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

SCM Repository

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

Diff of /sml/trunk/HISTORY

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

revision 657, Fri Jun 9 05:20:54 2000 UTC revision 1653, Fri Oct 15 20:45:02 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: 2004/10/15 15:45:00 CDT
18    Tag: blume-20041015-coverage
19    Description:
20    
21    * Test coverage tool added!
22    * Further reorganization of tracing-, debugging-, and profiling support:
23    
24        - moved original BTImp -- now called BackTrace -- into a separate
25          library called $/trace-debug-profile.cm
26        - eliminated all mentions of BTrace from SMLofNJ.Internals
27        - only the instrumentation mechanism is now left in the compiler proper
28        - BackTrace module is a plugin which is NOT plugged in by default
29        - Coverage module is another such plugin
30    
31    To get the benefits of any of these plugin modules, the code in
32    question must be compiled with tdp instrumentation turned on.  This
33    can be done by setting SMLofNJ.Internals.TDP.mode to true.
34    (The ref cell is also controlled via the -Ctdp.instrument=... switch.)
35    
36    Plugins are selected at link time.  (Pre-compiled instrumented code
37    can be re-loaded with different plugins in effect.)  When an
38    instrumented module is linked, whatever plugins are at that time
39    enabled will come into effect for that module.
40    
41    To enable the back-trace plugin, load library $/trace-debug-profile.cm
42    and invoke BackTrace.install() (e.g., from the interactive prompt).
43    To enable the coverage plugin, load the same library and invoke
44    Coverage.install().
45    
46    Back-traces are generated automatically on uncaught exceptions and
47    when the code in question explicitly invokes BackTrace.trigger().
48    
49    Coverage (and execution frequency-) information must be queried
50    explicitly by calling Coverage.not_covered and Coverage.hot_spots.
51    
52    ----------------------------------------------------------------------
53    Name: Matthias Blume (blume (at) tti - c (dot) org)
54    Date: 2004/10/14 17:40:00 CDT
55    Tag: blume-20041014-tdp-core
56    Description:
57    
58    Snapshot of a significant overhaul of how the trace/debug/profile support
59    is hooked into the system (specifically: Core and SMLofNJ.Internals).
60    
61    ----------------------------------------------------------------------
62    Name: Matthias Blume (blume (at) tti - c (dot) org)
63    Date: 2004/10/13 16:34:00 CDT
64    Tag: blume-20041013-tdp
65    Description:
66    
67    Some rationalization of names:
68    
69       structure BTrace -> structure TDPInstrument
70       etc.
71    
72    This is is preparation of using the original back-trace
73    instrumentation for other purposes.  "TDP" stands for
74    Trace/Debug/Profile.
75    
76    The control flag controlling whether instrumentation is on or off is now
77    registered under a different name, so instead of running sml as
78    
79     sml -Cinstrument.btrace-mode=true
80    
81    one has to say
82    
83     sml -Ctdp.instrument=true
84    
85    ----------------------------------------------------------------------
86    Name: Matthias Blume (blume (at) tti - c (dot) org)
87    Date: 2004/10/11 16:37:00 CDT
88    Tag: blume-20041011-regions
89    Description:
90    
91    Made some minor modifications to elabcore.sml to have source regions
92    be propagated more tightly -- resulting in better (i.e., smaller)
93    regions being reported in error- and debug messages.
94    
95    ----------------------------------------------------------------------
96    Name: Matthias Blume (blume (at) tti - c (dot) org)
97    Date: 2004/10/08 22:50:00 CDT
98    Tag: blume-20041008-cmkw
99    Description:
100    
101    Fixed handling of keywords in .cm files: After seeing "is" the lexer
102    treats subsequent occurrences of "group", "library", "source", "is",
103    "*", and "-" as ordinary identifiers rather than keywords.
104    
105    Most seriously, this fixes a problem with CM's "shell" tool.  The tool
106    is supposed to accept a tool argument called "source", but this did
107    not work because of the clash with the keyword.
108    
109    ----------------------------------------------------------------------
110    Name: Matthias Blume (blume (at) tti - c (dot) org)
111    Date: 2004/10/07 16:00:00 CDT
112    Tag: blume-20041007-cleanup
113    Description:
114    
115    Assorted cleanup work:
116    
117      - got rid of intstrmap in favor of using the library's
118        hash table implementation
119      - threw out most of the pathnames stuff, as it was not used anyway
120      - simplified tokentable implementation
121      - fixed some minor spelling errors
122    
123    ----------------------------------------------------------------------
124    Name: Matthias Blume (blume (at) tti - c (dot) org)
125    Date: 2004/10/06 15:15:15 CDT
126    Tag: blume-20041006-handler
127    Description:
128    
129    Cleaned up the absyn to reflect the invariant that HANDLE always
130    carries a FNexp as part of the type definition.  This eliminates some
131    superfluous sanity checks at runtime down the road.
132    
133    Some minor cleanup of the btrace code.
134    
135    ----------------------------------------------------------------------
136    Name: Matthias Blume (blume (at) tti - c (dot) org)
137    Date: 2004/10/01 10:20:30 CDT
138    Tag: blume-20041001-slave
139    Description:
140    
141    Added hack to make slave mode work in the presence of the version
142    tool.  (Still, since the master does two passes over the code for
143    CMB.make, the release number gets bumped twice when slaves are
144    attached.  I don't know if this is worth fixing...)
145    
146    ----------------------------------------------------------------------
147    Name: Matthias Blume (blume (at) tti - c (dot) org)
148    Date: 2004/09/30 10:55:00 CDT
149    Tag: blume-20040930-version
150    Description:
151    
152    * Moved the "version" magic into its own little library under
153    src/system/smlnj/internal.  This avoids expensive reconstruction
154    of a stable src/compiler/core.cm.
155    
156    * At the same time, structure CompilerVersion is now known as structure
157    SMLNJVersion.
158    
159    * Arranged for the version tool to NOT kick in when rebuilding the system
160      (makeml -rebuild, fixpt).  Otherwise one would never reach a fixpoint.
161      Also, loading the versiontool does not work when rebuilding the system
162      because CM is not properly initialized at that time.
163    
164    ----------------------------------------------------------------------
165    Name: Matthias Blume (blume (at) tti - c (dot) org)
166    Date: 2004/09/29 14:00:00 CDT
167    Tag: blume-20040929-autoversion
168    Description:
169    
170    Implemented some CM magic to have
171        file src/compiler/TopLevel/main/version.sml
172    generated automagically.
173    The version is taken from two files: config/version and config/release.
174    The first is expected to contain a two-part version number such as 110.49.
175    The second should contain a single number, but it may be missing.
176    
177    If the environment variable VERSIONTOOL_BUMP_RELEASE is defined at the
178    time the version tool is loaded (which is the first time you say CMB.make),
179    then the tool will increment the value stored in config/release every
180    time CMB.make is invoked.
181    
182    The binfile format is now insensitive to anything beyond the first
183    two components of a version number, so bumping the release does not render
184    binfiles incompatible.  Auto-bumping can be used to keep track of versions
185    during development without invalidating existing binfiles.
186    
187    In any case, every CMB.make updates the date information in version.sml.
188    (This is the date that is printed in the banner.)
189    
190    ----------------------------------------------------------------------
191    Name: Matthias Blume (blume (at) tti - c (dot) org)
192    Date: 2004/09/28 10:53:00 CDT
193    Tag: blume-20040928-controls
194    Description:
195    
196    Some cleanup of the controls code.
197    
198    ----------------------------------------------------------------------
199    Name: Matthias Blume (blume (at) tti - c (dot) org)
200    Date: 2004/09/27 22:08:00 CDT
201    Tag: blume-20040927-controls
202    Description:
203    
204    Added two pieces of functionality to the Controls interface:
205    
206    1.   val save'restore: 'a control -> unit -> unit
207    
208       grabs the current value of the control in stage 1 and restores it
209       in stage 2.
210    
211    2.   val set' : 'a control * 'a -> unit -> unit
212    
213       stores the given value into the control in stage 2 (i.e., delayed)
214       but does all error checking in stage 1.
215       (This is for string controls that need to do parse their argument --
216        something that might fail.  In some cases, notably in CM, one
217        already knows the intended argument but wants to delay the actual
218        assignment until a time when error recovery would be more difficult.)
219    
220    Changed the handling of controls in tool arguments to classes "sml" and
221    "lazysml":
222       - use Controls.save'restore as a more robust way of restoring the
223         old value (in particular: without having to re-parse the string)
224       - use controls to handle the "overload" keyword in the init group
225         (I believe this change actually fixes a long-standing obscure bug.)
226    
227    ----------------------------------------------------------------------
228    Name: Matthias Blume (blume (at) tti - c (dot) org)
229    Date: 2004/09/27 17:00:00 CDT
230    Tag: blume-20040927-lazysml
231    Description:
232    
233    Added a new tool class called "lazysml" to CM's tool chest.  The only
234    difference to "sml" is that compilation is done with Control.lazysml
235    set to true.  A source of class "lazysml" is automatically recognized
236    by a file name suffix of ".lml".
237    
238    In addition to the above feature, the original class "sml" now also
239    supports a tool argument "lazy" which has the same effect.  As a
240    result, the following three lines are equivalent:
241    
242        foo.sml : lazysml
243        foo.sml : sml (lazy)
244        foo.sml (lazy)
245    
246    The setting goes into effect both during parsing and during
247    compilation.  The original setting is restored right after parsing and
248    after compilation, respectively.
249    
250    In addition to all the above, there is also a general mechanism to set
251    ANY of the "controls" that are available at the command line via
252    "-C..." on a per-sml-file basis.  The same rules that apply for "lazy"
253    apply as well. (In fact, "lazy" is implemented as a special case of
254    the general mechanism.)
255    
256    The .cm file syntax uses a new keyword tool argument called "with".
257    There are several ways of indicating the desired settings:
258    
259        foo.sml (with:parser.quotations=true)
260        foo.sml (with:(name:parser.quotations value:true))
261        foo.sml (with:(name:name1 value:value1 name:name2 value:value2 ...))
262        foo.sml (with:(name1=value1 name2=value2 ...))
263        foo.sml (with:(name1=value1 name:name2 value:value2 name3=value3 ...))
264    
265        etc.
266    
267    Another possible abbreviation is to leave out the =v or value:v part
268    if the name refers to a boolean control (in which case the value is
269    taken to be true).  Thus, one could get lazy sml also by saying:
270    
271        foo.sml (with:parser.lazy-keyword=true)
272        foo.sml (with:parser.lazy-keyword)
273        foo.sml (with:(name:parser.lazy-keyword value:true))
274        foo.sml (with:(name:parser.lazy-keyword))
275    
276    ----------------------------------------------------------------------
277    Name: Matthias Blume (blume (at) tti - c (dot) org)
278    Date: 2004/09/24 16:22:00 CDT
279    Tag: blume-20040924-ppc-long-branch
280    Description:
281    
282    Turned message about "emiting long form of branch" off by default.
283    Added a control flag to turn it back on when desired.
284    
285    ----------------------------------------------------------------------
286    Name: Matthias Blume (blume (at) tti - c (dot) org)
287    Date: 2004/09/24 16:05:00 CDT
288    Tag: blume-20040924-rounding
289    Description:
290    
291    Applied patch for setting rounding modes under Mac OS X.  Thanks to
292    Melissa O'Neill for providing the code!
293    
294    ----------------------------------------------------------------------
295    Name: Matthias Blume (blume (at) tti - c (dot) org)
296    Date: 2004/09/23 17:30:00 CDT
297    Tag: blume-20040923-envvars
298    Description:
299    
300    1. Changed definition of type ControlRegistry.registry_tree to
301       include control_info (i.e., the name of the controlling environment
302       variable).
303    
304    2. Added command-line flags -e and -E to print the names of environment
305       variables that can be used to control internal settings.
306       (This uses the new API mentioned in 1.)
307    
308    ----------------------------------------------------------------------
309    Name: Matthias Blume (blume (at) tti - c (dot) org)
310    Date: 2004/09/13 16:50:00 CDT
311    Tag: Release_110_49
312    Description:
313    
314    New working version (110.49).  NEW BOOTFILES!
315    
316    ----------------------------------------------------------------------
317    Name: Matthias Blume
318    Date: 2004/09/13 16:20:00 CDT
319    Tag: blume-20040913-config-mlrisc
320    Description:
321    
322    Put target "mlrisc" back into the default list.
323    (There is no harm in having it, and some users have expressed their
324    wish to have "mlrisc" included by default.)
325    
326    ----------------------------------------------------------------------
327    Name: John Reppy
328    Date: 2004/09/13
329    Tag: jhr-20040913-signals
330    Description:
331    
332    Fixed the signal masking code to properly nest mask/unmask operations
333    on a per-signal basis.
334    
335    ----------------------------------------------------------------------
336    Name: Matthias Blume (blume (at) tti - c (dot) org)
337    Date: 2004/09/08 13:20:00 CDT
338    Tag: blume-20040908-heap-magic
339    Description:
340    
341    Bumped the heap macig number to 0x09082004 to account for the changed
342    layout of the ML frame under MacOS X.
343    
344    ----------------------------------------------------------------------
345    Name: Allen Leung (leunga (at) reservoir (dot) com)
346    Date: 2004/09/03 11:26:00 EST
347    Tag: leunga-20040903-cygwin-install
348    Description:
349    
350       Added a patch to _arch-n-opsys to enable the Cygwin runtime.
351    The Cygwin runtime is turned on by setting the environment
352    variable SMLNJ_CYGWIN_RUNTIME to 1.
353    
354    ----------------------------------------------------------------------
355    Name: Matthias Blume (blume (at) tti - c (dot) org)
356    Date: 2004/08/31 17:15:00 CDT
357    Tag: blume-20040831-core
358    Description:
359    
360    Added some exports to src/compiler/core.cm upon request by J. Joemann.
361    
362    ----------------------------------------------------------------------
363    Name: Matthias Blume (blume (at) tti - c (dot) org)
364    Date: 2004/08/30 17:55:00 CDT
365    Tag: blume-20040830-installer
366    Description:
367    
368    Upon request by Johannes Joemann:
369    
370    - improved ML code of installer to fall back to coping when renaming
371      fails (i.e., when source and target are on different file systems);
372      the code compiles but has yet to be tested in anger
373    - removed mlrisc from list of default targets (config/targets)
374    
375    ----------------------------------------------------------------------
376    Name: Matthias Blume (blume (at) tti - c (dot) org)
377    Date: 2004/08/27 17:20:00 CDT
378    Tag: blume-20040827-ptreql
379    Description:
380    
381    Added ptreql primop to structure InlineT (upon request from Larry
382    Paulson).
383    
384    ----------------------------------------------------------------------
385    Name: Allen Leung (leunga (at) reservoir (dot) com)
386    Date: 2004/08/15 21:21:00 EST
387    Tag: leunga-110_48-udgraph
388    Description:
389    
390    Another bug fix from Carl Hauser:
391    
392    diff /net/niflab/smlnj48/src/MLRISC/graphs/udgraph.sml udgraph.sml
393    > 48c48
394    > <              | rmv((e as (k,_))::es,L) = rmv(es,if k = i then es else
395    > e::L)
396    > ---
397    > >              | rmv((e as (k,_))::es,L) = rmv(es,if k = i then L else e::L)
398    > Without this, any deletion of an edge in an undirected graph does severe
399    > violence to the graph.
400    
401    ----------------------------------------------------------------------
402    Name: Allen Leung (leunga (at) reservoir (dot) com)
403    Date: 2004/08/10 23:35:00 EST
404    Tag: leunga-110_48-ppc
405    Description:
406    
407       The IBM/MacOS syntax switch on PPC was incorrectly swapped.
408    
409    ----------------------------------------------------------------------
410    Name: Matthias Blume (blume (at) tti - c (dot) org)
411    Date: 2004/08/10 12:00:00 CDT
412    Tag: Release_110_48
413    Description:
414    
415    New working version (110.48).  NEW BOOTFILES!
416    
417    ----------------------------------------------------------------------
418    Name: Allen Leung (leunga (at) reservoir (dot) com)
419    Date: 2004/08/09 12:21:00 EST
420    Tag: leunga-110_47-dijsktra
421    Description:
422    
423    Bug fix from Carl Hauser:
424    
425    single_source_shortest_paths in dijkstra.sml was observed to get wrong
426    answers (by comparing to single_source_shortest_paths in bellman-ford.sml).
427    
428    The problem is that following the expression A.update(dist,s,Num.zero)
429    it is necessary to update the priority queue using Q.decreaseWeight(Q,s).
430    
431    ----------------------------------------------------------------------
432    Name: Matthias Blume (blume (at) tti - c (dot) org)
433    Date: 2004/08/06 18:10:00 CDT
434    Tag: blume-20040806-cmdline
435    Description:
436    
437    Fiddled with handling of command-line options:
438    
439      * sml now quits after processing the command line
440        if -H, -S, -h<n>, or -s<n> appears as the last
441        command-line argument
442      * a new option -q terminates the session when encountered on
443        the command line; subsequent arguments will be ignored
444      * bug fixes: short (erroneous) arguments are no longer ignored
445        completely
446    
447    ----------------------------------------------------------------------
448    Name: Allen Leung (leunga (at) reservoir (dot) com)
449    Date: 2004/08/04 18:17:00 EST
450    Tag: leunga-110_47-ppc-ibm-asm
451    Description:
452    
453      - Added minimal IBM assembly syntax support for PowerPC.
454    
455      - Cygwin: manually changed the file cygwin.def.  Some exported symbols have
456        been altered in the runtime.  We need an automatic way to keep the file
457        in sync.
458    
459    ----------------------------------------------------------------------
460    Name: Matthias Blume (blume (at) tti - c (dot) org)
461    Date: 2004/08/04 14:00:00 CDT
462    Tag: Release_110_47
463    Description:
464    
465    New working version (110.47).  NEW BOOTFILES!
466    
467    ----------------------------------------------------------------------
468    Name: Matthias Blume (blume (at) tti - c (dot) org)
469    Date: 2004/08/03 14:25:00 CDT
470    Tag: blume-20040803-callingconv
471    Description:
472    
473    Added low-level support for choosing C calling conventions by
474    twiddling the type of rawccall.  (See
475    src/compiler/Semant/types/cproto.sml for details.)
476    
477    ----------------------------------------------------------------------
478    Name: Matthias Blume (blume (at) tti - c (dot) org)
479    Date: 2004/08/02 15:55:00 CDT
480    Tag: blume-20040802-backout
481    Description:
482    
483    Backed out of change to win32-filesys.c.  The earlier patch to
484    get_file_time caused CM to produce files with the wrong time stamp.
