interface to the
3201       register allocator for RISC architectures as a result of making
3202       parallel copy instructions explicit.
3203    
3204    ----------------------------------------------------------------------
3205    Name: Matthias Blume
3206    Date: 2002/01/22
3207    Tag: blume-20020122-x86-ccalls
3208    Description:
3209    
3210    Bug fix for c-calls on x86 (having to do with how char- and
3211    short-arguments are being handled).
3212    
3213    ----------------------------------------------------------------------
3214    Name: Matthias Blume
3215    Date: 2002/01/21
3216    Tag: blume-20020121-ff
3217    Description:
3218    
3219    Another day of fiddling with the FFI...
3220    
3221    1. Bug fix/workaround:  CKIT does not complain about negative array
3222       dimensions, so ml-nlffigen has to guard itself against this possibility.
3223       (Otherwise a negative dimension would send it into an infinite loop.)
3224    
3225    2. Some of the abstract types (light objects, light pointers, most "base"
3226       types) in structure C are now eqtypes.
3227    
3228    3. Added constructors and test functions for NULL function pointers.
3229    
3230    ----------------------------------------------------------------------
3231    Name: Matthias Blume
3232    Date: 2002/01/18
3233    Tag: blume-20020118-ready-for-new-release
3234    Description:
3235    
3236    Made config/srcarchiveurl point to a new place.  (Will provide boot
3237    files shortly.)
3238    
3239    Maybe we christen this to be 110.38?
3240    
3241    ----------------------------------------------------------------------
3242    Name: Matthias Blume
3243    Date: 2002/01/18
3244    Tag: blume-20020118-more-ffifiddle
3245    Description:
3246    
3247    Today's FFI fiddling:
3248    
3249      - Provided a structure CGetSet with "convenient" versions of C.Get.* and
3250        C.Set.* that use concrete (MLRep.*) arguments and results instead
3251        of abstract ones.
3252    
3253      - Provided word-style bit operations etc. for "int" representation
3254        types in MLRep.S<Foo>Bitops where <Foo> ranges over Char, Int, Short,
3255        and Long.
3256    
3257    ----------------------------------------------------------------------
3258    Name: Matthias Blume
3259    Date: 2002/01/18
3260    Tag: blume-20020118-use-x86-fp
3261    Description:
3262    
3263    Now that x86-fast-fp seems to be working, I turned it back on again
3264    by default.  (Seems to work fine now, even with the FFI.)
3265    
3266    Other than that, I added some documentation about the FFI to
3267    src/ml-nlffigen/README and updated the FFI test examples in
3268    src/ml-nlffi-lib/Tests/*.
3269    
3270    ----------------------------------------------------------------------
3271    Name: Allen Leung
3272    Date: 2002/01/17
3273    Tag: leunga-20020117-x86-fast-fp-call
3274    Description:
3275    
3276       1. Fixed a problem with handling return fp values when x86's fast fp
3277          mode is turned on.
3278    
3279       2. Minor pretty printing fix for cellset.  Print %st(0) as %st(0) instead
3280          of %f32.
3281    
3282       3. Added a constructor INT32lit to the ast of MLRISC tools.
3283    
3284    ----------------------------------------------------------------------
3285    Name: Matthias Blume
3286    Date: 2002/01/16
3287    Tag: blume-20020116-ffifiddle
3288    Description:
3289    
3290    More fiddling with the FFI interface:
3291    
3292     - Make constness 'c instead of rw wherever possible.  This eliminates
3293       the need for certain explicit coercions.  (However, due to ML's
3294       value polymorphism, there will still be many cases where explicit
3295       coercions are necessary.  Phantom types are not the whole answer
3296       to modeling a subtyping relationship in ML.)
3297    
3298     - ro/rw coersions for pointers added.  (Avoids the detour through */&.)
3299    
3300     - "printf" test example added to src/ml-nlffi-lib/Tests.  (Demonstrates
3301       clumsy workaround for varargs problem.)
3302    
3303    ----------------------------------------------------------------------
3304    Name: Lal George
3305    Date: 2002/01/15
3306    Tag: <none>
3307    Description:
3308    
3309    1. Since COPY instructions are no longer native to the architecture,
3310       a generic functor can be used to implement the expandCopies function.
3311    
3312    2. Allowed EXPORT and IMPORT pseudo-op declarations to appear inside a
3313       TEXT segment.
3314    
3315    ----------------------------------------------------------------------
3316    Name: Matthias Blume
3317    Date: 2002/01/15
3318    Tag: blume-20020115-ffiupdates
3319    Description:
3320    
3321    1. Fix for bug resulting in single-precision float values being returned
3322       incorrectly from FFI calls.
3323    
3324    2. Small modifications to C FFI API:
3325    
3326        - memory-allocation routines return straight objects (no options)
3327          and raise an exception in out-of-memory situations
3328        - unsafe extensions to cast between function pointers and pointers
3329          from/to ints
3330        - added structure C_Debug as an alternative to structure C where
3331          pointer-dereferencing (|*| and |*!) always check for null-pointers
3332        - added open_lib' to DynLinkage;  open_lib' works like open_lib
3333          but also takes a (possibly empty) list of existing library handles
3334          that the current library depends on
3335    
3336    ----------------------------------------------------------------------
3337    Name: Matthias Blume
3338    Date: 2002/01/10
3339    Tag: blume-20020110-newffigen
3340    Description:
3341    
3342    1. Updates to portable graph code.
3343    
3344    2. Major update to ml-nlffigen and ml-nlffi-lib.  Things are much
3345       more scalable now so that even huge interfaces such as the one
3346       for GTK compile in finite time and space. :-)
3347       See src/ml-nlffigen/README for details on what's new.
3348    
3349    ----------------------------------------------------------------------
3350    Name: Lal George
3351    Date: 2001/01/09 14:31:35 EST 2002
3352    Tag: george-20011206-rm-native-copy
3353    Description:
3354    
3355            Removed the native COPY and FCOPY instructions
3356            from all the architectures and replaced it with the
3357            explicit COPY instruction from the previous commit.
3358    
3359            It is now possible to simplify many of the optimizations
3360            modules that manipulate copies. This has not been
3361            done in this change.
3362    
3363    ----------------------------------------------------------------------
3364    Name: Lal George
3365    Date: 2001/12/06 16:50:13 EST 2001
3366    Tag: george-20011206-mlrisc-instruction
3367    Description:
3368    
3369    Changed the representation of instructions from being fully abstract
3370    to being partially concrete. That is to say:
3371    
3372      from
3373            type instruction
3374    
3375      to
3376            type instr                              (* machine instruction *)
3377    
3378            datatype instruction =
3379                LIVE of {regs: C.cellset, spilled: C.cellset}
3380              | KILL of {regs: C.cellset, spilled: C.cellset}
3381              | COPYXXX of {k: CB.cellkind, dst: CB.cell list, src: CB.cell list}
3382              | ANNOTATION of {i: instruction, a: Annotations.annotation}
3383              | INSTR of instr
3384    
3385    This makes the handling of certain special instructions that appear on
3386    all architectures easier and uniform.
3387    
3388    LIVE and KILL say that a list of registers are live or killed at the
3389    program point where they appear. No spill code is generated when an
3390    element of the 'regs' field is spilled, but the register is moved to
3391    the 'spilled' (which is present, more for debugging than anything else).
3392    
3393    LIVE replaces the (now deprecated) DEFFREG instruction on the alpha.
3394    We used to generate:
3395    
3396            DEFFREG f1
3397            f1 := f2 + f3
3398            trapb
3399    
3400    but now generate:
3401    
3402            f1 := f2 + f3
3403            trapb
3404            LIVE {regs=[f1,f2,f3], spilled=[]}
3405    
3406    Furthermore, the DEFFREG (hack) required that all floating point instruction
3407    use all registers mentioned in the instruction. Therefore f1 := f2 + f3,
3408    defines f1 and uses [f1,f2,f3]! This hack is no longer required resulting
3409    in a cleaner alpha implementation. (Hopefully, intel will not get rid of
3410    this architecture).
3411    
3412    COPYXXX is intended to replace the parallel COPY and FCOPY  available on
3413    all the architectures. This will result in further simplification of the
3414    register allocator that must be aware of them for coalescing purposes, and
3415    will also simplify certain aspects of the machine description that provides
3416    callbacks related to parallel copies.
3417    
3418    ANNOTATION should be obvious, and now INSTR represents the honest to God
3419    machine instruction set!