485    
486    ----------------------------------------------------------------------
487    Name: Matthias Blume (blume (at) tti - c (dot) org)
488    Date: 2004/08/02 14:45:00 CDT
489    Tag: blume-20040802-nlffi-win32
490    Description:
491    
492    Added NLFFI support for Win32, adapted from a patch provided by David
493    Hansel.  This is currently completely untested.  Also, the issue
494    concerning stdcall vs. ccall is still unresolved.
495    
496    ----------------------------------------------------------------------
497    Name: Matthias Blume (blume (at) tti - c (dot) org)
498    Date: 2004/07/30 17:55:00 CDT
499    Tag: blume-20040730-various
500    Description:
501    
502    Gearing up towards 110.47...
503    
504    - various minor bugfixes to ml-nlffigen
505    - a beginning of a manual for nlffi
506    
507    - eliminated 'export name=value' in config/install.sh as this does
508      not work with certain versions of /bin/sh
509      (Thanks to David King at Motorola for catching this.)
510    
511    - several bugfixes provided or suggested by David Hansel at Reactive Systems:
512      - added a test for tm==NULL to gmtime.c and localtime.c
513      - applied patch for incorrect GetFileTime under win32
514      - toSeconds -> toMilliseconds in Win32/win32-process.sml
515    
516    ----------------------------------------------------------------------
517    Name: Matthias Blume (blume (at) tti - c (dot) org)
518    Date: 2004/07/21 18:20:00 CDT
519    Tag: blume-20040721-nlffigen
520    Description:
521    
522    - Fixed minor issue in ml-nlffigen: Now generate structure T_foo
523      for a typedef to an incomplete type, but leave out the "typ" member.
524      (This is just for consistency.)
525    - Started to produce what is supposed to become better (i.e., comprehensive)
526      documentation of what ml-nlffigen does and produces.
527    
528    ----------------------------------------------------------------------
529    Name: Matthias Blume (blume (at) tti - c (dot) org)
530    Date: 2004/07/14 16:25:00 CDT
531    Tag: blume-20040714-union
532    Description:
533    
534    Added C_UNION to c-calls/c-types.sml and updated the machinery
535    (ml-nlffigen, cproto.sml) that conveys C function interface
536    information to the code generator.
537    
538    However, the actual architecture-specific implementation of function
539    arguments and results that are C unions is still not implemented.
540    
541    ----------------------------------------------------------------------
542    Name: Allen Leung (leunga (at) reservoir (dot) com)
543    Date: 2004/07/14 14:38:00 EST
544    Tag: leunga-110_46_1-ppc-lwzu
545    Description:
546    
547        Added these instructions to the PowerPC architecture:
548        LBZU(X), LHZU(X), LWZU(X),
549        STWU(X), STFDU, STFSU
550    
551        etc...
552    
553        Note: I haven't added their instruction encoding into the description.
554    
555    ----------------------------------------------------------------------
556    Name: Allen Leung (leunga (at) reservoir (dot) com)
557    Date: 2004/07/13 15:04:00 EST
558    Tag: leunga-110_46_1-ppc-lwarx
559    Description:
560    
561       Added the two instructions LWARX and STWCX to the PowerPC
562    instruction set.
563    
564       A (untested) rewrite of loop-structure.sml.  The old version
565    is completely broken.
566    
567    ----------------------------------------------------------------------
568    Name: Matthias Blume (blume (at) tti - c (dot) org)
569    Date: 2004/07/13 13:50:00 CDT
570    Tag: blume-20040713-nlffi
571    Description:
572    
573    - use paramAlloc to report c-calls with too many arguments
574      (for PPC version where parameter area is pre-allocated)
575    - added ccall_maxargspace to machspec (to implement the above)
576    - made "make" commend in CM's "make" tool configurable
577    - added option (default: on) for passing the name of the SML/NJ's "bin"
578      directory to "make"; the call looks like this:
579    
580         make <options> SMLNJ_BINDIR=<dir> <target>
581    
582      This can be used by the Makefile to, e.g., pick the "right" version
583      of ml-nlffigen.
584    - minor code tweaks
585    
586    ----------------------------------------------------------------------
587    Name: Matthias Blume (blume (at) tti - c (dot) org)
588    Date: 2004/07/12 22:50:00 CDT
589    Tag: blume-110_46_1-macosx-nlffi
590    Description:
591    
592    NLFFI under Mac OS X now working (sort of).  This is largely untested,
593    though.
594    
595    Note:  1. You have to make a new, clean build of the runtime system.
596           2. There are new BOOTFILES, you have to use them!
597              (Doing the bootstrap process yourself would be *very* painful!
598               If you absolutely have to do it, build the system under
599               a different architecture and then cross-compile.)
600    
601    Version bumped to 110.46.1 to account for runtime data format changes.
602    
603    ----------------------------------------------------------------------
604    Name: Matthias Blume (blume (at) tti - c (dot) org)
605    Date: 2004/06/18 14:30:00 CDT
606    Tag: blume-20040618-unix
607    Description:
608    
609    Changed the implementation of structure Unix so that the same stream
610    is returned every time one of the {text,bin}{In,Out}streamOf functions
611    is invoked on the same proc.  This is not what the spec currently
612    says -- although IMO it arguably should.  (See discussion below.)
613    
614    ----------------------------------------------------------------------
615    Name: Matthias Blume (blume (at) tti - c (dot) org)
616    Date: 2004/06/17 18:15:00 CDT
617    Tag: Release_110_46
618    Description:
619    
620    New working version (110.46).  NEW BOOTFILES!
621    
622    ----------------------------------------------------------------------
623    Name: Matthias Blume (blume (at) tti - c (dot) org)
624    Date: 2004/06/17 17:20:00 CDT
625    Tag: blume-20040617-timer-unix
626    Description:
627    
628    Changed the interface of structures Timer and Unix to match the most
629    recent Basis spec.
630    
631    In the case of Unix there still seems to be an open/weird issue:
632    
633       The {text,bin}{In,Out}streamOf functions are supposed to create
634       fresh streams whenever they are called -- as opposed to have them
635       return the same stream every time.  This design is supposed to
636       prevent space leaks caused by proc values hanging on to streams.
637    
638       The reap function, on the other hand, is supposed to close the
639       streams.  This cannot be done without having a handle on the
640       stream in proc after all...
641    
642    I took the liberty to implement the following stopgap solution:
643    
644      The proc value hangs on to the most recently created stream(s).
645      Reap closes those.  If either or both of the two streams hadn't
646      been created at all yet, then reap will close the corresponding
647      file descriptors directly.
648    
649    PS: I don't understand the original space leak argument anymore.  If
650    a proc hangs on to the imperative stream, then I/O operations on those
651    will advance the state of the cached stream and avoid the space leak.
652    
653    ----------------------------------------------------------------------
654    Name: Matthias Blume (blume (at) tti - c (dot) org)
655    Date: 2004/05/28 16:45:00 CDT
656    Tag: blume-20040528-basis
657    Description:
658    
659    Added signature PACK_REAL and exported functor PrimIO.
660    
661    ----------------------------------------------------------------------
662    Name: Matthias Blume (blume (at) tti - c (dot) org)
663    Date: 2004/05/25 16:00:00 CDT
664    Tag: blume-20040525-group-owner
665    Description:
666    
667    CM now ignores (but still accepts) the "owner" information in group
668    descriptions.  The owner of a group is its next enclosing
669    library. Each group must have a unique owner.  (There is a virtual
670    "toplevel" library that owns groups which are not nested within a real
671    library.)  Previously, each group had to explicitly declare its owner,
672    and CM would check that such a declaration is correct.  The new scheme
673    is to have CM check that for each group there is precisely one owning
674    library.
675    
676    The advantage of the new scheme is that the programmer no longer needs
677    to maintain the somewhat annoying owner information.  The downside is
678    that CM cannot enforce the ownership rule across multiple runs of
679    CM.make.  Fortunately, enclosing the same group in two different
680    libraries A and B which are not part of the same program does not
681    cause real problems.
682    
683    ----------------------------------------------------------------------
684    Name: Matthias Blume (blume (at) tti - c (dot) org)
685    Date: 2004/05/20 16:00:00 CDT
686    Tag: blume-20040520-win32
687    Description:
688    
689    Made the win32 version work again.  (Strangely, a misplaced comma had
690    slipped into win32-process.c which prevented the runtime from being
691    compiled correctly.)
692    
693    Also, included a minor addition to ml-build.bat analogous to what was
694    done in blume-20040519-ml-build.
695    
696    ----------------------------------------------------------------------
697    Name: Matthias Blume (blume (at) tti - c (dot) org)
698    Date: 2004/05/19 22:10:00 CDT
699    Tag: blume-20040519-ml-build
700    Description:
701    
702    Arranged for ml-build to clean up after itself a little bit better.
703    The script generates a temporary SML source file and compiles it using
704    CM, so CM generates metadata (GUID, SKEL, objectfile) for it.  It now
705    gets rid of those at the end, so they don't accumulate under .cm.
706    
707    This required a minor change to install.sh because the name of the
708    metadata directory (default: .cm) is actually configurable at
709    installation time.
710    
711    ----------------------------------------------------------------------
712    Name: Matthias Blume (blume (at) tti - c (dot) org)
713    Date: 2004/05/18 15:50:00 CDT
714    Tag: blume-20040518-mkreader
715    Description:
716    
717    Added Posix.IO.mk{Bin,Text}{Reader,Writer} by lifting their respective
718    implementations from internal modules PosixBinPrimIO and PosixTextPrimIO.
719    
720    ----------------------------------------------------------------------
721    Name: Matthias Blume (blume (at) tti - c (dot) org)
722    Date: 2004/05/11 14:35:00 CDT
723    Tag: blume-20040511-win32sock
724    Description:
725    
726    Added previously missing support for many socket-related functions
727    under win32.  Thanks to David Hansel <hansel@reactive-systems.com>
728    for the voluminous patch!
729    
730    (I have not tested this patch under win32 yet.)
731    
732    Here is David's e-mail:
733    
734    Hi,
735    
736    Attached to this email you find a diff against sml/nj 110.45
737    that will enable socket support under Windows.
738    
739    To apply the patch (using unix or cygwin)
740    1) gunzip runtime.diff.gz
741    2) "cd" into "src/runtime" in the source tree of a fresh
742       110.45 installation.
743    3) patch -p 1 < [your/path/to]runtime.diff
744    
745    The code compiles fine but has NOT yet been extensively tested.
746    I only ran a few tests for basic socket client functionality
747    (which worked fine).  Especially the functions that use ioctl
748    are not tested at all and might not work (see below).
749    
750    I implemented this since we want to move to a newer version of sml/nj
751    but need socket support in order to use it.  This is the first time I
752    even had a look at the sml/nj source,  so please review my changes
753    before making this part of the distribution!  Here are a few issues
754    that I think might be better for someone to solve who is more
755    familiar with the sml/nj source (and socket programming):
756    
757    - getnetbyaddr.c and getnetbyname.c will raise a "not implemented"
758      exception since I could not figure out what the windows equivalent
759      of these functions is
760    
761    - In sockets-osdep.h there are a some #include statements that are
762      only used in a few files that include sockets-osdep.h
763    
764    - In smlnj-sock-lib.c, function init_fn() calls WSAStartup() but
765      does not process its return value since I don't know how to
766      report an error upwards.
767    
768    - It would probably be good to have a call to WSACleanup() when
769      the library is unloaded (if there is such a possibility).
770      Otherwise I think Windows will take care of this automatically
771      when the process finishes.
772    
773    - I used ioctlsocket() as a replacement for ioctl() but I have
774      no idea if that is actually the proper replacement on Windows.
775    
776    - All these issues are marked in the code by "FIXME" comments.
777    
778    We use sml/nj extensively in our products and are quite happy
779    with it.  I hope this contribution will help you.
780    
781    Keep up the good work!
782    
783    David
784    
785    ----------------------------------------------------------------------
786    Name: Matthias Blume (blume (at) tti - c (dot) org)
787    Date: 2004/05/11 14:20:00 CDT
788    Tag: blume-20040511-installml
789    Description:
790    
791    Fixed two bugs in installml script.  (Thanks to Vesa A. Norrman for
792    the patch.)
793    
794    ----------------------------------------------------------------------
795    Name: Matthias Blume (blume (at) tti - c (dot) org)
796    Date: 2004/05/11 14:05:00 CDT
797    Tag: blume-20040511-nlffi-netbsd
798    Description:
799    
800    Added support for nlffi under netbsd.  (Thanks to Vesa A. Norrman for
801    the patch.)
802    
803    ----------------------------------------------------------------------
804    Name: Matthias Blume (blume (at) tti - c (dot) org)
805    Date: 2004/05/11 12:05:00 CDT
806    Tag: blume-20040511-exports
807    Description:
808    
809    As per request by Adam Chlipala <adam@hcoop.net>, extended various
810    export lists in compiler-related .cm-files.
811    
812    ----------------------------------------------------------------------
813    Name: Matthias Blume (blume (at) tti - c (dot) org)
814    Date: 2004/05/11 11:35:00 CDT
815    Tag: blume-20040511-allsource
816    Description:
817    
818    The installer now honors the "src-smlnj" target again, although its meaning
819    has changed from "all sources required for the compiler" to "all sources
820    the installer knows about".  In other words, if you enable "src-smlnj"
821    in the "targets" file, then the installer will pull in sources for
822    everything.  (Notice that this refers to source code only.  Compiled
823    code is still only installed for modules that were requested explicitly
824    or which are required for other modules that were requested explicitly.)
825    
826    ----------------------------------------------------------------------
827    Name: Matthias Blume (blume (at) tti - c (dot) org)
828    Date: 2004/04/23 17:40:00 CDT
829    Tag: blume-20040423-ieee-scan
830    Description:
831    
832    Fixed IEEEReal.scan (and .fromString) so that if there is an overflow
833    in the exponent calculation we get INF or ZERO (depending on the mantissa
834    and the sign of the exponent).
835    
836    ----------------------------------------------------------------------
837    Name: Matthias Blume (blume (at) tti - c (dot) org)
838    Date: 2004/04/23 10:40:00 CDT
839    Tag: blume-20040423-ml-build
840    Description:
841    
842    The ml-build script now terminates with a non-0 status when something
843    goes wrong.
844    
845    ----------------------------------------------------------------------
846    Name: Matthias Blume (blume (at) tti - c (dot) org)
847    Date: 2004/04/22 16:35:00 CDT
848    Tag: blume-20040422-Option
849    Description:
850    
851    Made exception Option to be the same as exception Option.Option
852    (as it should be).
853    
854    ----------------------------------------------------------------------
855    Name: Allen Leung (leunga (at) reservoir (dot) com)
856    Date: 2004/03/19 14:40:00 EST
857    Tag: leunga-20040319-cygwin-nlffi
858    Description:
859    
860    Fixed the runtime so that ml-nlffi-lib runs on the cygwin version
861    of SML/NJ.  The problem is that
862    
863       lib = dlopen(NULL, ...)
864       f   = dlsym(lib, "malloc");
865    
866    does not work on Windows unless we explicitly export symbols
867    such as 'malloc' during linking.  We fixed this by explicitly
868    exporting the required symbols with the magic gcc incantation:
869    
870        -Wl,--export-all cygwin.def
871    
872    where cygwin.def is a file containing all the symbols that
873    we wish to export.
874    
875    I suspect this is a Windows problem and we'll have to
876    do the same (somehow with windows compilers) when
877    we build the native win32 version with the system
878    calls LoadLibrary/GetProcAddress.
879    
880    ----------------------------------------------------------------------
881    Name: Matthias Blume (blume (at) tti - c (dot) org)
882    Date: 2004/03/04 16:35:00 CST
883    Tag: blume-20040304-intinf-fmt
884    Description:
885    
886    Fixed problem with IntInf.fmt (sign would show up on the right instead
887    of on the left for BIN, OCT, and HEX).
888    
889    ----------------------------------------------------------------------
890    Name: Matthias Blume (blume (at) tti - c (dot) org)
891    Date: 2004/03/04 11:25:00 CST
892    Tag: blume-20040304-symlinks
893    Description:
894    
895    Fixed problem with installer script (unix only) where bin/ml-yacc and
896    friends pointed (via symlinks) to absolute locations instead of just
897    .run-sml.  This was reported by Vesa A Norrman.
898    
899    ----------------------------------------------------------------------
900    Name: Matthias Blume (blume (at) tti - c (dot) org)
901    Date: 2004/02/13 14:50:00 CST
902    Tag: Release_110_45
903    Description:
904    
905    New working version (110.45). New bootfiles.
906    
907    ----------------------------------------------------------------------
908    Name: Matthias Blume (blume (at) tti - c (dot) org)
909    Date: 2004/01/26 15:15:15 CST
910    Tag: blume-20040126-toplevel
911    Description:
912    
913    Improved handling of exceptions at the interactive toplevel.
914    
915    ----------------------------------------------------------------------
916    Name: Matthias Blume (blume (at) tti - c (dot) org)
917    Date: 2004/01/26 11:25:00
918    Tag: blume-20040126-app
919    Description:
920    
921    Type of top-level "app" corrected.
922    Added code for setting vp_limitPtrMask to Win32-specific runtime.
923    
924    ----------------------------------------------------------------------
925    Name: Matthias Blume (blume (at) tti - c (dot) org)
926    Date: 2003/11/18 17:10 CST
927    Tag: blume-20031118-basis-fiddle
928    Description:
929    
930    - changed Timer interface to what might become the spec
931    - POSIX_FLAGS -> BIT_FLAGS according to spec
932    - some other minor discrepancies wrt. spec eliminated
933    
934    ----------------------------------------------------------------------
935    Name: Matthias Blume (blume (at) tti - c (dot) org)
936    Date: 2003/11/06 12:00:00 CST
937    Tag: Release_110_44
938    Description:
939    
940    New working version (110.44). New bootfiles.
941    
942    ----------------------------------------------------------------------
943    Name: Matthias Blume (blume (at) tti - c (dot) org)
944    Date: 2003/11/04 11:50:00 CST
945    Tag: blume-20031104-move-libraries
946    Description:
947    
948    Eliminated the "dont_move_libraries" directive in config/targets.
949    (The mechanism was broken and could not be fixed easily.  Moreover,
950    there does not seem to be any reason not to move all libraries into
951    lib during installation.  I originally implemented this directive as a
952    backward-compatibility feature when I first introduced the new CM.