3420    
3421    The <arch>/instructions/<arch>Instr.sml files define certain utility
3422    functions for making porting easier -- essentially converting upper case
3423    to lower case. All machine instructions (of type instr) are in upper case,
3424    and the lower case form generates an MLRISC instruction. For example on
3425    the alpha we have:
3426    
3427      datatype instr =
3428         LDA of {r:cell, b:cell, d:operand}
3429       | ...
3430    
3431      val lda : {r:cell, b:cell, d:operand} -> instruction
3432        ...
3433    
3434    where lda is just (INSTR o LDA), etc.
3435    
3436    ----------------------------------------------------------------------
3437    Name: Matthias Blume
3438    Date: 2001/11/22 21:40:00 EST
3439    Tag: Release_110_37
3440    Description:
3441    
3442    Release 110.37.  This time for real.
3443    
3444    ----------------------------------------------------------------------
3445    Name: Matthias Blume
3446    Date: 2001/11/21 16:35:00 EST
3447    Tag: blume-20011121-foot-in-mouth
3448    Description:
3449    
3450    Removed the "Release_110_37" tag because of a serious bug.
3451    This will be re-tagged once the bug is fixed.
3452    
3453    ----------------------------------------------------------------------
3454    Name: Matthias Blume
3455    Date: 2001/11/21 16:14:00 EST
3456    Tag: blume-20011121-forgottenfile
3457    Description:
3458    
3459    Forgot to add a file.  (Just a .tex-file -- part of
3460    the CM manual source.)
3461    
3462    ----------------------------------------------------------------------
3463    Name: Matthias Blume
3464    Date: 2001/11/21 16:10:00 EST
3465    Tag: blume-20011121-invalid_110_37
3466    Description:
3467    
3468    Note: I removed the original tag "Release_110_37" from this commit
3469          because we found a serious bug in all non-x86 backends.
3470          - Matthias
3471    
3472    1. Modifications to the SML/NJ code generator and to the runtime system
3473       so that code object name strings are directly inserted into code
3474       objects at code generation time.  The only business the runtime system
3475       has with this is now to read the name strings on occasions.
3476       (The encoding of the name string has also changed somewhat.)
3477    
3478    2. CM now implements a simple "set calculus" for specifying export lists.
3479       In particular, it is now possible to refer to the export lists of
3480       other libraries/groups/sources and form unions as well as differences.
3481       See the latest CM manual for details.
3482    
3483    3. An separate notion of "proxy" libraries has again be eliminated from
3484       CM's model.  (Proxy libraries are now simply a special case of using
3485       the export list calculus.)
3486    
3487    4. Some of the existing libraries now take advantage of the new set
3488       calculus.
3489       (Notice that not all libraries have been converted because some
3490       of the existing .cm-files are supposed to be backward compatible
3491       with 110.0.x.)
3492    
3493    5. Some cleanup in stand-alone programs.  (Don't use "exnMessage" -- use
3494       "General.exnMessage"!  The former relies on a certain hook to be
3495       initialized, and that often does not happen in the stand-alone case.)
3496    
3497    ----------------------------------------------------------------------
3498    Name: Lal George
3499    Date: 2001/11/21  13:56:18 EST
3500    Tag: george-2001121-pseudo-ops
3501    Description:
3502    
3503      Implemented a complete redesign of MLRISC pseudo-ops. Now there
3504      ought to never be any question of incompatabilities with
3505      pseudo-op syntax expected by host assemblers.
3506    
3507      For now, only modules supporting GAS syntax are implemented
3508      but more should follow, such as MASM, and vendor assembler
3509      syntax, e.g. IBM as, Sun as, etc.
3510    
3511    ----------------------------------------------------------------------
3512    Name: Matthias Blume
3513    Date: 2001/11/14 11:52:00 EST
3514    Tag: blume-20011114-srcname
3515    Description:
3516    
3517    1. Routed the name of the current source file to mlriscgen where it
3518       should be directly emitted into the code object.  (This last part
3519       is yet to be done.)
3520    
3521    2. Some cleanup of the pgraph code to make it match the proposal that
3522       I put out the other day.  (The proposal notwithstanding, things are
3523       still in flux here.)
3524    
3525    ----------------------------------------------------------------------
3526    Name: Lal George
3527    Date: 2001/11/14 09:44:04 EST
3528    Tag:
3529    Description:
3530    
3531      Fix for a backpatching bug reported by Allen.
3532    
3533      Because the boundary between short and long span-dependent
3534      instructions is +/- 128, there are an astounding number of
3535      span-dependent instructions whose size is over estimated.
3536    
3537      Allen came up with the idea of letting the size of span
3538      dependent instructions be non-monotonic, for a maxIter
3539      number of times, after which the size must be monotonically
3540      increasing.
3541    
3542      This table shows the number of span-dependent instructions
3543      whose size was over-estimated as a function of maxIter, for the
3544      file Parse/parse/ml.grm.sml:
3545    
3546         maxIter            # of instructions:
3547            10                      687
3548            20                      438
3549            30                      198
3550            40                        0
3551    
3552      In compiling the compiler, there is no significant difference in
3553      compilation speed between maxIter=10 and maxIter=40. Actually,
3554      my measurements showed that maxIter=40 was a tad faster than
3555      maxIter=10! Also 96% of the  files in the compiler reach a fix
3556      point within 13 iterations, so fixing maxIter at 40, while high,
3557      is okay.
3558    
3559    ----------------------------------------------------------------------
3560    Name: Matthias Blume
3561    Date: 2001/10/31 15:25:00 EST
3562    Tag: blume-20011031-pgraph
3563    Description:
3564    
3565    CKIT:
3566    * Changed the "Function" constructor of type Ast.ctype to carry optional
3567      argument identifiers.
3568    * Changed the return type of TypeUtil.getFunction accordingly.
3569    * Type equality ignores the argument names.
3570    * TypeUtil.composite tries to preserve argument names but gives up quickly
3571      if there is a mismatch.
3572    
3573    installation script:
3574    * attempts to use "curl" if available (unless "wget" is available as well)
3575    
3576    CM:
3577    * has an experimental implementation of "portable graphs" which I will
3578      soon propose as an implementation-independent library format
3579    * there are also new libraries $/pgraph.cm and $/pgraph-util.cm
3580    
3581    NLFFI-LIB:
3582    * some cleanup (all cosmetic)
3583    
3584    NLFFIGEN:
3585    * temporarily disabled the mechanism that suppresses ML output for
3586      C definitions whose identifiers start with an underscore character
3587    * generate val bindings for enum constants
3588    * user can request that only one style (light or heavy) is being used;
3589      default is to use both (command-line arguments: -heavy and -light)
3590    * fixed bug in handling of function types involving incomplete pointers
3591    * generate ML entry points that take record arguments (i.e., using
3592      named arguments) for C functions that have a prototype with named
3593      arguments
3594      (see changes to CKIT)
3595    
3596    ----------------------------------------------------------------------
3597    Name: Allen Leung
3598    Date: 2001/10/27 20:34:00 EDT
3599    Tag: leunga-20011027-x86-fast-fp-call
3600    Description:
3601    
3602       Fixed the bug described in blume-20010920-slowfp.
3603    
3604       The fix involves
3605          1. generating FCOPYs in FSTP in ia32-svid
3606          2. marking a CALL with the appropriate annotation
3607    
3608    ----------------------------------------------------------------------
3609    Name: Matthias Blume
3610    Date: 2001/10/16 11:32:00 EDT
3611    Tag: blume-20011016-netbsd
3612    Description:
3613    
3614    Underscore patch from Chris Richards (fixing problem with compiling
3615    runtime system under recent NetBSD).
3616    
3617    ----------------------------------------------------------------------
3618    Name: Allen Leung
3619    Date: 2001/10/12 17:18:32 EDT 2001
3620    Tag: leung-20011012-x86-printflowgraph
3621    Description:
3622    
3623    X86RA now uses a valid (instead of dummy) PrintFlowgraph module.
3624    
3625    ----------------------------------------------------------------------
3626    Name: Lal George
3627    Date: 2001/10/11 23:51:34 EDT
3628    Tag: george-20011011-too-many-instrs
3629    Description:
3630    
3631    The representation of a program point never expected to see more
3632    than 65536 instructions in a basic block!
3633    
3634    ----------------------------------------------------------------------
3635    Name: Lal George
3636    Date: 2001/10/09 09:41:37 EDT
3637    Tag: george-20011008-mlrisc-labels
3638    Description:
3639    
3640    Changed the machine description files to support printing of
3641    local and global labels in assembly code, based on host assembler
3642    conventions.