953    Now that things have been stable for a long time and going back to the
954    old CM is not an option, there is no reason to keep it around.)
955    
956    ----------------------------------------------------------------------
957    Name: Matthias Blume (blume (at) tti - c (dot) org)
958    Date: 2003/11/03 16:00:00 CST
959    Tag: blume-20031103-installdir
960    Description:
961    
962    Made installer honor INSTALLDIR variable again.  (Thanks to Chris
963    Richards for pointing out the problem and providing the solution.)
964    
965    ----------------------------------------------------------------------
966    Name: Matthias Blume (blume (at) tti - c (dot) org)
967    Date: 2003/10/01 17:05:00 CDT
968    Tag: blume-20031001-lal-mlrisc
969    Description:
970    
971    MLRISC bug fix from Lal.
972    
973    ----------------------------------------------------------------------
974    Name: Matthias Blume (blume (at) tti - c (dot) org)
975    Date: 2003/09/30 16:10:00 CDT
976    Tag: blume-20030930-primio-bat
977    Description:
978    
979    1. Added openVector, nullRd, and nullWr to PRIM_IO.
980    2. Improved .bat files (for Win32 port) to make things work under Win95.
981       (thanks to Aaron S. Hawley for this one)
982    
983    ----------------------------------------------------------------------
984    Name: Matthias Blume (blume (at) tti - c (dot) org)
985    Date: 2003/09/26 16:05:00 CDT
986    Tag: blume-20030926-wrappriv
987    Description:
988    
989    Added missing wrapper for privilege "primitive" in $smlnj/viscomp/core.cm.
990    
991    ----------------------------------------------------------------------
992    Name: Matthias Blume (blume (at) tti - c (dot) org)
993    Date: 2003/09/26 15:00:00 CDT
994    Tag: blume-20030926-110_43_3
995    Description:
996    
997    - additional cleanup
998    - version number bump, NEW BOOTFILES
999    
1000    ----------------------------------------------------------------------
1001    Name: Matthias Blume (blume (at) tti - c (dot) org)
1002    Date: 2003/09/26 12:00:00 CDT
1003    Tag: blume-20030926-ppautoload
1004    Description:
1005    
1006    I modified the read-eval-print loop so that the autoloader gets
1007    invoked whenever the prettyprinter tries to look up a symbol that
1008    is not currently defined in the toplevel environment but which
1009    appears in CM's autoload registry.  As a result, we see far fewer of
1010    those ?.Foo.Bar.xxx names in the prettyprinter's output.
1011    
1012    In addition to this I tried to clean up some pieces of the Basis
1013    implementation (e.g., Socket, Word8Array) in order to prevent other
1014    instances of these ?.Foo.Bar.xxx names from being printed.
1015    
1016    The mechanism that picks names for types still needs some work, though.
1017    (Right now it seems that if there is a type A.t which is defined to
1018    be B.u, but B is unavailable at toplevel, then A.t gets printed as
1019    "?.B.u" although the perhaps more sensible solution would be to use
1020    "A.t" in this case.  In other words, the prettyprinter should follow
1021    a chain of DEFtycs not farther than there are corresponding toplevel
1022    names in the current environment.)
1023    
1024    ----------------------------------------------------------------------
1025    Name: Matthias Blume (blume (at) tti - c (dot) org)
1026    Date: 2003/09/24 16:31:00 CDT
1027    Tag: blume-20030924-installer
1028    Description:
1029    
1030    Another installer tweak:  All the ML code for the installer is now
1031    compiled during CMB.make and put into a little library called
1032    $smlnj/installer.cm.  The installation then simply invokes
1033    
1034       sml -m $smlnj/installer.cm
1035    
1036    and everything happens automagically.
1037    
1038    Win32:  ML code senses value of environment variable SMLNJ_HOME.
1039    Unix:   ML code senses values of environment variables ROOT, CONFIGDIR,
1040            and BINDIR.
1041    
1042    The new scheme guarantees that the ML code responsible for the installation
1043    is in sync with the APIs of the main system.  Also, the installer is
1044    somewhat faster because the installer script is precompiled.
1045    
1046    ----------------------------------------------------------------------
1047    Name: Matthias Blume (blume (at) tti - c (dot) org)
1048    Date: 2003/09/24 15:35:00 CDT
1049    Tag: blume-20030924-synsock
1050    Description:
1051    
1052    Added a signature SYNCHRONOUS_SOCKET to basis.cm.  This is like SOCKET
1053    but excludes all non-blocking operations.  Defined SOCKET (in Basis)
1054    and CML_SOCKET in terms of SYNCHRONOUS_SOCKET.  Removed superfluous
1055    implementations of non-blocking operations from CML's Socket
1056    structure.
1057    
1058    ----------------------------------------------------------------------
1059    Name: Matthias Blume (blume (at) tti - c (dot) org)
1060    Date: 2003/09/24 15:10:05 CDT
1061    Tag: blume-20030924-sockets
1062    Description:
1063    
1064    1. Fixed SOCKET API and implementation to match Basis spec.
1065       This required changing the internal representation of sockets to one
1066       that remembers (for each socket file descriptor) whether it is currently
1067       blocking or non-blocking.  This state is maintained lazily (i.e., a system
1068       call is made only if the state actually needs to change).
1069    
1070    2. OS-specific details of sockets were moved into separate files, thus
1071       making it possible to unify the bulk of the socket implementations
1072       between Unix and Win32.
1073    
1074    3. CML's socket API changed accordingly.
1075       (Note that we need to remove non-blocking functions from this API
1076       since they are redundant in the case of CML!)
1077    
1078    4. CML's socket implementation now makes use of non-blocking functions
1079       provided by Basis, thus removing all OS-dependent code from this part
1080       of CML.
1081    
1082    5. Changed Real64.precision from 52 to 53.  Minor cleanup in Real64 code.
1083    
1084    ----------------------------------------------------------------------
1085    Name: Matthias Blume (blume (at) tti - c (dot) org)
1086    Date: 2003/09/22 12:10:00 CDT
1087    Tag: blume-20030922-110_43_2
1088    Description:
1089    
1090    Made a new interim version and bootfiles for developer's bootstrapping
1091    convenience.
1092    
1093       110.43.2 -- NEW BOOTFILES
1094    
1095    ----------------------------------------------------------------------
1096    Name: Matthias Blume (blume (at) tti - c (dot) org)
1097    Date: 2003/09/19 15:55:00 CDT
1098    Tag: blume-20030919-cmdir
1099    Description:
1100    
1101    1. new-install.sh -> install.sh
1102    2. changed default CM "metadata" directory name to ".cm" (instead of "CM")
1103    3. tweaked installer so that another name instead of .cm can be chosen
1104       at install time (by setting the CM_DIR_ARC environment variable
1105       during installation); once installation is complete, the name is
1106       fixed
1107    
1108    ----------------------------------------------------------------------
1109    Name: Matthias Blume (blume (at) tti - c (dot) org)
1110    Date: 2003/09/18 16:00:00 CDT
1111    Tag: blume-20030918-110_43_1
1112    Description:
1113    
1114    Made a new interim version and bootfiles for developer's bootstrapping
1115    convenience.
1116    
1117       110.43.1 -- NEW BOOTFILES
1118    
1119    ----------------------------------------------------------------------
1120    Name: Matthias Blume (blume (at) tti - c (dot) org)
1121    Date: 2003/09/18 15:20:00 CDT
1122    Tag: blume-20030918-misc
1123    Description:
1124    
1125    1. Exported fractionsPerSecond etc. from TimeImp (but not from Time as
1126       this seems to be controversial at the moment) and used those in
1127       Posix.ProcEnv.times.
1128    
1129    2. Added Time.{from,to}Nanoseconds to Time.
1130    
1131    3. Improved Real.{from,to}LargeInt by avoiding needless calculations.
1132       For example, fromLargeInt never needs to look at more than 3 "big
1133       digits" to get its 53 bits of precision.
1134    
1135    ----------------------------------------------------------------------
1136    Name: Matthias Blume (blume (at) tti - c (dot) org)
1137    Date: 2003/09/17 16:30:00 CDT
1138    Tag: blume-20030917-real32-slices
1139    Description:
1140    
1141    Added an entry to the primitive environment
1142    (compiler/Semant/statenv/prim.sml) for int32->real64 conversion and
1143    added code to compiler/CodeGen/main/mlriscGen.sml to implement it.
1144    
1145    Removed some of the "magic" constants in real64.sml and replaced them
1146    with code that generates these values from their corresponding
1147    integer counterparts.
1148    
1149    Made all(?) the slice-related changes to the Basis and made everything
1150    compile again...
1151    
1152    ----------------------------------------------------------------------
1153    Name: Matthias Blume (blume (at) tti - c (dot) org)
1154    Date: 2003/09/15 17:45:00 CDT
1155    Tag: blume-20030915-rbase
1156    Description:
1157    
1158    Fixed bug in Real.fromLargeInt.
1159    
1160    ----------------------------------------------------------------------
1161    Name: Matthias Blume (blume (at) tti - c (dot) org)
1162    Date: 2003/09/13 18:11:00 CDT
1163    Tag: blume-20030913-libinstall
1164    Description:
1165    
1166    Minor bugfix in config/libinstall (set anchor with path to
1167    standalone tool after installing it, otherwise libraries that
1168    need ml-lex or ml-yacc won't compile the first time the installer
1169    runs).
1170    
1171    ----------------------------------------------------------------------
1172    Name: Matthias Blume (blume (at) tti - c (dot) org)
1173    Date: 2003/09/12 11:45:00 CDT
1174    Tag: blume-20030912-various
1175    Description:
1176    
1177     - fixed bug in Real.toLargeInt
1178     - fixed bug in Posix.ProcEnv.times
1179     - changed inputLine functions to return an option
1180     - minor installer improvements / bugfixes
1181     - changed default @SMLalloc parameter for x86/celeron to 64k
1182    
1183    ----------------------------------------------------------------------
1184    Name: Matthias Blume (blume (at) tti - c (dot) org)
1185    Date: 2003/09/09 22:00:00 CDT
1186    Tag: Release_110_43
1187    Description:
1188    
1189        New working release 110.43.  New bootfiles.
1190    
1191    ----------------------------------------------------------------------
1192    Name: Matthias Blume (blume (at) tti - c (dot) org)
1193    Date: 2003/09/09 19:20:00 CDT
1194    Tag: blume-20030909-installer
1195    Description:
1196    
1197    Rewrote large parts of config/install.sh in SML (config/libinstall.sml).
1198    Modified config/install.bat to take advantage of it.  Also modified
1199    config/install.sh (and called it config/new-install.sh) to take advantage
1200    of it on Unix systems.  (The SML code is (supposed to be) platform-
1201    independent.)
1202    
1203    The installer can now install everything under Win32
1204    as well as under *nix as long as it compiles.
1205    
1206    Other changes:
1207    
1208      - made CML compile again under Win32
1209      - made eXene compile under Win32 (by providing a fake structure UnixSock
1210        and by using OS.Process.getEnv instead of Posix.ProcEnv.getenv)
1211      - fixed a bug in nowhere: it assumed that type OS.Process.status is the
1212        same as type int;  under Win32 it isn't
1213      - fixed some slice-related problems in the win32-specific parts of CML
1214      - added a functor argument "sameVol" to os-path-fn.sml in the Basis
1215        (under Win32, the volume name is case-insensitive, and the
1216        OS.Path code compares volume names for equality)
1217    
1218    ----------------------------------------------------------------------
1219    Name: Matthias Blume (blume (at) tti - c (dot) org)
1220    Date: 2003/09/08 11:55:00 CDT
1221    Tag: blume-20030908-fullpath
1222    Description:
1223    
1224    Made Win32 version of OS.FileSys.fullPath return current directory
1225    when given an empty string.  This is what the spec says, and incidentally,
1226    CM depends on it.  (CM otherwise goes into an infinite loop in certain
1227    cases when presented with the name of a non-existing .cm file.)
1228    
1229    ----------------------------------------------------------------------
1230    Name: Matthias Blume (blume (at) tti - c (dot) org)
1231    Date: 2003/09/04 16:30:00 CDT
1232    Tag: blume-20030905-slices-etc
1233    Description:
1234    
1235      1. Changed interface to vectors and arrays in Basis to match
1236         (draft) Basis spec.
1237      2. Added signatures and implementations of slices according to
1238         Basis spec.
1239      3. Edited source code throughout the system to make it compile again
1240         under 1. and 2.  (In some cases code had to be added to have it
1241         match the new signatures.)
1242      4. MLRISC should be backward-compatible: the copies of the originals
1243         of files that needed to change under 3. were retained, the .cm files
1244         check the compiler version number and use old versions when
1245         appropriate.
1246      5. Changed type of OS.FileSys.readDir and Posix.FileSys.readdir to
1247         dirstream -> string option (in accordance with Basis spec).
1248      6. When generating code that counts lines, ml-lex used function
1249         CharVector.foldli, taking advantage of its old interface.
1250         This has been replaced with the corresponding code from
1251         CharVectorSlice.  (html-lex must be re-lexed!)
1252      7. BitArray in smlnj-lib/Util has been extended/modified to match the
1253         new MONO_ARRAY signature.  (Do we need BitArraySlice?)
1254      8. Removed temporary additions (fromInternal, toInternal) from the
1255         (now obsolete) IntInf in smlnj-lib/Util.
1256      9. Cleaned up structure Byte.
1257     10. Added localOffset, scan, and fromString to Date (according to spec).
1258         Cleaned/corrected implementation of Date.
1259         (Still need to check for correctness; implement better canonicalizeDate.)
1260     11. Added "scan" to signature IEEE_REAL.
1261     12. Some improvements to IntInf [in particular: efficiency-hack for
1262         mod and rem when second operand is 2 (for parity checks).]
1263     13. Changed representation of type Time.time, using a single IntInf.int
1264         value counting microseconds.  This considerably simplified the
1265         implementation of structure Time.  We now support negative time
1266         values; scan and fromString handle signs.
1267     14. Functor PrimIO now takes two additional arguments (VectorSlice and
1268         ArraySlice).
1269    
1270    ----------------------------------------------------------------------
1271    Name: Matthias Blume (blume (at) tti - c (dot) org)
1272    Date: 2003/08/28 17:00:00 CDT
1273    Tag: blume-20030828-intinf
1274    Description:
1275    
1276    This is a major update which comes with a version number bump
1277    (110.42.99 -- yes, we are really close to 110.43 :-), NEW BOOTFILES,
1278    and an implementation of IntInf in the Basis.
1279    
1280    There are a fairly large number of related changes and updates throughout
1281    the system:
1282    
1283    Basis:
1284            - Implemented IntInf.
1285            - Made LargeInt a projection of IntInf (by filtering through INTEGER).
1286            - Added some missing Real64 operations, most notably Real.toLargeInt.
1287            - Added FixedInt as a synonym for Int32.
1288    
1289    compiler:
1290            * Added support for a built-in intinf type.
1291              - literals
1292              - pattern matching
1293              - conversion shortcuts (Int32.fromLarge o Int.toLarge etc.)
1294              - overloading on literals and operations
1295    
1296            This required adding a primitive type intinf, some additional
1297            primops, and implementations for several non-trivial intinf
1298            operations in Core.  (The intinf type is completely abstract
1299            to the compiler; all operations get delegated back to the Core.)
1300    
1301            * Intinf equality is handled by polyequal.  However, the compiler
1302            does not print its usual warning in this case (since polyequal
1303            is the right thing to do there).
1304    
1305            * Improved the organization of structure InlineT.
1306    
1307            * A word about conversion primops:
1308            If conversions involving intinf do not cancel out during
1309            CPS contract, then the compiler must insert calls to Core functions.
1310            Since all core access must be resolved already during the FLINT
1311            translate phase, it would be too late a the time of CPS contract
1312            to add new Core calls.  For this reason, conversion primops
1313            for intinf carry two arguments:  1. the numeric argument that
1314            they are supposed to convert, and 2. the Core function that
1315            can help with this conversion if necessary.  If CPS contract
1316            eliminates a primop, then the associated Core function becomes
1317            dead and goes away.  Intinf conversion primops that do not get
1318            eliminated by CPS contract get rewritten into calls of their
1319            core functions by a separate, new phase.
1320    
1321    interactive system:
1322            - Control.Print.intinfDepth controls max length of intinf constants
1323            being printed. (Analogous to Control.Print.stringDepth.)
1324            - Cleanup in printutil and pputil: got rid of unused stuff and
1325            duplicates; replaced some of the code with code that makes better
1326            use of library functionality.
1327    
1328    CM:
1329            Bugfix: parse-errors in init group (system/smlnj/init/init.cmi)
1330            are no longer silent.
1331    
1332    CKIT:
1333            Fixed mismatched uses of Int32 and LargeInt.  I always decided
1334            in favor of LargeInt -- which is now the same as IntInf.
1335            CKIT-knowledgable people should check whether this is what's
1336            intended and otherwise change things back to using Int32 or
1337            FixedInt.
1338    
1339    Throughout the code:
1340            Started using IntInf.int literals and built-in operations
1341            (e.g., comparison with 0) where this seems appropriate.
1342    
1343    
1344    ----------------------------------------------------------------------
1345    Name: Dave MacQueen (dbm@cs.uchicago.edu)
1346    Date: 2003/08/13 11:36:00 CDT
1347    Tag: dbm-20030813-mcz-merge1
1348    Description:
1349    
1350    Merging changes from the mcz-branch development branch into trunk.
1351    These changes involve replacement of the emulated old prettyprinter
1352    interface with direct use of the SML/NJ Lib PP library, and fixing
1353    of a couple of bugs (895, 1186) relating to error messages.  A new
1354    prettyprinter for ast datatypes (Elaborator/print/ppast.{sig,sml})
1355    has been added.
1356    
1357    ----------------------------------------------------------------------
1358    Name: Matthias Blume (blume (at) tti - c (dot) org)
1359    Date: 2003/08/11 15:45:00 CDT
1360    Tag: blume-20030811-windows
1361    Description:
1362    
1363    Version number bumped to 110.42.9.  NEW BOOTFILES!!!
1364    
1365        http://smlnj.cs.uchicago.edu/dist/working/110.42.9/
1366    
1367    This patch restores SML/NJ's ability to run under win32.  There are a
1368    number of changes, including fixes for several bugs that had gone
1369    unnoticed until now:
1370    
1371      - uname "CYGWIN_NT*" is recognized as win32 (This is relevant only when
1372        trying to run the win32 version from within cygwin.)
1373    
1374      - There are a number of simple .bat scripts that substitute for their
1375        corresponding Unix shell-scripts.  (See below.)
1376    
1377      - The internals of ml-build have been modified slightly.  The main
1378        difference is that instead of calling ".link-sml" (or link-sml.bat)
1379        using OS.Process.system, the ML process delegates this task back
1380        to the script.  Otherwise problems arise in mixed environments such
1381        as Cygwin where scripts look and work like Unix scripts, but
1382        where OS.Process.system cannot run them.
1383    
1384      - In CM, the srcpath pickler used native pathname syntax -- which
1385        is incorrect in the case of cross-compilation.  The new pickle format
1386        is independent of platform-specific naming conventions.
1387    
1388      - Path configuration files (such as lib/pathconfig) can now choose
1389        between native and standard syntax.  Placing a line of the form
1390    
1391            standard!
1392    
1393        into the file causes all subsequent paths to be interpreted using
1394        CM standard pathname syntax (= Unix conventions); a line
1395    
1396            native!
1397    
1398        switches back to native style.  This was needed so that
1399        path config files can be written portably, see src/system/pathconfig.
1400    
1401      - Runtime system:
1402    
1403          - win32-filesys.c:  get_file_time and set_file_time now
1404            access modification time, not creation time.
1405    
1406          - I/O code made aware of new array representation.
1407    
1408          - Bug fixes in X86.prim.masm.
1409    
1410       - src/system/makeml made aware of win32. (For use under cygwin
1411            and other Unix-environments for windows.)
1412    
1413       - In Basis, fixed off-by-one error in win32-io.sml (function vecF)
1414         which caused BinIO.inputAll to fail consistently.
1415    
1416    .bat scripts:
1417    
1418       Windows .bat scripts assume that SMLNJ_HOME is defined.
1419    
1420       - sml.bat, ml-yacc.bat, ml-lex.bat:  Driver scripts for standalone
1421         applications (sml, ml-yacc, ml-lex).
1422       - ml-build.bat: analogous to ml-build.
1423       - config\install.bat:  Analogous to config/install.sh.  This requires
1424         that SMLNJ_HOME is set and that Microsoft Visual C is ready to use.
1425         (nmake etc. must be on the path, and vcvars32 must have been run.)
1426         Moreover, sources for ml-lex and ml-yacc need to exist under src,
1427         and the bootfile hierarchy must have been unpacked under
1428         sml.boot.x86-win32.
1429         The script is very primitive and does a poor job at error checking.
1430         It only installs the base system, ml-lex, and ml-yacc.  No other
1431         libraries are being installed (i.e., you get only those that
1432         are part of the compiler.)
1433       - link-sml.bat:  analogous to .link-sml, but not currently used
1434    
1435    Unrelated bug fixes:
1436    
1437      - ml-nlffigen now exports structures ST_* corresponding to incomplete
1438        types.
1439      - Added getDevice to PP/src/pp-debug-fn.sml.  (Would not compile
1440        otherwise.)
1441    
1442    ----------------------------------------------------------------------
1443    Name: Dave MacQueen (macqueen@cs.uchicago.edu)
1444    Date: 2003/06/17
1445    Tag: macqueen-20030617-bug895
1446    Description:
1447    
1448    Modified compiler/Elaborator/print/pptype.sml to fix bug 895.
1449    Tag will be used for new development branch (mcz-branch) for
1450    use by MacQueen, (Lucasz) Zairek, and (George) Cao at uchicago.
1451    
1452    ----------------------------------------------------------------------
1453    Name: Matthias Blume (blume (at) tti - c (dot) org)
1454    Date: 2003/05/27 16:55:00 CDT
1455    Tag: blume-20030527-polyeq
1456    Description:
1457    
1458    Tried to eliminated most cases of polymorphic equality.
1459    
1460    ----------------------------------------------------------------------
1461    Name: Matthias Blume (blume (at) tti - c (dot) org)
1462    Date: 2003/05/21 17:45:00 CDT
1463    Tag: blume-20030517-complete
1464    Description:
1465    
1466    Two changes:
1467    
1468      1. Added a flag for controlling whether non-exhaustive bindings will
1469         be treated as errors (default is false).
1470      2. Cleaned up the *entire* source tree so that CMB.make goes through
1471         without a single non-exhaustive match- or bind warning.
1472    
1473    ----------------------------------------------------------------------
1474    Name: Matthias Blume (blume (at) tti - c (dot) org)
1475    Date: 2003/05/17 10:20:00 CDT
1476    Tag: blume-20030517-absyn
1477    Description:
1478    
1479    1. Added cases for IF, WHILE, ANDALSO, and ORELSE to Absyn.
1480    
1481       This mainly affects the quality of error messages.  However, some
1482       of the code is now more straightforward than before. (Treatment of
1483       the above four constructs in translate.sml is much simpler than
1484       the "macro-expansion" that was going on before.  Plus, the mach-
1485       compiler no longer gets invoked just to be able to compile an
1486       if-expression.)
1487    
1488    2. The ErrorMsg.Error exception is now caught and absorbed by the
1489       interactive loop.
1490    
1491    ----------------------------------------------------------------------
1492    Name: Allen Leung
1493    Date: 2003/05/16 13:05:00 CDT
1494    Tag: leunga-20030516-cygwin-runtime
1495    Description:
1496    
1497        Ported the runtime system to cygwin, which uses the unix
1498    x86-unix bin files.  Missing/buggy features:
1499    
1500       o getnetbyname, getnetbyaddr: these functions seem to be missing in
1501         the Cygwin library.
1502       o Ctrl-C handling may be flaky.
1503       o Windows system calls and Windows I/O are not supported.
1504    
1505       A new set of binfiles is located at:
1506    
1507         http://www.dorsai.org/~leunga/boot.x86-unix.tgz
1508    
1509       This is only needed for bootstrapping the cygwin version of smlnj.
1510       Other x86 versions can use the existing binfiles.
1511    
1512    ----------------------------------------------------------------------
1513    Name: Matthias Blume
1514    Date: 2003/04/08 15:42:00 CDT
1515    Tag: blume-20030408-listpair
1516    Description:
1517    
1518    1. Added a target 'mlrisc' to installer.
1519    
1520    2. Added missing elements to structure ListPair.
1521    
1522    ----------------------------------------------------------------------
1523    Name: Allen Leung
1524    Date: 2003/01/07 10:40:00 EST
1525    Tag: leunga-20030107-int-rem
1526    Description:
1527    
1528    Fixed a bug in Int.rem(x,y) where y is a power of 2 on x86.
1529    The arguments to the SUBL instruction were swapped.
1530    
1531    ----------------------------------------------------------------------
1532    Name: Matthias Blume
1533    Date: 2002/12/12 16:25:00 EST
1534    Tag: blume-20021212-risc-ra
1535    Description:
1536    
1537    Fixed a serious bug in the rewrite code for FP spilling/reloading that
1538    sent the RA into an infinite loop when floating point registers get
1539    spilled.  (Because of this bug, e.g., nucleic stopped compiling between
1540    110.37 and 110.38.)
1541    There was another set of potential problems related to the handling of
1542    MLRISC annotations (but those did not yet cause real problems, apparently).
1543    
1544    ----------------------------------------------------------------------
1545    Name: Matthias Blume
1546    Date: 2002/12/06 22:40:00 EST
1547    Tag: blume-20021206-cm-fileid
1548    Description:
1549    
1550    Added a call of SrcPath.sync at the beginning of Parse.parse (in CM).
1551    This fixes the problem of CM getting confused by files that suddenly
1552    change their identity (e.g., by getting unlinked and recreated by some
1553    text editor such as vi).  There might be a better/cheaper/cleaner way
1554    of doing this, but for now this will have to do.
1555    
1556    ----------------------------------------------------------------------
1557    Name: Matthias Blume
1558    Date: 2002/10/28 09:50:00 EST
1559    Tag: blume-20021028-typecheck
1560    Description:
1561    
1562    Exported structure Typecheck from $smlnj/viscomp/core.cm.
1563    
1564    ----------------------------------------------------------------------
1565    Name: Matthias Blume
1566    Date: 2002/10/17 09:10:00 EDT
1567    Tag: Release_110_42
1568    Description:
1569    
1570    In good old tradition, there has been a slight hiccup so that we have
1571    to patch 110.42 after the fact.   The old release tag has been replaced
1572    (see below).
1573    
1574    The change solves a problem with two competing approaches the
1575    configuration problem regarding MacOS 10.1 vs. MacOS 10.2 which got in
1576    each other's way.
1577    
1578    This change only affects the runtime system code and the installer script.
1579    (No new bootfiles.)
1580    
1581    ----------------------------------------------------------------------
1582    Name: Matthias Blume
1583    Date: 2002/10/16 12:00:00 EDT
1584    Tag: Release_110_42_removed
1585    Description:
1586    
1587    New working release.  New bootfiles.
1588    
1589    ----------------------------------------------------------------------
1590    Name: Matthias Blume
1591    Date: 2002/10/10 13:10:00 EDT
1592    Tag: blume-20021010-ppc-divs
1593    Description:
1594    
1595    The mltree operator DIVS must be implemented with an overflow check on
1596    the PPC because the hardware indicates divide-by-zero using "overflow" as
1597    well.
1598    
1599    ----------------------------------------------------------------------
1600    Name: Matthias Blume
1601    Date: 2002/07/23 11:45:00
1602    Tag: blume-20020723-smlnj-home
1603    Description:
1604    
1605    Sml now senses the SMLNJ_HOME environment variable.  If this is set,
1606    then the bin dir is assumed to be in $SMLNJ_HOME/bin and (unless
1607    CM_PATHCONFIG is also set), the path configuration file is assumed
1608    to be in $SMLNJ_HOME/lib/pathconfig.  This way one can easily move
1609    the entire tree to some other place and everything will "just work".
1610    
1611    (Companion commands such as ml-build and ml-makedepend also sense this
1612    variable.)
1613    
1614    ----------------------------------------------------------------------
1615    Name: Matthias Blume
1616    Date: 2002/07/12 21:19:00 EDT
1617    Tag: blume-20020712-liveness
1618    Description:
1619    
1620    Exported two useful "step" functions from liveness module (MLRISC).
1621    
1622    ----------------------------------------------------------------------
1623    Name: Matthias Blume
1624    Date: 2002/07/05 16:00 EDT
1625    Tag: Release_110_41
1626    Description:
1627    
1628    New working release.  New bootfiles.
1629    
1630    ----------------------------------------------------------------------
1631    Name: Matthias Blume
1632    Date: 2002/07/05 10:25:00 EDT
1633    Tag: blume-20020705-btimp
1634    Description:
1635    
1636    Exported structure BTImp from $smlnj/viscomp/debugprof.cm so that
1637    other clients can set up backtracing support.
1638    
1639    ----------------------------------------------------------------------
1640    Name: Matthias Blume
1641    Date: 2002/06/25 17:23:00 EDT
1642    Tag: blume-20020625-fpmax
1643    Description:
1644    
1645    Fixed a bug in translation of INLMAX (and INLMIN) for the floating-point
1646    case.  (The sense of the isNaN test was reversed -- which made min and
1647    max always return their first argument.)
1648    
1649    ----------------------------------------------------------------------
1650    Name: Matthias Blume
1651    Date: 2002/06/11
1652    Tag: blume-20020611-unixpath
1653    Description:
1654    
1655    Back-ported OS.Path.{from,to}UnixPath from idlbasis-devel branch.
1656    
1657    ----------------------------------------------------------------------
1658    Name: Matthias Blume
1659    Date: 2002/06/10 16:35:00 EDT
1660    Tag: blume-20020610-ieeereal
1661    Description:
1662    
1663    I back-ported my implementation of IEEEReal.fromString from the
1664    idlbasis-devel branch so that we can test it.
1665    
1666    Another small change is that ppDec tries to give more information
1667    than just "<sig>" in the case of functors.  However, this code is
1668    broken in some mysterious way if the functor's body's signature
1669    has not been declared by ascription but gets inferred from the
1670    implementation.  This needs fixing...
1671    
1672    ----------------------------------------------------------------------
1673    Name: Matthias Blume
1674    Date: 2002/05/31
1675    Tag: blume-20020531-btrace-mode
1676    Description:
1677    
1678    Resurrected SMLofNJ.Internals.BTrace.mode.  (It accidentally fell by
1679    the wayside when I switched over to using Controls everywhere.)
1680    
1681    ----------------------------------------------------------------------
1682    Name: Lal George
1683    Date: 2002/05/23 12:21:40 EDT
1684    Tag: george-20020523-visual-labels
1685    Description:
1686    
1687    Labels are now displayed in the graphical output to make
1688    the fall-through and target blocks obvious.
1689    
1690    ----------------------------------------------------------------------
1691    Name: Matthias Blume
1692    Date: 2002/05/22 11:03:00 EDT
1693    Tag: blume-20020522-shrink
1694    Description:
1695    
1696    John tweaked yesterday's fix for 1131 to handle an out-of-memory
1697    situation that comes up when allocating huge arrays.
1698    
1699    ----------------------------------------------------------------------
1700    Name: Matthias Blume
1701    Date: 2002/05/21 16:00:00 EDT
1702    Tag: Release_110_40
1703    Description:
1704    
1705    New working release (110.40).  New bootfiles.
1706    
1707    [Also: John Reppy fixed GC bug 1131.]
1708    
1709    ----------------------------------------------------------------------
1710    Name: Matthias Blume
1711    Date: 2002/05/21 12:35:00 EDT
1712    Tag: blume-20020521-cmdoc
1713    Description:
1714    
1715    CM documentation update.
1716    
1717    ----------------------------------------------------------------------
1718    Name: Matthias Blume
1719    Date: 2002/05/21 10:55:00 EDT
1720    Tag: blume-20020521-misc
1721    Description:
1722    
1723    - John tweaked runtime to be silent on heap export (except when
1724      GC messages are on).
1725    - I added a few more things (cross-compiling versions of CMB) to
1726      config/preloads (as suggestions).
1727    
1728    ----------------------------------------------------------------------
1729    Name: Matthias Blume
1730    Date: 2002/05/20 22:25:00 EDT
1731    Tag: blume-20020520-controls
1732    Description:
1733    
1734    - Added ControlUtil structure to control-lib.cm.
1735    - Use it throughout.
1736    - Used Controls facility to define MLRISC controls (as opposed to
1737      registering MLRISC control ref cells with Controls after the fact)
1738    - Fixed messed-up controls priorities.
1739    
1740    * Removed again all the stuff from config/preloads that one wouldn't
1741      be able to preload at the time the initial heap image is built.
1742      (Many libraries, e.g., CML, do not exist yet at this time.  The
1743       only libraries that can be preloaded via config/preloads are those
1744       that come bundled with the bootfiles.)
1745    
1746    ----------------------------------------------------------------------
1747    Name: Matthias Blume
1748    Date: 2002/05/20 10:59:00 EDT
1749    Tag: blume-20020520-preloads
1750    Description:
1751    
1752    Added a lot of commented-out suggestions for things to be included
1753    in config/preloads.
1754    
1755    ----------------------------------------------------------------------
1756    Name: Allen Leung
1757    Date: 2002/05/18 14:20:00 EDT
1758    Tag: leunga-20020518-mdl
1759    Description:
1760    
1761    o Made the mdl tool stuff compile and run again.
1762    o I've disabled all the stuff that depends on RTL specifications; they
1763      are all badly broken anyway.
1764    
1765    ----------------------------------------------------------------------
1766    Name: Matthias Blume
1767    Date: 2002/05/17 16:49:00 EDT
1768    Tag: blume-20020517-controls
1769    Description:
1770    
1771    0. John Reppy made several modifications to the SML/NJ library.
1772       In particular, there is a shiny new controls-lib.cm.
1773    
1774    1. Pushed new controls interface through compiler so that everything
1775       compiles again.
1776    
1777    2. Added FormatComb and FORMAT_COMB to the CML version of the
1778       SML/NJ library (so that CML compiles again).
1779    
1780    3. Modified init scripts because XXX_DEFAULT environment variables
1781       are no longer with us.  (Boot-time initialization is now done
1782       using the same environment variables that are also used for
1783       startup-time initialization of controls.)
1784    
1785    ----------------------------------------------------------------------
1786    Name: Lal George
1787    Date: 2002/05/15 09:20:10 EDT
1788    Tag: george-20020515-pseudo-op-decls
1789    Description:
1790    
1791            All pseudo-ops emitted before the first segment declaration
1792    such as TEXT, DATA, and BSS directives are assumed to be global
1793    declarations and are emitted first in the assembly file. This is
1794    useful in a number of situations where one has pseudo-ops that are not
1795    specific to any segment, and also works around the constraint that one
1796    cannot have client pseudo-ops in the TEXT segment.
1797    
1798            Because no segment is associated with these declarations it is
1799    an error to allocate any space or objects before the first segment
1800    directive and an exception will be raised. However, we cannot make
1801    this check for client pseudo-ops.
1802    
1803            These top level declarations are a field in the CFG graph_info.
1804    In theory you can continue to add to this field after the CFG has been
1805    built -- provided you know what you are doing;-)
1806    
1807    ----------------------------------------------------------------------
1808    Name: Matthias Blume
1809    Date: 2002/05/13 16:40:00 EDT
1810    Tag: blume-20020513-pp-etc
1811    Description:
1812    
1813    A few minor bugfixes:
1814    
1815      - Stopgap measure for bug recently reported by Elsa Gunter (ppDec).
1816        (Bogus printouts for redefined bindings still occur.  Compiler
1817        bug should no longer occur now.  We need to redo the prettyprinter
1818        from scratch.)
1819    
1820      - CM pathname printer now also adds escape sequences for ( and )
1821    
1822      - commend and docu fixes for ml-nlffi
1823    
1824    ----------------------------------------------------------------------
1825    Name: Matthias Blume
1826    Date: 2002/05/10 16:40:00 EDT
1827    Tag: blume-20020510-erg-textio
1828    Description:
1829    
1830    Applied the following bugfix provided by Emden Gansner:
1831    
1832        Output is corrupted when outputSubstr is used rather than output.
1833    
1834        The problem occurs when a substring
1835    
1836            ss = (s, dataStart, dataLen)
1837    
1838        where dataStart > 0, fills a stream buffer with avail bytes left.
1839        avail bytes of s, starting at index dataStart, are copied into the
1840        buffer, the buffer is flushed, and then the remaining dataLen-avail
1841        bytes of ss are copied into the beginning of the buffer. Instead of
1842        starting this copy at index dataStart+avail in s, the current code
1843        starts the copy at index avail.