3643    
3644    ----------------------------------------------------------------------
3645    Name: Matthias Blume
3646    Date: 2001/09/25 15:25:00 EDT
3647    Tag: blume-20010925-exninfo
3648    Description:
3649    
3650    I provided a non-hook implementation of exnName (at the toplevel) and
3651    made the "dummy" implementation of exnMessage (at the toplevel) more
3652    useful: if nothing gets "hooked in", then at least you are going to
3653    see the exception name and a message indicating why you don't see more.
3654    
3655    [For the time being, programs that need exnMessage and want to use
3656    ml-build should either use General.exnMessage (strongly recommended) or
3657    refer to structure General at some other point so that CM sees a
3658    static dependency.]
3659    
3660    [Similar remarks go for "print" and "use":  If you want to use their
3661    functionality in stand-alone programs generated by ml-build, then use
3662    TextIO.output and Backend.Interact.useFile (from $smlnj/compiler.cm).]
3663    
3664    ----------------------------------------------------------------------
3665    Name: Matthias Blume
3666    Date: 2001/09/20 17:28:00 EDT
3667    Tag: blume-20010920-slowfp
3668    Description:
3669    
3670    Allen says that x86-fast-fp is not safe yet, so I turned it off again...
3671    
3672    ----------------------------------------------------------------------
3673    Name: Matthias Blume
3674    Date: 2001/09/20 17:20:00 EDT
3675    Tag: blume-20010920-canonicalpaths
3676    Description:
3677    
3678    0. Updated the BOOT file (something that I forgot to do earlier).
3679    
3680    1. Small internal change to CM so that it avoids "/../" in filenames
3681       as much as possible (but only where it is safe).
3682    
3683    2. Changed config/_run-sml (resulting in a changed bin/.run-sml) so
3684       that arguments that contain delimiters are passed through correctly.
3685       This change also means that all "special" arguments of the form
3686       @SMLxxx... must come first.
3687    
3688    3. Changed install script to put relative anchor names for tool commands
3689       into pathconfig.
3690    
3691    ----------------------------------------------------------------------
3692    Name: Matthias Blume
3693    Date: 2001/09/18 15:35:00 EDT
3694    Tag: blume-20010918-readme11036
3695    Description:
3696    
3697    Added README files.
3698    
3699    ----------------------------------------------------------------------
3700    Name: Matthias Blume
3701    Date: 2001/09/18 11:45:00 EDT
3702    Tag: Release_110_36 (retag)
3703    Description:
3704    
3705    Fixed mistake in config/preloads. Retagged as 110.36.
3706    
3707    ----------------------------------------------------------------------
3708    Name: Matthias Blume
3709    Date: 2001/09/18 09:40:00 EDT
3710    Tag: Release_110_36_orig (tag changed)
3711    Description:
3712    
3713    New version (110.36).  New bootfiles.
3714    
3715    ----------------------------------------------------------------------
3716    Name: Matthias Blume
3717    Date: 2001/09/14 16:15:00 EDT
3718    Tag: blume-20010914-x86fastfp
3719    Description:
3720    
3721    John committed some changes that Allen made, in particular a (hopefully)
3722    correctly working version of the x86-fp module.
3723    
3724    I changed the default setting of the Control.MLRISC.getFlag "x86-fast-fp"
3725    flag to "true".  Everything seems to compile to a fixpoint ok, and
3726    "mandelbrot" speeds up by about 15%.
3727    
3728    ----------------------------------------------------------------------
3729    Name: Matthias Blume
3730    Date: 2001/09/13 11:20:00 EDT
3731    Tag: blume-20010913-minimal
3732    Description:
3733    
3734    1. Stefan Monnier's patch to fix a miscompilation problem that
3735       was brought to light by John Reppy's work on Moby.
3736    
3737    2. Implemented a minimal "structure Compiler" that contains just
3738       "version" and "architecture".  The minimal version will be
3739       available when the full version is not.  This is for backward-
3740       compatibility with code that wants to test Compiler.version.