1844    
1845        Fix:
1846        In text-io-fn.sml, change line 695 from
1847             val needsFlush = copyVec(v, avail, dataLen-avail, buf, 0)
1848        to
1849             val needsFlush = copyVec(v, dataStart+avail, dataLen-avail, buf, 0)
1850    
1851    ----------------------------------------------------------------------
1852    Name: Matthias Blume
1853    Date: 2002/04/12 13:55:00 EDT
1854    Tag: blume-20020412-assyntax
1855    Description:
1856    
1857    1. Grabbed newer assyntax.h from the XFree86 project.
1858    2. Fiddled with how to compile X86.prim.asm without warnings.
1859    3. (Very) Minor cleanup in CM.
1860    
1861    ----------------------------------------------------------------------
1862    Name: Matthias Blume
1863    Date: 2002/04/01 (no joke!) 17:07:00 EST
1864    Tag: blume-20020401-x86div
1865    Description:
1866    
1867    Added full support for div/mod/rem/quot on the x86, using the machine
1868    instruction's two results (without clumsily recomputing the remainder)
1869    directly where appropriate.
1870    
1871    Some more extensive power-of-two support was added to the x86 instruction
1872    selector (avoiding expensive divs, mods, and muls where they can be
1873    replaced with cheaper shifts and masks).  However, this sort of thing
1874    ought to be done earlier, e.g., within the CPS optimizer so that
1875    all architectures benefit from it.
1876    
1877    The compiler compiles to a fixed point, but changes might be somewhat
1878    fragile nevertheless.  Please, report any strange things that you might
1879    see wrt. div/mod/quot/rem...
1880    
1881    ----------------------------------------------------------------------
1882    Name: Matthias Blume
1883    Date: 2002/03/29 17:22:00
1884    Tag: blume-20020329-div
1885    Description:
1886    
1887    Fixed my broken div/mod logic.  Unfortunately, this means that the
1888    inline code for div/mod now has one more comparison than before.
1889    Fast paths (quotient > 0 or remainder = 0) are not affected, though.
1890    The problem was with quotient = 0, because that alone does not tell
1891    us which way the rounding went.  One then has to look at whether
1892    remainder and divisor have the same sign...  :(
1893    
1894    Anyway, I replaced the bootfiles with fresh ones...
1895    
1896    ----------------------------------------------------------------------
1897    Name: Matthias Blume
1898    Date: 2002/03/29 14:10:00 EST
1899    Tag: blume-20020329-inlprims
1900    Description:
1901    
1902    NEW BOOTFILES!!!    Version number bumped to 110.39.3.
1903    
1904    Primops have changed. This means that the bin/boot-file formats have
1905    changed as well.
1906    
1907    To make sure that there is no confusion, I made a new version.
1908    
1909    
1910    CHANGES:
1911    
1912    * removed REMT from mltree (remainder should never overflow).
1913    
1914    * added primops to deal with divisions of all flavors to the frontend
1915    
1916    * handled these primops all the way through so they map to their respective
1917      MLRISC support
1918    
1919    * used these primops in the implementation of Int, Int32, Word, Word32
1920    
1921    * removed INLDIV, INLMOD, and INLREM as they are no longer necessary
1922    
1923    * parameterized INLMIN, INLMAX, and INLABS by a numkind
1924    
1925    * translate.sml now deals with all flavors of INL{MIN,MAX,ABS}, including
1926      floating point
1927    
1928    * used INL{MIN,MAX,ABS} in the implementation of Int, Int32, Word, Word32,
1929      and Real (but Real.abs maps to a separate floating-point-only primop)
1930    
1931    
1932    TODO items:
1933    
1934    * Hacked Alpha32 instruction selection, disabling the selection of REMx
1935      instructions because the machine instruction encoder cannot handle
1936      them.  (Hppa, PPC, and Sparc instruction selection did not handle
1937      REM in the first place, and REM is supported by the x86 machine coder.)
1938    
1939    * Handle DIV and MOD with DIV_TO_NEGINF directly in the x86 instruction
1940      selection phase.  (The two can be streamlined because the hardware
1941      delivers both quotient and remainder at the same time anyway.)
1942    
1943    * Think about what to do with "valOf(Int32.minInt) div ~1" and friends.
1944      (Currently the behavior is inconsistent both across architectures and
1945      wrt. the draft Basis spec.)
1946    
1947    * Word8 should eventually be handled natively, too.
1948    
1949    * There seems to be one serious bug in mltree-gen.sml.  It appears, though,
1950      as if there currently is no execution path that could trigger it in
1951      SML/NJ.  (The assumptions underlying functions arith and promotable do not
1952      hold for things like multiplication and division.)
1953    
1954    ----------------------------------------------------------------------
1955    Name: Matthias Blume
1956    Date: 2002/03/27 16:27:00 EST
1957    Tag: blume-20020327-mlrisc-divisions
1958    Description:
1959    
1960    Added support for all four division operations (ML's div, mod, quot,
1961    and rem) to MLRISC.  In the course of doing so, I also rationalized
1962    the naming (no more annoying switch-around of DIV and QUOT), by
1963    parameterizing the operation by div_rounding_mode (which can be either
1964    DIV_TO_ZERO or DIV_TO_NEGINF).
1965    
1966    The generic MLTreeGen functor takes care of compiling all four
1967    operations down to only round-to-zero div.
1968    
1969    Missing pieces:
1970    
1971      * Doing something smarter than relying on MLTreeGen on architectures
1972        like, e.g., the x86 where hardware division delivers both quotient and
1973        remainder at the same time.  With this, the implementation of the
1974        round-to-neginf operations could be further streamlined.
1975    
1976      * Remove inlining support for div/mod/rem from the frontend and replace it
1977        with primops that get carried through to the backend.  Do this for all
1978        int and word types.
1979    
1980    ----------------------------------------------------------------------
1981    Name: Matthias Blume
1982    Date: 2002/03/25 17:25:00 EST
1983    Tag: blume-20020325-divmod
1984    Description:
1985    
1986    I improved (hopefully without breaking them) the implementation of Int.div,
1987    Int.mod, and Int.rem.   For this, the code in translate.sml now takes
1988    advantage of the following observations:
1989    
1990      Let  q = x quot y      r = x rem y
1991           d = x div  y      m = x mod y
1992    
1993    where "quot" is the round-to-zero version of integer division that
1994    hardware usually provides.  Then we have:
1995    
1996         r = x - q * y        where neither the * nor the - will overflow
1997         d = if q >= 0 orelse x = q * y then q else q - 1
1998                              where neither the * nor the - will overflow
1999         m = if q >= 0 orelse r = 0 then r else r + y
2000                              where the + will not overflow
2001    
2002    This results in substantial simplification of the generated code.
2003    The following table shows the number of CFG nodes and edges generated
2004    for
2005            fun f (x, y) = x OPER y
2006            (* with OPER \in div, mod, quot, rem *)
2007    
2008    
2009        OPER | nodes(old) | edges(old) | nodes(new) | edges(new)
2010        --------------------------------------------------------
2011         div |         24 |         39 |         12 |         16
2012         mod |         41 |         71 |         12 |         16
2013        quot |          8 |         10 |          8 |         10
2014         rem |         10 |         14 |          8 |         10
2015    
2016    
2017    ----------------------------------------------------------------------
2018    Name: Matthias Blume
2019    Date: 2002/03/25 22:06:00 EST
2020    Tag: blume-20020325-cprotobug
2021    Description:
2022    
2023    Fixed a bug in cproto (c prototype decoder).
2024    
2025    ----------------------------------------------------------------------
2026    Name: Matthias Blume
2027    Date: 2002/03/25 16:00:00 EST
2028    Tag: blume-20020325-raw-primops
2029    Description:
2030    
2031    I did some cleanup to Allen's new primop code and
2032    replaced yesterday's bootfiles with new ones.
2033    (But they are stored in the same place.)
2034    
2035    ----------------------------------------------------------------------
2036    Name: Matthias Blume
2037    Date: 2002/03/24 22:40:00 EST
2038    Tag: blume-20020324-bootfiles
2039    Description:
2040    
2041    Made the bootfiles that Allen asked for.
2042    
2043    ----------------------------------------------------------------------
2044    Name: Allen Leung
2045    Date: 2002/03/23 15:50:00 EST
2046    Tag: leunga-20020323-flint-cps-rcc-primops
2047    Description:
2048    
2049      1. Changes to FLINT primops:
2050    
2051        (* make a call to a C-function;
2052         * The primop carries C function prototype information and specifies
2053         * which of its (ML-) arguments are floating point. C prototype
2054         * information is for use by the backend, ML information is for
2055         * use by the CPS converter. *)
2056      | RAW_CCALL of { c_proto: CTypes.c_proto,
2057                       ml_args: ccall_type list,
2058                       ml_res_opt: ccall_type option,
2059                       reentrant : bool
2060                     } option
2061       (* Allocate uninitialized storage on the heap.
2062        * The record is meant to hold short-lived C objects, i.e., they
2063        * are not ML pointers.  With the tag, the representation is
2064        * the same as RECORD with tag tag_raw32 (sz=4), or tag_fblock (sz=8)
2065        *)
2066      | RAW_RECORD of {tag:bool,sz:int}
2067      and ccall_type = CCALL_INT32 | CCALL_REAL64 | CCALL_ML_PTR
2068    
2069      2.  These CPS primops are now overloaded:
2070    
2071           rawload of {kind:numkind}
2072           rawstore of {kind:numkind}
2073    
2074          The one argument form is:
2075    
2076             rawload {kind} address
2077    
2078          The two argument form is:
2079    
2080             rawload {kind} [ml object, byte-offset]
2081    
2082      3. RAW_CCALL/RCC now takes two extra arguments:
2083    
2084         a. The first is whether the C call is reentrant, i.e., whether
2085            ML state should be saved and restored.
2086         b. The second argument is a string argument specifying the name of
2087            library and the C function.
2088    
2089         These things are currently not handled in the code generator, yet.
2090    
2091      4. In CProto,
2092    
2093         An encoding type of "bool" means "ml object" and is mapped into
2094         C prototype of PTR.  Note that "bool" is different than "string",
2095         even though "string" is also mapped into PTR, because "bool"
2096         is assigned an CPS type of BOGt, while "string" is assigned INT32t.
2097    
2098      5. Pickler/unpicker
2099    
2100         Changed to handle RAW_RECORD and newest RAW_CCALL
2101    
2102      6. MLRiscGen,
2103    
2104         1. Changed to handle the new rawload/rawstore/rawrecord operators.
2105         2. Code for handling C Calls has been moved to a new module CPSCCalls,
2106            in the file CodeGen/cpscompile/cps-c-calls.sml
2107    
2108      7. Added the conditional move operator
2109    
2110             condmove of branch
2111    
2112         to cps.  Generation of this is still buggy so it is currently
2113         disabled.
2114    
2115    ----------------------------------------------------------------------
2116    Name: Lal George
2117    Date: 2002/03/22 14:18:25 EST
2118    Tag: george-20020322-cps-branch-prob
2119    Description:
2120    
2121    Implemented the Ball-Larus branch prediction-heuristics, and
2122    incorporated graphical viewers for control flow graphs.
2123    
2124    Ball-Larus Heuristics:
2125    ---------------------
2126    See the file compiler/CodeGen/cpscompile/cpsBranchProb.sml.
2127    
2128    By design it uses the Dempster-Shafer theory for combining
2129    probabilities.  For example, in the function:
2130    
2131        fun f(n,acc) = if n = 0 then acc else f(n-1, n*acc)
2132    
2133    the ball-larus heuristics predicts that the n=0 is unlikely
2134    (OH-heuristic), and the 'then' branch is unlikely because of the
2135    RH-heuristic -- giving the 'then' branch an even lower combined
2136    probability using the Dempster-Shafer theory.
2137    
2138    Finally, John Reppy's loop analysis in MLRISC, further lowers the
2139    probability of the 'then' branch because of the loop in the else
2140    branch.
2141    
2142    
2143    Graphical Viewing:
2144    ------------------
2145    I merely plugged in Allen's graphical viewers into the compiler. The
2146    additional code is not much. At the top level, saying:
2147    
2148            Control.MLRISC.getFlag "cfg-graphical-view" := true;
2149    
2150    will display the graphical view of the control flow graph just before
2151    back-patching.  daVinci must be in your path for this to work. If
2152    daVinci is not available, then the default viewer can be changed
2153    using:
2154    
2155            Control.MLRISC.getString "viewer"
2156    
2157    which can be set to "dot" or "vcg" for the corresponding viewers. Of
2158    course, these viewers must be in your path.
2159    
2160    The above will display the compilation unit at the level of clusters,
2161    many of which are small, boring, and un-interesting. Also setting:
2162    
2163            Control.MLRISC.getInt "cfg-graphical-view_size"
2164    
2165    will display clusters that are larger than the value set by the above.
2166    
2167    
2168    ----------------------------------------------------------------------
2169    Name: Matthias Blume
2170    Date: 2002/03/21 22:20:00 EST
2171    Tag: blume-20020321-kmp-bugfix
2172    Description:
2173    
2174    Changed the interface to the KMP routine in PreString and fixed
2175    a minor bug in one place where it was used.
2176    
2177    ----------------------------------------------------------------------
2178    Name: Allen Leung
2179    Date: 2002/03/21 20:30:00 EST
2180    Tag: leunga-20020321-cfg
2181    Description:
2182    
2183      Fixed a potential problem in cfg edge splitting.
2184    
2185    ----------------------------------------------------------------------
2186    Name: Allen Leung
2187    Date: 2002/03/21 17:15:00 EST
2188    Tag: leunga-20020321-x86-fp-cfg
2189    Description:
2190    
2191      1. Recoded the buggy parts of x86-fp.
2192    
2193         a. All the block reordering code has been removed.
2194            We now depend on the block placement phases to do this work.
2195    
2196         b. Critical edge splitting code has been simplified and moved into the
2197            CFG modules, as where they belong.
2198    
2199         Both of these were quite buggy and complex.  The code is now much, much
2200         simpler.
2201    
2202      2. X86 backend.
2203    
2204         a. Added instructions for 64-bit support.  Instruction selection for
2205            64-bit has not been committed, however, since that
2206            requires changes to MLTREE which haven't been approved by
2207            Lal and John.
2208    
2209         b. Added support for FUCOMI and FUCOMIP when generating code for
2210            PentiumPro and above.  We only generate these instructions in
2211            the fast-fp mode.
2212    
2213         c. Added cases for JP and JNP in X86FreqProps.
2214    
2215      3. CFG
2216    
2217         CFG now has a bunch of methods for edge splitting and merging.
2218    
2219      4. Machine description.
2220    
2221         John's simplification of MLTREE_BASIS.fcond broke a few machine
2222         description things:
2223    
2224         rtl-build.{sig,sml} and hppa.mdl fixed.
2225    
2226         NOTE: the machine description stuff in the repository is still broken.
2227               Again, I can't put my fixes in because that involves
2228               changes to MLTREE.
2229    
2230    ----------------------------------------------------------------------
2231    Name: Matthias Blume
2232    Date: 2002/03/20 15:55:00 EST
2233    Tag: blume-20020320-kmp
2234    Description:
2235    
2236    Implemented Knuth-Morris-Pratt string matching in PreString and used
2237    it for String.isSubstring, Substring.isSubstring, and
2238    Substring.position.
2239    
2240    (Might need some stress-testing.  Simple examples worked fine.)
2241    
2242    ----------------------------------------------------------------------
2243    Name: Matthias Blume
2244    Date: 2002/03/19 16:37:00 EST
2245    Tag: blume-20020319-witnesses
2246    Description:
2247    
2248    Added a structure C.W and functions convert/Ptr.convert to ml-nlffi-lib.
2249    
2250    This implements a generic mechanism for changing constness qualifiers
2251    anywhere within big C types without resorting to outright "casts".
2252    (So far, functions such as C.rw/C.ro or C.Ptr.rw/C.Ptr.ro only let you
2253    modify the constness at the outermost level.)
2254    The implementation of "convert" is based on the idea of "witness"
2255    values -- values that are not used by the operation but whose types
2256    "testify" to their applicability.  On the implementation side, "convert"
2257    is simply a projection (returning its second curried argument).  With
2258    cross-module inlining, it should not result in any machine code being
2259    generated.
2260    
2261    ----------------------------------------------------------------------
2262    Name: Matthias Blume
2263    Date: 2002/03/15 16:40:00 EST
2264    Tag: blume-20020315-basis
2265    Description:
2266    
2267    Provided (preliminary?) implementations for
2268    
2269      {String,Substring}.{concatWith,isSuffix,isSubstring}
2270    
2271    and
2272    
2273      Substring.full
2274    
2275    Those are in the Basis spec but they were missing in SML/NJ.
2276    
2277    ----------------------------------------------------------------------
2278    Name: Matthias Blume
2279    Date: 2002/03/14 21:30:00 EST
2280    Tag: blume-20020314-controls
2281    Description:
2282    
2283    Controls:
2284    ---------
2285    
2286    1. Factored out the recently-added Controls : CONTROLS stuff and put
2287       it into its own library $/controls-lib.cm.  The source tree for
2288       this is under src/smlnj-lib/Controls.
2289    
2290    2. Changed the names of types and functions in this interface, so they
2291       make a bit more "sense":
2292    
2293          module -> registry
2294          'a registry -> 'a group
2295    
2296    3. The interface now deals in ref cells only.  The getter/setter interface
2297       is (mostly) gone.
2298    
2299    4. Added a function that lets one register an already-existing ref cell.
2300    
2301    5. Made the corresponding modifications to the rest of the code so that
2302       everything compiles again.
2303    
2304    6. Changed the implementation of Controls.MLRISC back to something closer
2305       to the original.  In particular, this module (and therefore MLRISC)
2306       does not depend on Controls.  There now is some link-time code in
2307       int-sys.sml that registers the MLRISC controls with the Controls
2308       module.
2309    
2310    CM:
2311    ---
2312    
2313      * One can now specify the lambda-split aggressiveness in init.cmi.
2314    
2315    ----------------------------------------------------------------------
2316    Name: Allen Leung
2317    Date: 2002/03/13 17:30:00 EST
2318    Tag: leunga-20020313-x86-fp-unary
2319    Description:
2320    
2321    Bug fix for:
2322    
2323    > leunga@weaselbane:~/Yale/tmp/sml-dist{21} bin/sml
2324    > Standard ML of New Jersey v110.39.1 [FLINT v1.5], March 08, 2002
2325    > - fun f(x,(y,z)) = Real.~ y;
2326    > [autoloading]
2327    > [autoloading done]
2328    >       fchsl   (%eax), 184(%esp)
2329    > Error: MLRisc bug: X86MCEmitter.emitInstr
2330    >
2331    > uncaught exception Error
2332    >   raised at: ../MLRISC/control/mlriscErrormsg.sml:16.14-16.19
2333    
2334    The problem was that the code generator did not generate any fp registers
2335    in this case, and the ra didn't know that it needed to run the X86FP phase to
2336    translate the pseudo fp instruction.   This only happened with unary fp
2337    operators in certain situations.
2338    
2339    ----------------------------------------------------------------------
2340    Name: Matthias Blume
2341    Date: 2002/03/13 14:00:00 EST
2342    Tag: blume-20020313-overload-etc
2343    Description:
2344    
2345    1. Added _overload as a synonym for overload for backward compatibility.
2346       (Control.overloadKW must be true for either version to be accepted.)
2347    
2348    2. Fixed bug in install script that caused more things to be installed
2349       than what was requested in config/targets.
2350    
2351    3. Made CM aware of the (_)overload construct so that autoloading
2352       works.
2353    
2354    ----------------------------------------------------------------------
2355    Name: Matthias Blume
2356    Date: 2002/03/12 22:03:00 EST
2357    Tag: blume-20020312-url
2358    Description:
2359    
2360    Forgot to update BOOT and srcarchiveurl.
2361    
2362    ----------------------------------------------------------------------
2363    Name: Matthias Blume
2364    Date: 2002/03/12 17:30:00 EST
2365    Tag: blume-20020312-version110392
2366    Description:
2367    
2368    Yet another version number bump (because of small changes to the
2369    binfile format).  Version number is now 110.39.2.  NEW BOOTFILES!
2370    
2371    Changes:
2372    
2373      The new pid generation scheme described a few weeks ago was overly
2374      complicated.  I implemented a new mechanism that is simpler and
2375      provides a bit more "stability":  Once CM has seen a compilation
2376      unit, it keeps its identity constant (as long as you do not delete
2377      those crucial CM/GUID/* files).  This means that when you change
2378      an interface, compile, then go back to the old interface, and
2379      compile again, you arrive at the original pid.
2380    
2381      There now also is a mechanism that instructs CM to use the plain
2382      environment hash as a module's pid (effectively making its GUID
2383      the empty string).  For this, "noguid" must be specified as an
2384      option to the .sml file in question within its .cm file.
2385      This is most useful for code that is being generated by tools such
2386      as ml-nlffigen (because during development programmers tend to
2387      erase the tool's entire output directory tree including CM's cached
2388      GUIDs).  "noguid" is somewhat dangerous (since it can be used to locally
2389      revert to the old, broken behavior of SML/NJ, but in specific cases
2390      where there is no danger of interface confusion, its use is ok
2391      (I think).
2392    
2393      ml-nlffigen by default generates "noguid" annotations.  They can be
2394      turned off by specifying -guid in its command line.
2395    
2396    ----------------------------------------------------------------------
2397    Name: Lal George
2398    Date: 2002/03/12 12 14:42:36 EST
2399    Tag: george-20020312-frequency-computation
2400    Description:
2401    
2402    Integrated jump chaining and static block frequency into the
2403    compiler. More details and numbers later.
2404    
2405    ----------------------------------------------------------------------
2406    Name: Lal George
2407    Date: 2002/03/11 11 22:38:53 EST
2408    Tag: george-20020311-jump-chain-elim
2409    Description:
2410    
2411    Tested the jump chain elimination on all architectures (except the
2412    hppa).  This is on by default right now and is profitable for the
2413    alpha and x86, however, it may not be profitable for the sparc and ppc
2414    when compiling the compiler.
2415    
2416    The gc test will typically jump to a label at the end of the cluster,
2417    where there is another jump to an external cluster containing the actual
2418    code to invoke gc. This is to allow factoring of common gc invocation
2419    sequences. That is to say, we generate:
2420    
2421            f:
2422               testgc
2423               ja   L1      % jump if above to L1
2424    
2425            L1:
2426               jmp L2
2427    
2428    
2429    After jump chain elimination the 'ja L1' instructions is converted to
2430    'ja L2'. On the sparc and ppc, many of the 'ja L2' instructions may end
2431    up being implemented in their long form (if L2 is far away) using:
2432    
2433            jbe     L3      % jump if below or equal to L3
2434            jmp     L2
2435         L3:
2436            ...
2437    
2438    
2439    For large compilation units L2  may be far away.
2440    
2441    
2442    ----------------------------------------------------------------------
2443    Name: Matthias Blume
2444    Date: 2002/03/11 13:30:00 EST
2445    Tag: blume-20020311-mltreeeval
2446    Description:
2447    
2448    A functor parameter was missing.
2449    
2450    ----------------------------------------------------------------------
2451    Name: Allen Leung
2452    Date: 2002/03/11 10:30:00 EST
2453    Tag: leunga-20020311-runtime-string0
2454    Description:
2455    
2456       The representation of the empty string now points to a
2457    legal null terminated C string instead of unit.  It is now possible
2458    to convert an ML string into C string with InlineT.CharVector.getData.
2459    This compiles into one single machine instruction.
2460    
2461    ----------------------------------------------------------------------
2462    Name: Allen Leung
2463    Date: 2002/03/10 23:55:00 EST
2464    Tag: leunga-20020310-x86-call
2465    Description:
2466    
2467       Added machine generation for CALL instruction (relative displacement mode)
2468    
2469    ----------------------------------------------------------------------
2470    Name: Matthias Blume
2471    Date: 2002/03/08 16:05:00
2472    Tag: blume-20020308-entrypoints
2473    Description:
2474    
2475    Version number bumped to 110.39.1.  NEW BOOTFILES!
2476    
2477    Entrypoints: non-zero offset into a code object where execution should begin.
2478    
2479    - Added the notion of an entrypoint to CodeObj.
2480    - Added reading/writing of entrypoint info to Binfile.
2481    - Made runtime system bootloader aware of entrypoints.
2482    - Use the address of the label of the first function given to mlriscGen
2483      as the entrypoint.  This address is currently always 0, but it will
2484      not be 0 once we turn on block placement.
2485    - Removed the linkage cluster code (which was The Other Way(tm) of dealing
2486      with entry points) from mlriscGen.
2487    
2488    ----------------------------------------------------------------------
2489    Name: Allen Leung
2490    Date: 2002/03/07 20:45:00 EST
2491    Tag: leunga-20020307-x86-cmov
2492    Description:
2493    
2494       Bug fixes for CMOVcc on x86.
2495    
2496       1. Added machine code generation for CMOVcc
2497       2. CMOVcc is now generated in preference over SETcc on PentiumPro or above.
2498       3. CMOVcc cannot have an immediate operand as argument.
2499    
2500    ----------------------------------------------------------------------
2501    Name: Matthias Blume
2502    Date: 2002/03/07 16:15:00 EST
2503    Tag: blume-20020307-controls
2504    Description:
2505    
2506    This is a very large but mostly boring patch which makes (almost)
2507    every tuneable compiler knob (i.e., pretty much everything under
2508    Control.* plus a few other things) configurable via both the command
2509    line and environment variables in the style CM did its configuration
2510    until now.
2511    
2512    Try starting sml with '-h' (or, if you are brave, '-H')
2513    
2514    To this end, I added a structure Controls : CONTROLS to smlnj-lib.cm which
2515    implements the underlying generic mechanism.
2516    
2517    The interface to some of the existing such facilities has changed somewhat.
2518    For example, the MLRiscControl module now provides mkFoo instead of getFoo.
2519    (The getFoo interface is still there for backward-compatibility, but its
2520    use is deprecated.)
2521    
2522    The ml-build script passes -Cxxx=yyy command-line arguments through so
2523    that one can now twiddle the compiler settings when using this "batch"
2524    compiler.
2525    
2526    TODO items:
2527    
2528    We should go through and throw out all controls that are no longer
2529    connected to anything.  Moreover, we should go through and provide
2530    meaningful (and correct!) documentation strings for those controls
2531    that still are connected.
2532    
2533    Currently, multiple calls to Controls.new are accepted (only the first
2534    has any effect).  Eventually we should make sure that every control
2535    is being made (via Controls.new) exactly once.  Future access can then
2536    be done using Controls.acc.
2537    
2538    Finally, it would probably be a good idea to use the getter-setter
2539    interface to controls rather than ref cells.  For the time being, both
2540    styles are provided by the Controls module, but getter-setter pairs are
2541    better if thread-safety is of any concern because they can be wrapped.
2542    
2543    *****************************************
2544    
2545    One bug fix: The function blockPlacement in three of the MLRISC
2546    backpatch files used to be hard-wired to one of two possibilities at
2547    link time (according to the value of the placementFlag).  But (I
2548    think) it should rather sense the flag every time.
2549    
2550    *****************************************
2551    
2552    Other assorted changes (by other people who did not supply a HISTORY entry):
2553    
2554    1. the cross-module inliner now works much better (Monnier)
2555    2. representation of weights, frequencies, and probabilities in MLRISC
2556       changed in preparation of using those for weighted block placement
2557       (Reppy, George)
2558    
2559    ----------------------------------------------------------------------
2560    Name: Lal George
2561    Date: 2002/03/07 14:44:24 EST 2002
2562    Tag: george-20020307-weighted-block-placement
2563    
2564    Tested the weighted block placement optimization on all architectures
2565    (except the hppa) using AMPL to generate the block and edge frequencies.
2566    Changes were required in the machine properties to correctly
2567    categorize trap instructions. There is an MLRISC flag
2568    "weighted-block-placement" that can be used to enable weighted block
2569    placement, but this will be ineffective without block/edge
2570    frequencies (coming soon).
2571    
2572    
2573    ----------------------------------------------------------------------
2574    Name: Lal George
2575    Date: 2002/03/05 17:24:48 EST
2576    Tag: george-20020305-linkage-cluster
2577    
2578    In order to support the block placement optimization, a new cluster
2579    is generated as the very first cluster (called the linkage cluster).
2580    It contains a single jump to the 'real' entry point for the compilation
2581    unit. Block placement has no effect on the linkage cluster itself, but
2582    all the other clusters  have full freedom in the manner in which they
2583    reorder blocks or functions.
2584    
2585    On the x86 the typical linkage code that is generated is:
2586       ----------------------
2587            .align 2
2588       L0:
2589            addl    $L1-L0, 72(%esp)
2590            jmp     L1
2591    
2592    
2593            .align  2
2594       L1:
2595       ----------------------
2596    
2597    72(%esp) is the memory location for the stdlink register. This
2598    must contain the address of the CPS function being called. In the
2599    above example, it contains the address of  L0; before
2600    calling L1 (the real entry point for the compilation unit), it
2601    must contain the address for L1, and hence
2602    
2603            addl $L1-L0, 72(%esp)
2604    
2605    I have tested this on all architectures except the hppa.The increase
2606    in code size is of course negligible
2607    
2608    ----------------------------------------------------------------------
2609    Name: Allen Leung
2610    Date: 2002/03/03 13:20:00 EST
2611    Tag: leunga-20020303-mlrisc-tools
2612    
2613      Added #[ ... ] expressions to mlrisc tools
2614    
2615    ----------------------------------------------------------------------
2616    Name: Matthias Blume
2617    Date: 2002/02/27 12:29:00 EST
2618    Tag: blume-20020227-cdebug
2619    Description:
2620    
2621    - made types in structure C and C_Debug to be equal
2622    - got rid of code duplication (c-int.sml vs. c-int-debug.sml)
2623    - there no longer is a C_Int_Debug (C_Debug is directly derived from C)
2624    
2625    ----------------------------------------------------------------------
2626    Name: Matthias Blume
2627    Date: 2002/02/26 12:00:00 EST
2628    Tag: blume-20020226-ffi
2629    Description:
2630    
2631    1. Fixed a minor bug in CM's "noweb" tool:
2632       If numbering is turned off, then truly don't number (i.e., do not
2633       supply the -L option to noweb).  The previous behavior was to supply
2634       -L'' -- which caused noweb to use the "default" line numbering scheme.
2635       Thanks to Chris Richards for pointing this out (and supplying the fix).
2636    
2637    2. Once again, I reworked some aspects of the FFI:
2638    
2639       A. The incomplete/complete type business:
2640    
2641       - Signatures POINTER_TO_INCOMPLETE_TYPE and accompanying functors are
2642         gone!
2643       - ML types representing an incomplete type are now *equal* to
2644         ML types representing their corresponding complete types (just like
2645         in C).  This is still safe because ml-nlffigen will not generate
2646         RTTI for incomplete types, nor will it generate functions that
2647         require access to such RTTI.   But when ML code generated from both
2648         incomplete and complete versions of the C type meet, the ML types
2649         are trivially interoperable.
2650    
2651         NOTE:  These changes restore the full generality of the translation
2652         (which was previously lost when I eliminated functorization)!
2653    
2654       B. Enum types:
2655    
2656       - Structure C now has a type constructor "enum" that is similar to
2657         how the "su" constructor works.  However, "enum" is not a phantom
2658         type because each "T enum" has values (and is isomorphic to
2659         MLRep.Signed.int).
2660       - There are generic access operations for enum objects (using
2661         MLRep.Signed.int).
2662       - ml-nlffigen will generate a structure E_foo for each "enum foo".
2663         * The structure contains the definition of type "mlrep" (the ML-side
2664         representation type of the enum).  Normally, mlrep is the same
2665         as "MLRep.Signed.int", but if ml-nlffigen was invoked with "-ec",
2666         then mlrep will be defined as a datatype -- thus facilitating
2667         pattern matching on mlrep values.
2668         ("-ec" will be suppressed if there are duplicate values in an
2669          enumeration.)
2670         * Constructors ("-ec") or values (no "-ec") e_xxx of type mlrep
2671         will be generated for each C enum constant xxx.
2672         * Conversion functions m2i and i2m convert between mlrep and
2673         MLRep.Signed.int.  (Without "-ec", these functions are identities.)
2674         * Coversion functions c and ml convert between mlrep and "tag enum".
2675         * Access functions (get/set) fetch and store mlrep values.
2676       - By default (unless ml-nlffigen was invoked with "-nocollect"), unnamed
2677         enumerations are merged into one single enumeration represented by
2678         structure E_'.
2679    
2680    ----------------------------------------------------------------------
2681    Name: Allen Leung
2682    Date: 2002/02/25 04:45:00 EST
2683    Tag: leunga-20020225-cps-spill
2684    
2685    This is a new implementation of the CPS spill phase.
2686    The new phase is in the new file compiler/CodeGen/cpscompile/spill-new.sml
2687    In case of problems, replace it with the old file spill.sml
2688    
2689    The current compiler runs into some serious performance problems when
2690    constructing a large record.  This can happen when we try to compile a
2691    structure with many items.  Even a very simple structure like the following
2692    makes the compiler slow down.
2693    
2694        structure Foo = struct
2695           val x_1 = 0w1 : Word32.int
2696           val x_2 = 0w2 : Word32.int
2697           val x_3 = 0w3 : Word32.int
2698           ...
2699           val x_N = 0wN : Word32.int
2700        end
2701    
2702    The following table shows the compile time, from N=1000 to N=4000,
2703    with the old compiler:
2704    
2705    N
2706    1000   CPS 100 spill                           0.04u  0.00s  0.00g
2707           MLRISC ra                               0.06u  0.00s  0.05g
2708              (spills = 0 reloads = 0)
2709           TOTAL                                   0.63u  0.07s  0.21g
2710    
2711    1100   CPS 100 spill                           8.25u  0.32s  0.64g
2712           MLRISC ra                               5.68u  0.59s  3.93g
2713              (spills = 0 reloads = 0)
2714           TOTAL                                   14.71u  0.99s  4.81g
2715    
2716    1500   CPS 100 spill                           58.55u  2.34s  1.74g
2717           MLRISC ra                               5.54u  0.65s  3.91g
2718              (spills = 543 reloads = 1082)
2719           TOTAL                                   65.40u  3.13s  6.00g
2720    
2721    2000   CPS 100 spill                           126.69u  4.84s  3.08g
2722           MLRISC ra                               0.80u  0.10s  0.55g
2723              (spills = 42 reloads = 84)
2724           TOTAL                                   129.42u  5.10s  4.13g
2725    
2726    3000   CPS 100 spill                           675.59u  19.03s  11.64g
2727           MLRISC ra                               2.69u  0.27s  1.38g
2728              (spills = 62 reloads = 124)
2729           TOTAL                                   682.48u  19.61s  13.99g
2730    
2731    4000   CPS 100 spill                           2362.82u  56.28s  43.60g
2732           MLRISC ra                               4.96u  0.27s  2.72g
2733              (spills = 85 reloads = 170)
2734           TOTAL                                   2375.26u  57.21s  48.00g
2735    
2736    As you can see the old cps spill module suffers from some serious
2737    performance problem.  But since I cannot decipher the old code fully,
2738    instead of patching the problems up, I'm reimplementing it
2739    with a different algorithm.  The new code is more modular,
2740    smaller when compiled, and substantially faster
2741    (O(n log n) time and O(n) space).  Timing of the new spill module:
2742    
2743    4000  CPS 100 spill                           0.02u  0.00s  0.00g
2744          MLRISC ra                               0.25u  0.02s  0.15g
2745             (spills=1 reloads=3)
2746          TOTAL                                   7.74u  0.34s  1.62g
2747    
2748    Implementation details:
2749    
2750    As far as I can tell, the purpose of the CPS spill module is to make sure the
2751    number of live variables at any program point (the bandwidth)
2752    does not exceed a certain limit, which is determined by the
2753    size of the spill area.
2754    
2755    When the bandwidth is too large, we decrease the register pressure by
2756    packing live variables into spill records.  How we achieve this is
2757    completely different than what we did in the old code.
2758    
2759    First, there is something about the MLRiscGen code generator
2760    that we should be aware of:
2761    
2762    o MLRiscGen performs code motion!
2763    
2764       In particular, it will move floating point computations and
2765       address computations involving only the heap pointer to
2766       their use sites (if there is only a single use).
2767       What this means is that if we have a CPS record construction
2768       statement
2769    
2770           RECORD(k,vl,w,e)
2771    
2772       we should never count the new record address w as live if w
2773       has only one use (which is often the case).
2774    
2775       We should do something similar to floating point, but the transformation
2776       there is much more complex, so I won't deal with that.
2777    
2778    Secondly, there are now two new cps primops at our disposal:
2779    
2780     1. rawrecord of record_kind option
2781        This pure operator allocates some uninitialized storage from the heap.
2782        There are two forms:
2783    
2784         rawrecord NONE [INT n]  allocates a tagless record of length n
2785         rawrecord (SOME rk) [INT n] allocates a tagged record of length n
2786                                     and initializes the tag.