3741    
3742    ----------------------------------------------------------------------
3743    Name: Matthias Blume
3744    Date: 2001/08/28 14:03:00 EDT
3745    Tag: blume-20010828-ml-lex
3746    Description:
3747    
3748    Fix for bug 1581, received from Neophytos Michael.
3749    
3750    ----------------------------------------------------------------------
3751    Name: Matthias Blume
3752    Date: 2001/08/27 11:20:00 EDT
3753    Tag: blume-20010827-readme11035
3754    Description:
3755    
3756    Fleshed out the README file for 110.35.
3757    
3758    ----------------------------------------------------------------------
3759    Name: Matthias Blume
3760    Date: 2001/08/24 17:10:00 EDT
3761    Tag: Release_110_35
3762    Description:
3763    
3764    New version number (110.35).  New bootfiles.
3765    
3766    ----------------------------------------------------------------------
3767    Name: Lal George
3768    Date: 2001/08/24 13:47:18 EDT 2001
3769    Tag: george-20010824-MLRISC-graphs
3770    Description:
3771    
3772     removed clusters from MLRISC completely and replaced with graphs.
3773    
3774    ----------------------------------------------------------------------
3775    Name: Matthias Blume
3776    Date: 2001/08/23 17:50:00 EDT
3777    Tag: blume-20010823-toplevel
3778    Description:
3779    
3780    - some reorganization of the code that implements various kinds of
3781      environments in the compiler (static, dynamic, symbolic, combined)
3782    - re-implemented the EnvRef module so that evalStream works properly
3783      (if the stream contains references to "use", "CM.make", etc.)
3784    - cleaned up evalloop.sml and interact.sml (but they need more cleaning)
3785    
3786    ----------------------------------------------------------------------
3787    Name: Matthias Blume
3788    Date: 2001/08/20 15:50 EDT
3789    Tag: blume20010820-slipup
3790    Description:
3791    
3792    I forgot to commit a few files.  Here they are...
3793    
3794    ----------------------------------------------------------------------
3795    Name: Matthias Blume
3796    Date: 2001/08/20 15:35:00 EDT
3797    Tag: blume-20010820-debugprof
3798    Description:
3799    
3800    !!!! NEW BOOTFILES !!!!
3801    
3802    This is another round of reorganizing the compiler sources.  This
3803    time the main goal was to factor out all the "instrumentation"
3804    passes (for profiling and backtracing) into their own library.
3805    The difficulty was to do it in such a way that it does not depend
3806    on elaborate.cm but only on elabdata.cm.
3807    
3808    Therefore there have been further changes to both elaborate.cm and
3809    elabdata.cm -- more "generic" things have been moved from the former
3810    to the latter.  As a result, I was forced to split the assignment
3811    of numbers indicating "primtyc"s into two portions: SML-generic and
3812    SML/NJ-specific.  Since it would have been awkward to maintain,
3813    I bit the bullet and actually _changed_ the mapping between these
3814    numbers and primtycs.  The bottom line of this is that you need
3815    a new set of bin- and bootfiles.
3816    
3817    I have built new bootfiles for all architectures, so doing a fresh
3818    checkout and config/install.sh should be all you need.
3819    
3820    The newly created library's name is
3821    
3822        $smlnj/viscomp/debugprof.cm
3823    
3824    and its sources live under
3825    
3826        src/compiler/DebugProf
3827    
3828    ----------------------------------------------------------------------
3829    Name: Matthias Blume
3830    Date: 2001/08/15 17:15:00 EDT
3831    Tag: blume-20010815-compreorg
3832    Description:
3833    
3834    This is a first cut at reorganizing the CM libraries that make up the
3835    core of the compiler.  The idea is to separate out pieces that could
3836    be used independently by tools, e.g., the parser, the typechecker, etc.
3837    
3838    The current status is a step in this direction, but it is not quite
3839    satisfactory yet.  Expect more changes in the future.
3840    
3841    Here is the current (new) organization...
3842    
3843        What used to be $smlnj/viscomp/core.cm is now divided into
3844        six CM libraries:
3845    
3846             $smlnj/viscomp/basics.cm
3847                           /parser.cm
3848                           /elabdata.cm
3849                           /elaborate.cm
3850                           /execute.cm
3851                           /core.cm
3852    
3853        The CM files for these libraries live under src/system/smlnj/viscomp.
3854        All these libraries are proxy libraries that contain precisely
3855        one CM library component.  Here are the locations of the components
3856        (all within the src/compiler tree):
3857    
3858             Basics/basics.cm
3859             Parse/parser.cm
3860             ElabData/elabdata.cm
3861             Elaborator/elaborate.cm
3862             Execution/execute.cm
3863             core.cm
3864    
3865         [This organization is the same that has been used already
3866         for a while for the architecture-specific parts of the visible
3867         compiler and for the old version of core.cm.]