2787    
2788     2. rawupdate of cty
2789          rawupdate cty (v,i,x)
2790          Assigns to x to the ith component of record v.
2791          The storelist is not updated.
2792    
2793    We use these new primops for both spilling and increment record construction.
2794    
2795     1. Spilling.
2796    
2797        This is implemented with a linear scan algorithm (but generalized
2798        to trees).  The algorithm will create a single spill record at the
2799        beginning of the cps function and use rawupdate to spill to it,
2800        and SELECT or SELp to reload from it.  So both spills and reloads
2801        are fine-grain operations.  In contrast, in the old algorithm
2802        "spills" have to be bundled together in records.
2803    
2804        Ideally, we should sink the spill record construction to where
2805        it is needed.  We can even split the spill record into multiple ones
2806        at the places where they are needed.  But CPS is not a good
2807        representation for global code motion, so I'll keep it simple and
2808        am not attempting this.
2809    
2810     2. Incremental record construction (aka record splitting).
2811    
2812        Long records with many component values which are simulatenously live
2813        (recall that single use record addresses are not considered to
2814         be live) are constructed with rawrecord and rawupdate.
2815        We allocate space on the heap with rawrecord first, then gradually
2816        fill it in with rawupdate.  This is the technique suggested to me
2817        by Matthias.
2818    
2819        Some restrictions on when this is applicable:
2820        1. It is not a VECTOR record.  The code generator currently does not handle
2821           this case. VECTOR record uses double indirection like arrays.
2822        2. All the record component values are defined in the same "basic block"
2823           as the record constructor.  This is to prevent speculative
2824           record construction.
2825    
2826    ----------------------------------------------------------------------
2827    Name: Allen Leung
2828    Date: 2002/02/22 01:02:00 EST
2829    Tag: leunga-20020222-mlrisc-tools
2830    
2831    Minor bug fixes in the parser and rewriter
2832    
2833    ----------------------------------------------------------------------
2834    Name: Allen Leung
2835    Date: 2002/02/21 20:20:00 EST
2836    Tag: leunga-20020221-peephole
2837    
2838    Regenerated the peephole files.  Some contained typos in the specification
2839    and some didn't compile because of pretty printing bugs in the old version
2840    of 'nowhere'.
2841    
2842    ----------------------------------------------------------------------
2843    Name: Allen Leung
2844    Date: 2002/02/19 20:20:00 EST
2845    Tag: leunga-20020219-mlrisc-tools
2846    Description:
2847    
2848       Minor bug fixes to the mlrisc-tools library:
2849    
2850       1.  Fixed up parsing colon suffixed keywords
2851       2.  Added the ability to shut the error messages up
2852       3.  Reimplemented the pretty printer and fixed up/improved
2853           the pretty printing of handle and -> types.
2854       4.  Fixed up generation of literal symbols in the nowhere tool.
2855       5.  Added some SML keywords to to sml.sty
2856    
2857    ----------------------------------------------------------------------
2858    Name: Matthias Blume
2859    Date: 2002/02/19 16:20:00 EST
2860    Tag: blume-20020219-cmffi
2861    Description:
2862    
2863    A wild mix of changes, some minor, some major:
2864    
2865    * All C FFI-related libraries are now anchored under $c:
2866        $/c.cm      --> $c/c.cm
2867        $/c-int.cm  --> $c/internals/c-int.cm
2868        $/memory.cm --> $c/memory/memory.cm
2869    
2870    * "make" tool (in CM) now treats its argument pathname slightly
2871      differently:
2872        1. If the native expansion is an absolute name, then before invoking
2873           the "make" command on it, CM will apply OS.Path.mkRelative
2874           (with relativeTo = OS.FileSys.getDir()) to it.
2875        2. The argument will be passed through to subsequent phases of CM
2876           processing without "going native".  In particular, if the argument
2877           was an anchored path, then "make" will not lose track of that anchor.
2878    
2879    * Compiler backends now "know" their respective C calling conventions
2880      instead of having to be told about it by ml-nlffigen.  This relieves
2881      ml-nlffigen from one of its burdens.
2882    
2883    * The X86Backend has been split into X86CCallBackend and X86StdCallBackend.
2884    
2885    * Export C_DEBUG and C_Debug from $c/c.cm.
2886    
2887    * C type encoding in ml-nlffi-lib has been improved to model the conceptual
2888      subtyping relationship between incomplete pointers and their complete
2889      counterparts.  For this, ('t, 'c) ptr has been changed to 'o ptr --
2890      with the convention of instantiating 'o with ('t, 'c) obj whenever
2891      the pointer target type is complete.  In the incomplete case, 'o
2892      will be instantiated with some "'c iobj" -- a type obtained by
2893      using one of the functors PointerToIncompleteType or PointerToCompleteType.
2894    
2895      Operations that work on both incomplete and complete pointer types are
2896      typed as taking an 'o ptr while operations that require the target to
2897      be known are typed as taking some ('t, 'c) obj ptr.
2898    
2899      voidptr is now a bit "more concrete", namely "type voidptr = void ptr'"
2900      where void is an eqtype without any values.  This makes it possible
2901      to work on voidptr values using functions meant to operate on light
2902      incomplete pointers.
2903    
2904    * As a result of the above, signature POINTER_TO_INCOMPLETE_TYPE has
2905      been vastly simplified.
2906    
2907    ----------------------------------------------------------------------
2908    Name: Matthias Blume
2909    Date: 2002/02/19 10:48:00 EST
2910    Tag: blume-20020219-pqfix
2911    Description:
2912    
2913    Applied Chris Okasaki's bug fix for priority queues.
2914    
2915    ----------------------------------------------------------------------
2916    Name: Matthias Blume
2917    Date: 2002/02/15 17:05:00
2918    Tag: Release_110_39
2919    Description:
2920    
2921    Last-minute retagging is becoming a tradition... :-(
2922    
2923    This is the working release 110.39.
2924    
2925    ----------------------------------------------------------------------
2926    Name: Matthias Blume
2927    Date: 2002/02/15 16:00:00 EST
2928    Tag: Release_110_39-orig
2929    Description:
2930    
2931    Working release 110.39.  New bootfiles.
2932    
2933    (Update: There was a small bug in the installer so it wouldn't work
2934    with all shells.  So I retagged. -Matthias)
2935    
2936    ----------------------------------------------------------------------
2937    Name: Matthias Blume
2938    Date: 2002/02/15 14:17:00 EST
2939    Tag: blume-20020215-showbindings
2940    Description:
2941    
2942    Added EnvRef.listBoundSymbols and CM.State.showBindings.  Especially
2943    the latter can be useful for exploring what bindings are available at
2944    the interactive prompt.  (The first function returns only the list
2945    of symbols that are really bound, the second prints those but also the
2946    ones that CM's autoloading mechanism knows about.)
2947    
2948    ----------------------------------------------------------------------
2949    Name: Matthias Blume
2950    Date: 2002/02/15 12:08:00 EST
2951    Tag: blume-20020215-iptrs
2952    Description:
2953    
2954    Two improvements to ml-nlffigen:
2955    
2956      1. Write files only if they do not exist or if their current contents
2957         do not coincide with what's being written.  (That is, avoid messing
2958         with the time stamps unless absolutely necessary.)
2959    
2960      2. Implement a "repository" mechanism for generated files related
2961         to "incomplete pointer types".   See the README file for details.
2962    
2963    ----------------------------------------------------------------------
2964    Name: Matthias Blume
2965    Date: 2002/02/14 11:50:00 EST
2966    Tag: blume-20020214-quote
2967    Description:
2968    
2969    Added a type 't t_' to tag.sml (in ml-nlffi-lib.cm).  This is required
2970    because of the new and improved tag generation scheme.  (Thanks to Allen
2971    Leung for pointing it out.)
2972    
2973    ----------------------------------------------------------------------
2974    Name: Lal George
2975    Date: 2002/02/14 09:55:27 EST 2002
2976    Tag: george-20020214-isabelle-bug
2977    Description:
2978    
2979    Fixed the MLRISC bug sent by Markus Wenzel regarding the compilation
2980    of Isabelle on the x86.
2981    
2982    From Allen:
2983    -----------
2984     I've found the problem:
2985    
2986         in ra-core.sml, I use the counter "blocked" to keep track of the
2987         true number of elements in the freeze queue.  When the counter goes
2988         to zero, I skip examining the queue.  But I've messed up the
2989         bookkeeping in combine():
2990    
2991             else ();
2992             case !ucol of
2993               PSEUDO => (if !cntv > 0 then
2994                     (if !cntu > 0 then blocked := !blocked - 1 else ();
2995                                        ^^^^^^^^^^^^^^^^^^^^^^^
2996                      moveu := mergeMoveList(!movev, !moveu)
2997                     )
2998                  else ();
2999    
3000         combine() is called to coalesce two nodes u and v.
3001         I think I was thinking that if the move counts of u and v are both
3002         greater than zero then after they are coalesced then one node is
3003         removed from the freeze queue.  Apparently I was thinking that
3004         both u and v are of low degree, but that's clearly not necessarily true.
3005    
3006    
3007    02/12/2002:
3008        Here's the patch.  HOL now compiles.
3009    
3010        I don't know how this impact on performance (compile
3011        time or runtime).  This bug caused the RA (especially on the x86)
3012        to go thru the potential spill phase when there are still nodes on the
3013        freeze queue.
3014    
3015    
3016    
3017    
3018    ----------------------------------------------------------------------
3019    Name: Matthias Blume
3020    Date: 2002/02/13 22:40:00 EST
3021    Tag: blume-20020213-fptr-rtti
3022    Description:
3023    
3024    Fixed a bug in ml-nlffigen that was introduced with one of the previous
3025    updates.
3026    
3027    ----------------------------------------------------------------------
3028    Name: Matthias Blume
3029    Date: 2002/02/13 16:41:00 EST
3030    Tag: blume-20020213-cmlpq
3031    Description:
3032    
3033    Added new priority queue export symbols (which have just been added to
3034    smlnj-lib.cm) to CML's version of smlnj-lib.cm.  (Otherwise CML would
3035    not compile and the installer would choke.)
3036    
3037    ----------------------------------------------------------------------
3038    Name: Matthias Blume
3039    Date: 2002/02/13 16:15:00 EST
3040    Tag: blume-20020213-various
3041    Description:
3042    
3043    1. More tweaks to ml-nlffigen:
3044    
3045       - better internal datastructures (resulting in slight speedup)
3046       - "-match" option requires exact match
3047       - "localized" gensym counters (untagged structs/unions nested within
3048         other structs/unions or within typedefs get a fresh counter; their
3049         tag will be prefixed by a concatenation of their parents' tags)
3050       - bug fixes (related to calculation of transitive closure of types
3051         to be included in the output)
3052    
3053    2. Minor Basis updates:
3054    
3055       - added implementations for List.collate and Option.app
3056    
3057    ----------------------------------------------------------------------
3058    Name: Matthias Blume
3059    Date: 2002/02/11 15:55:00 EST
3060    Tag: blume-20020211-gensym
3061    Description:
3062    
3063    Added a "-gensym" option to command line of ml-nlffigen.  This can be
3064    used to specify a "stem" -- a string that is inserted in all "gensym'd"
3065    names (ML structure names that correspond to unnamed C structs, unions,
3066    and enums), so that separate runs of ml-nlffigen do not clash.
3067    
3068    ----------------------------------------------------------------------
3069    Name: Matthias Blume
3070    Date: 2002/02/11 12:05:00 EST
3071    Tag: blume-20020211-gensml
3072    Description:
3073    
3074    A quick fix for a problem with GenSML (in the pgraph-util library):
3075    Make generation of toplevel "local" optional.  (Strictly speaking,
3076    signature definitions within "local" are not legal SML.)
3077    
3078    Other than that: updates to INSTALL and cm/TODO.
3079    
3080    ----------------------------------------------------------------------
3081    Name: Matthias Blume
3082    Date: 2002/02/08 15:00:00 EST
3083    Tag: blume-20020208-uniquepid
3084    Description:
3085    
3086    0. Version number has been bumped to 110.38.1.  NEW BOOTFILES!!!
3087    
3088    1. The installer (config/install.sh) has gotten smarter:
3089    
3090         - Configuration options are a bit easier to specify now
3091           (in config/targets).
3092         - Bug in recognizing .tar.bz2 files fixed.
3093         - Installer automatically resolves dependencies between
3094           configuration options (e.g., if you ask for eXene, you will
3095           also get cml -- regardless whether you asked for it or not).
3096         - Installer can run in "quieter mode" by setting the environment
3097           variable INSTALL_QUIETLY to "true".  "Quieter" does not mean
3098           "completely silent", though.
3099         - Build HashCons library as part of smlnj-lib.
3100    
3101    2. A new scheme for assigning persistent identifiers to compilation
3102       units (and, by extension, to types etc.) has been put into place.
3103       This fixes a long-standing bug where types and even dynamic values
3104       can get internally confused, thereby compromising type safety
3105       (abstraction) and dynamic correctness.  See
3106    
3107         http://cm.bell-labs.com/cm/cs/who/blume/pid-confusion.tgz
3108    
3109       for an example of how things could go wrong until now.
3110    
3111       The downside of the new scheme is that pids are not quite as
3112       persistent as they used to be: CM will generate a fresh pid
3113       for every compilation unit that it thinks it sees for the first
3114       time.  That means that if you compile starting from a clean, fresh
3115       source tree at two different times, you end up with different
3116       binaries.
3117    
3118       Cutoff recompilation, however, has not been compromised because
3119       CM keeps pid information in special caches between runs.
3120    
3121    ----------------------------------------------------------------------
3122    Name: Lal George
3123    Date: 2002/02/07 15:34:13 EST 2002
3124    Tag: <none>
3125    Description:
3126    
3127    Compilers that generate assembly code may produce  global labels
3128    whose value is resolved at link time. The various peephole optimization
3129    modules did not take this in account.
3130    
3131    TODO. The Labels.addrOf function should really return an option
3132    type so that clients are forced to deal with this issue, rather
3133    than an exception being raised.
3134    
3135    ----------------------------------------------------------------------
3136    Name: Lal George
3137    Date: 2002/02/06 13:55:02 EST
3138    Tag: george-20020206-ra-breakup
3139    Description:
3140    
3141    1. A bug fix from Allen.
3142    
3143        A typo causes extra fstp %st(0)'s to be generated at compensation
3144        edges, which might cause stack underflow traps at runtime.  This
3145        occurs in fft where there are extraneous fstps right before the 'into'
3146        trap instruction (in this case they are harmless since none of the
3147        integers overflow.)
3148    
3149    2. Pulled out various utility modules that were embedded in the modules
3150       of the register allocator. I need these modules for other purposes, but
3151       they are not complete enough to put into a library (just yet).
3152    ----------------------------------------------------------------------
3153    Name: Matthias Blume
3154    Date: 2002/01/31 16:05:00 EST
3155    Tag: blume-20020131-sparc-ccalls
3156    Description:
3157    
3158    1. C-calls on Sparc needlessly allocated a huge chunk (96 bytes)
3159       of extra stack space by mistake.  Fixed.
3160    
3161    2. Bug in logic of handling of command-line options in ml-nlffigen fixed.
3162    
3163    ----------------------------------------------------------------------
3164    Name: Allen Leung
3165    Date: 2002/01/30
3166    Tag: leunga-20020130-nowhere-bug-fix
3167    Description:
3168    
3169       MLRISC bug fixes:
3170       1. Fixed a bindings computation bug in the 'nowhere' program generator tool.
3171       2. MachineInt.fromString was negating its value.
3172    
3173    ----------------------------------------------------------------------
3174    Name: Matthias Blume
3175    Date: 2002/01/29
3176    Tag: blume-20020129-INSTALL
3177    Description:
3178    
3179    - Added somewhat detailed installation instructions (file INSTALL).
3180    - Fixed curl-detection bug in config/install.sh.
3181    - It is now possible to select the URL getter using the URLGETTER
3182      environment variable:
3183    
3184          not set / "unknown"      --> automatic detection (script tries wget,
3185                                       curl, and lynx)
3186          "wget" / "curl" / "lynx" --> use the specified program (script "knows"
3187                                       how to properly invoke them)
3188          other                    --> use $URLGETTER directly, it must take
3189                                       precisely two command-line arguments
3190                                       (source URL and destination file name)
3191    
3192    ----------------------------------------------------------------------
3193    Name: Matthias Blume
3194    Date: 2002/01/28
3195    Tag: blume-20020128-sparc-ccalls
3196    Description:
3197    
3198    - Fixed problem with calculation of "used" registers in sparc-c-calls.
3199    - Make use of the allocParam argument in sparc-c-calls.
3200    
3201    ----------------------------------------------------------------------
3202    Name: Matthias Blume
3203    Date: 2002/01/28
3204    Tag: blume-20020128-allocParam
3205    Description:
3206    
3207    John Reppy:  Changes c-calls API to accept client-callback for
3208    allocating extra stack space.
3209    me: Corresponding changes to mlriscGen (using a dummy argument that
3210        does not change the current behavior).
3211    
3212    ----------------------------------------------------------------------
3213    Name: Matthias Blume
3214    Date: 2002/01/28 12:00:00
3215    Tag: Release_110_38
3216    Description:
3217    
3218    This time for real!!!
3219    
3220    ----------------------------------------------------------------------
3221    Name: Matthias Blume
3222    Date: 2002/01/28 10:56:00 EST
3223    Tag: blume-20020128-retraction
3224    Description:
3225    
3226    0. Retracted earlier 110.38.  (The Release_110_38 tag has been replaced
3227       with blume-Release_110_38-retracted.)
3228    
3229    1. Fixed a problem with incorrect rounding modes in real64.sml.
3230       (Thanks to Andrew Mccreight <andrew.mccreight@yale.edu>.)
3231    
3232    2. A bug in ml-nlffigen related to the handling of unnamed structs, unions,
3233       and enums fixed.  The naming of corresponding ML identifiers should
3234       now be consistent again.
3235    
3236    ----------------------------------------------------------------------
3237    Name: Allen Leung
3238    Date: 2002/01/27
3239    Tag: leunga-20020127-nowhere
3240    Description:
3241    
3242       Added a target called nowhere in the configuration scripts.
3243       Enabling this will build the MLRISC 'nowhere' tool (for translating
3244       programs with where-clauses into legal SML code) during installation.