3868    
3869         As you will notice, many source files have been moved from their
3870         respective original locations to a new home in one of the above
3871         subtrees.
3872    
3873         The division of labor between the new libraries is the following:
3874    
3875             basics.cm:
3876                - Simple, basic definitions that pertain to many (or all) of
3877                  the other libraries.
3878             parser.cm:
3879                - The SML parser, producing output of type Ast.dec.
3880                - The type family for Ast is also defined and exported here.
3881             elabdata.cm:
3882                - The datatypes that describe input and output of the elaborator.
3883                  This includes types, absyn, and static environments.
3884             elaborator.cm:
3885                - The SML/NJ type checker and elaborator.
3886                  This maps an Ast.dec (with a given static environment) to
3887                  an Absyn.dec (with a new static environment).
3888                - This libraries implements certain modules that used to be
3889                  structures as functors (to remove dependencies on FLINT).
3890             execute.cm:
3891                - Everything having to do with executing binary code objects.
3892                - Dynamic environments.
3893             core.cm:
3894                - SML/NJ-specific instantiations of the elaborator and MLRISC.
3895                - Top-level modules.
3896                - FLINT (this should eventually become its own library)
3897    
3898    Notes:
3899    
3900    I am not 100% happy with the way I separated the elaborator (and its
3901    data structures) from FLINT.  Two instances of the same problem:
3902    
3903        1. Data structures contain certain fields that carry FLINT-specific
3904           information.  I hacked around this using exn and the property list
3905           module from smlnj-lib.  But the fact that there are middle-end
3906           specific fields around at all is a bit annoying.
3907    
3908        2. The elaborator calculates certain FLINT-related information.  I tried
3909           to make this as abstract as I could using functorization, but, again,
3910           the fact that the elaborator has to perform calculations on behalf
3911           of the middle-end at all is not nice.
3912    
3913        3. Having to used exn and property lists is unfortunate because it
3914           weakens type checking.  The other alternative (parameterizing
3915           nearly *everything*) is not appealing, though.
3916    
3917    I removed the "rebinding =" warning hack because due to the new organization
3918    it was awkward to maintain it.  As a result, the compiler now issues some of
3919    these warnings when compiling init.cmi during bootstrap compilation. On
3920    the plus side, you also get a warning when you do, for example:
3921       val op = = Int32.+
3922    which was not the case up to now.
3923    
3924    I placed "assign" and "deref" into the _Core structure so that the
3925    code that deals with the "lazy" keyword can find them there.  This
3926    removes the need for having access to the primitive environment
3927    during elaboration.
3928    
3929    ----------------------------------------------------------------------
3930    Name: Matthias Blume
3931    Date: 2001/08/13
3932    Tag: blume-20010813-closures
3933    Description:
3934    
3935    This fix was sent to us by Zhong Shao.  It is supposed to improve the
3936    performance of certain loops by avoiding needless closure allocation.
3937    
3938    ----------------------------------------------------------------------
3939    Name: Lal George
3940    Date: 2001/07/31 10:03:23 EDT 2001
3941    Tag: george-20010731-x86-fmalloc
3942    Description: Fixed bug in x86 calls
3943    
3944        There was a bug where call instructions would mysteriously
3945        vanish. The call instruction had to be one that returned
3946        a floating point value.
3947    
3948    ----------------------------------------------------------------------
3949  Name: Lal George  Name: Lal George
3950  Date: 2001/07/19 16:36:29 EDT 2001  Date: 2001/07/19 16:36:29 EDT 2001
3951  Tag: george-20010719-simple-cells  Tag: george-20010719-simple-cells
# Line 2733  Line 6666 
6666       elaborator). There were a lot of changes during my "linkpath" trials       elaborator). There were a lot of changes during my "linkpath" trials
6667       that could have been reverted to their original state but weren't.       that could have been reverted to their original state but weren't.
6668       Please, don't be too harsh on me for messing with this code a bit more       Please, don't be too harsh on me for messing with this code a bit more
6669       than what was strictly necessary...  (I _did_ resist the tempation       than what was strictly necessary...  (I _did_ resist the temptation
6670       of doing any "global reformatting" to avoid an untimely death at       of doing any "global reformatting" to avoid an untimely death at
6671       Dave's hands. :)       Dave's hands. :)
6672    

Legend:
Removed from v.890  
changed lines
  Added in v.1644

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