3245    
3246    ----------------------------------------------------------------------
3247    Name: Matthias Blume
3248    Date: 2002/01/25 21:27:00 EST
3249    Tag: blume-Release_110_38-retracted
3250    Description:
3251    
3252    Call it a (working) release!  Version is 110.38. Bootfiles are ready.
3253    
3254    README will be added later.
3255    
3256    !!! NOTE:  Re-tagged as blume-Release_110_38-retracted. Original tag
3257    (Release_110_38) removed.  Reason: Last-minute bug fixes.
3258    
3259    ----------------------------------------------------------------------
3260    Name: Matthias Blume
3261    Date: 2002/01/25
3262    Tag: blume-20020125-ffi
3263    Description:
3264    
3265    A large number of tweaks and improvements to ml-nlffi-lib and
3266    ml-nlffigen:
3267    
3268       - ML represenation types have been streamlined
3269       - getter and setter functions work with concrete values, not abstract
3270         ones where possible
3271       - ml-nlffigen command line more flexible (see README file there)
3272       - some bugs have been fixed (hopefully)
3273    
3274    ----------------------------------------------------------------------
3275    Name: Lal George
3276    Date: 2002/01/24
3277    Tag: george-20020124-risc-ra-interface
3278    Description:
3279    
3280       There is a dramatic simplification in the interface to the
3281       register allocator for RISC architectures as a result of making
3282       parallel copy instructions explicit.
3283    
3284    ----------------------------------------------------------------------
3285    Name: Matthias Blume
3286    Date: 2002/01/22
3287    Tag: blume-20020122-x86-ccalls
3288    Description:
3289    
3290    Bug fix for c-calls on x86 (having to do with how char- and
3291    short-arguments are being handled).
3292    
3293    ----------------------------------------------------------------------
3294    Name: Matthias Blume
3295    Date: 2002/01/21
3296    Tag: blume-20020121-ff
3297    Description:
3298    
3299    Another day of fiddling with the FFI...
3300    
3301    1. Bug fix/workaround:  CKIT does not complain about negative array
3302       dimensions, so ml-nlffigen has to guard itself against this possibility.
3303       (Otherwise a negative dimension would send it into an infinite loop.)
3304    
3305    2. Some of the abstract types (light objects, light pointers, most "base"
3306       types) in structure C are now eqtypes.
3307    
3308    3. Added constructors and test functions for NULL function pointers.
3309    
3310    ----------------------------------------------------------------------
3311    Name: Matthias Blume
3312    Date: 2002/01/18
3313    Tag: blume-20020118-ready-for-new-release
3314    Description:
3315    
3316    Made config/srcarchiveurl point to a new place.  (Will provide boot
3317    files shortly.)
3318    
3319    Maybe we christen this to be 110.38?
3320    
3321    ----------------------------------------------------------------------
3322    Name: Matthias Blume
3323    Date: 2002/01/18
3324    Tag: blume-20020118-more-ffifiddle
3325    Description:
3326    
3327    Today's FFI fiddling:
3328    
3329      - Provided a structure CGetSet with "convenient" versions of C.Get.* and
3330        C.Set.* that use concrete (MLRep.*) arguments and results instead
3331        of abstract ones.
3332    
3333      - Provided word-style bit operations etc. for "int" representation
3334        types in MLRep.S<Foo>Bitops where <Foo> ranges over Char, Int, Short,
3335        and Long.
3336    
3337    ----------------------------------------------------------------------
3338    Name: Matthias Blume
3339    Date: 2002/01/18
3340    Tag: blume-20020118-use-x86-fp
3341    Description:
3342    
3343    Now that x86-fast-fp seems to be working, I turned it back on again
3344    by default.  (Seems to work fine now, even with the FFI.)
3345    
3346    Other than that, I added some documentation about the FFI to
3347    src/ml-nlffigen/README and updated the FFI test examples in
3348    src/ml-nlffi-lib/Tests/*.
3349    
3350    ----------------------------------------------------------------------
3351    Name: Allen Leung
3352    Date: 2002/01/17
3353    Tag: leunga-20020117-x86-fast-fp-call
3354    Description:
3355    
3356       1. Fixed a problem with handling return fp values when x86's fast fp
3357          mode is turned on.
3358    
3359       2. Minor pretty printing fix for cellset.  Print %st(0) as %st(0) instead
3360          of %f32.
3361    
3362       3. Added a constructor INT32lit to the ast of MLRISC tools.
3363    
3364    ----------------------------------------------------------------------
3365    Name: Matthias Blume
3366    Date: 2002/01/16
3367    Tag: blume-20020116-ffifiddle
3368    Description:
3369    
3370    More fiddling with the FFI interface:
3371    
3372     - Make constness 'c instead of rw wherever possible.  This eliminates
3373       the need for certain explicit coercions.  (However, due to ML's
3374       value polymorphism, there will still be many cases where explicit
3375       coercions are necessary.  Phantom types are not the whole answer
3376       to modeling a subtyping relationship in ML.)
3377    
3378     - ro/rw coersions for pointers added.  (Avoids the detour through */&.)
3379    
3380     - "printf" test example added to src/ml-nlffi-lib/Tests.  (Demonstrates
3381       clumsy workaround for varargs problem.)
3382    
3383    ----------------------------------------------------------------------
3384    Name: Lal George
3385    Date: 2002/01/15
3386    Tag: <none>
3387    Description:
3388    
3389    1. Since COPY instructions are no longer native to the architecture,
3390       a generic functor can be used to implement the expandCopies function.
3391    
3392    2. Allowed EXPORT and IMPORT pseudo-op declarations to appear inside a
3393       TEXT segment.
3394    
3395    ----------------------------------------------------------------------
3396    Name: Matthias Blume
3397    Date: 2002/01/15
3398    Tag: blume-20020115-ffiupdates
3399    Description:
3400    
3401    1. Fix for bug resulting in single-precision float values being returned
3402       incorrectly from FFI calls.
3403    
3404    2. Small modifications to C FFI API:
3405    
3406        - memory-allocation routines return straight objects (no options)
3407          and raise an exception in out-of-memory situations
3408        - unsafe extensions to cast between function pointers and pointers
3409          from/to ints
3410        - added structure C_Debug as an alternative to structure C where
3411          pointer-dereferencing (|*| and |*!) always check for null-pointers
3412        - added open_lib' to DynLinkage;  open_lib' works like open_lib
3413          but also takes a (possibly empty) list of existing library handles
3414          that the current library depends on
3415    
3416    ----------------------------------------------------------------------
3417    Name: Matthias Blume
3418    Date: 2002/01/10
3419    Tag: blume-20020110-newffigen
3420    Description:
3421    
3422    1. Updates to portable graph code.
3423    
3424    2. Major update to ml-nlffigen and ml-nlffi-lib.  Things are much
3425       more scalable now so that even huge interfaces such as the one
3426       for GTK compile in finite time and space. :-)
3427       See src/ml-nlffigen/README for details on what's new.
3428    
3429    ----------------------------------------------------------------------
3430    Name: Lal George
3431    Date: 2001/01/09 14:31:35 EST 2002
3432    Tag: george-20011206-rm-native-copy
3433    Description:
3434    
3435            Removed the native COPY and FCOPY instructions
3436            from all the architectures and replaced it with the
3437            explicit COPY instruction from the previous commit.
3438    
3439            It is now possible to simplify many of the optimizations
3440            modules that manipulate copies. This has not been
3441            done in this change.
3442    
3443    ----------------------------------------------------------------------
3444    Name: Lal George
3445    Date: 2001/12/06 16:50:13 EST 2001
3446    Tag: george-20011206-mlrisc-instruction
3447    Description:
3448    
3449    Changed the representation of instructions from being fully abstract
3450    to being partially concrete. That is to say:
3451    
3452      from
3453            type instruction
3454    
3455      to
3456            type instr                              (* machine instruction *)
3457    
3458            datatype instruction =
3459                LIVE of {regs: C.cellset, spilled: C.cellset}
3460              | KILL of {regs: C.cellset, spilled: C.cellset}
3461              | COPYXXX of {k: CB.cellkind, dst: CB.cell list, src: CB.cell list}
3462              | ANNOTATION of {i: instruction, a: Annotations.annotation}
3463              | INSTR of instr
3464    
3465    This makes the handling of certain special instructions that appear on
3466    all architectures easier and uniform.
3467    
3468    LIVE and KILL say that a list of registers are live or killed at the
3469    program point where they appear. No spill code is generated when an
3470    element of the 'regs' field is spilled, but the register is moved to
3471    the 'spilled' (which is present, more for debugging than anything else).
3472    
3473    LIVE replaces the (now deprecated) DEFFREG instruction on the alpha.
3474    We used to generate:
3475    
3476            DEFFREG f1
3477            f1 := f2 + f3
3478            trapb
3479    
3480    but now generate:
3481    
3482            f1 := f2 + f3
3483            trapb
3484            LIVE {regs=[f1,f2,f3], spilled=[]}
3485    
3486    Furthermore, the DEFFREG (hack) required that all floating point instruction
3487    use all registers mentioned in the instruction. Therefore f1 := f2 + f3,
3488    defines f1 and uses [f1,f2,f3]! This hack is no longer required resulting
3489    in a cleaner alpha implementation. (Hopefully, intel will not get rid of
3490    this architecture).
3491    
3492    COPYXXX is intended to replace the parallel COPY and FCOPY  available on
3493    all the architectures. This will result in further simplification of the
3494    register allocator that must be aware of them for coalescing purposes, and
3495    will also simplify certain aspects of the machine description that provides
3496    callbacks related to parallel copies.
3497    
3498    ANNOTATION should be obvious, and now INSTR represents the honest to God
3499    machine instruction set!
3500    
3501    The <arch>/instructions/<arch>Instr.sml files define certain utility
3502    functions for making porting easier -- essentially converting upper case
3503    to lower case. All machine instructions (of type instr) are in upper case,
3504    and the lower case form generates an MLRISC instruction. For example on
3505    the alpha we have:
3506    
3507      datatype instr =
3508         LDA of {r:cell, b:cell, d:operand}
3509       | ...
3510    
3511      val lda : {r:cell, b:cell, d:operand} -> instruction
3512        ...
3513    
3514    where lda is just (INSTR o LDA), etc.
3515    
3516    ----------------------------------------------------------------------
3517    Name: Matthias Blume
3518    Date: 2001/11/22 21:40:00 EST
3519    Tag: Release_110_37
3520    Description:
3521    
3522    Release 110.37.  This time for real.
3523    
3524    ----------------------------------------------------------------------
3525    Name: Matthias Blume
3526    Date: 2001/11/21 16:35:00 EST
3527    Tag: blume-20011121-foot-in-mouth
3528    Description:
3529    
3530    Removed the "Release_110_37" tag because of a serious bug.
3531    This will be re-tagged once the bug is fixed.
3532    
3533    ----------------------------------------------------------------------
3534    Name: Matthias Blume
3535    Date: 2001/11/21 16:14:00 EST
3536    Tag: blume-20011121-forgottenfile
3537    Description:
3538    
3539    Forgot to add a file.  (Just a .tex-file -- part of
3540    the CM manual source.)
3541    
3542    ----------------------------------------------------------------------
3543    Name: Matthias Blume
3544    Date: 2001/11/21 16:10:00 EST
3545    Tag: blume-20011121-invalid_110_37
3546    Description:
3547    
3548    Note: I removed the original tag "Release_110_37" from this commit
3549          because we found a serious bug in all non-x86 backends.
3550          - Matthias
3551    
3552    1. Modifications to the SML/NJ code generator and to the runtime system
3553       so that code object name strings are directly inserted into code
3554       objects at code generation time.  The only business the runtime system
3555       has with this is now to read the name strings on occasions.
3556       (The encoding of the name string has also changed somewhat.)
3557    
3558    2. CM now implements a simple "set calculus" for specifying export lists.
3559       In particular, it is now possible to refer to the export lists of
3560       other libraries/groups/sources and form unions as well as differences.
3561       See the latest CM manual for details.
3562    
3563    3. An separate notion of "proxy" libraries has again be eliminated from
3564       CM's model.  (Proxy libraries are now simply a special case of using
3565       the export list calculus.)
3566    
3567    4. Some of the existing libraries now take advantage of the new set
3568       calculus.
3569       (Notice that not all libraries have been converted because some
3570       of the existing .cm-files are supposed to be backward compatible
3571       with 110.0.x.)
3572    
3573    5. Some cleanup in stand-alone programs.  (Don't use "exnMessage" -- use
3574       "General.exnMessage"!  The former relies on a certain hook to be
3575       initialized, and that often does not happen in the stand-alone case.)
3576    
3577    ----------------------------------------------------------------------
3578    Name: Lal George
3579    Date: 2001/11/21  13:56:18 EST
3580    Tag: george-2001121-pseudo-ops
3581    Description:
3582    
3583      Implemented a complete redesign of MLRISC pseudo-ops. Now there
3584      ought to never be any question of incompatabilities with
3585      pseudo-op syntax expected by host assemblers.
3586    
3587      For now, only modules supporting GAS syntax are implemented
3588      but more should follow, such as MASM, and vendor assembler
3589      syntax, e.g. IBM as, Sun as, etc.
3590    
3591    ----------------------------------------------------------------------
3592    Name: Matthias Blume
3593    Date: 2001/11/14 11:52:00 EST
3594    Tag: blume-20011114-srcname
3595    Description:
3596    
3597    1. Routed the name of the current source file to mlriscgen where it
3598       should be directly emitted into the code object.  (This last part
3599       is yet to be done.)
3600    
3601    2. Some cleanup of the pgraph code to make it match the proposal that
3602       I put out the other day.  (The proposal notwithstanding, things are
3603       still in flux here.)
3604    
3605    ----------------------------------------------------------------------
3606    Name: Lal George
3607    Date: 2001/11/14 09:44:04 EST
3608    Tag:
3609    Description:
3610    
3611      Fix for a backpatching bug reported by Allen.
3612    
3613      Because the boundary between short and long span-dependent
3614      instructions is +/- 128, there are an astounding number of
3615      span-dependent instructions whose size is over estimated.
3616    
3617      Allen came up with the idea of letting the size of span
3618      dependent instructions be non-monotonic, for a maxIter
3619      number of times, after which the size must be monotonically
3620      increasing.
3621    
3622      This table shows the number of span-dependent instructions
3623      whose size was over-estimated as a function of maxIter, for the
3624      file Parse/parse/ml.grm.sml:
3625    
3626         maxIter            # of instructions:
3627            10                      687
3628            20                      438
3629            30                      198
3630            40                        0
3631    
3632      In compiling the compiler, there is no significant difference in
3633      compilation speed between maxIter=10 and maxIter=40. Actually,
3634      my measurements showed that maxIter=40 was a tad faster than
3635      maxIter=10! Also 96% of the  files in the compiler reach a fix
3636      point within 13 iterations, so fixing maxIter at 40, while high,
3637      is okay.
3638    
3639    ----------------------------------------------------------------------
3640    Name: Matthias Blume
3641    Date: 2001/10/31 15:25:00 EST
3642    Tag: blume-20011031-pgraph
3643    Description:
3644    
3645    CKIT:
3646    * Changed the "Function" constructor of type Ast.ctype to carry optional
3647      argument identifiers.
3648    * Changed the return type of TypeUtil.getFunction accordingly.
3649    * Type equality ignores the argument names.
3650    * TypeUtil.composite tries to preserve argument names but gives up quickly
3651      if there is a mismatch.
3652    
3653    installation script:
3654    * attempts to use "curl" if available (unless "wget" is available as well)
3655    
3656    CM:
3657    * has an experimental implementation of "portable graphs" which I will
3658      soon propose as an implementation-independent library format
3659    * there are also new libraries $/pgraph.cm and $/pgraph-util.cm
3660    
3661    NLFFI-LIB:
3662    * some cleanup (all cosmetic)
3663    
3664    NLFFIGEN:
3665    * temporarily disabled the mechanism that suppresses ML output for
3666      C definitions whose identifiers start with an underscore character
3667    * generate val bindings for enum constants
3668    * user can request that only one style (light or heavy) is being used;
3669      default is to use both (command-line arguments: -heavy and -light)
3670    * fixed bug in handling of function types involving incomplete pointers
3671    * generate ML entry points that take record arguments (i.e., using
3672      named arguments) for C functions that have a prototype with named
3673      arguments
3674      (see changes to CKIT)
3675    
3676    ----------------------------------------------------------------------
3677    Name: Allen Leung
3678    Date: 2001/10/27 20:34:00 EDT
3679    Tag: leunga-20011027-x86-fast-fp-call
3680    Description:
3681    
3682       Fixed the bug described in blume-20010920-slowfp.
3683    
3684       The fix involves
3685          1. generating FCOPYs in FSTP in ia32-svid
3686          2. marking a CALL with the appropriate annotation
3687    
3688    ----------------------------------------------------------------------
3689    Name: Matthias Blume
3690    Date: 2001/10/16 11:32:00 EDT
3691    Tag: blume-20011016-netbsd
3692    Description:
3693    
3694    Underscore patch from Chris Richards (fixing problem with compiling
3695    runtime system under recent NetBSD).
3696    
3697    ----------------------------------------------------------------------
3698    Name: Allen Leung
3699    Date: 2001/10/12 17:18:32 EDT 2001
3700    Tag: leung-20011012-x86-printflowgraph
3701    Description:
3702    
3703    X86RA now uses a valid (instead of dummy) PrintFlowgraph module.
3704    
3705    ----------------------------------------------------------------------
3706    Name: Lal George
3707    Date: 2001/10/11 23:51:34 EDT
3708    Tag: george-20011011-too-many-instrs
3709    Description:
3710    
3711    The representation of a program point never expected to see more
3712    than 65536 instructions in a basic block!
3713    
3714    ----------------------------------------------------------------------
3715    Name: Lal George
3716    Date: 2001/10/09 09:41:37 EDT
3717    Tag: george-20011008-mlrisc-labels
3718    Description:
3719    
3720    Changed the machine description files to support printing of
3721    local and global labels in assembly code, based on host assembler
3722    conventions.
3723    
3724    ----------------------------------------------------------------------
3725    Name: Matthias Blume
3726    Date: 2001/09/25 15:25:00 EDT
3727    Tag: blume-20010925-exninfo
3728    Description:
3729    
3730    I provided a non-hook implementation of exnName (at the toplevel) and
3731    made the "dummy" implementation of exnMessage (at the toplevel) more
3732    useful: if nothing gets "hooked in", then at least you are going to
3733    see the exception name and a message indicating why you don't see more.
3734    
3735    [For the time being, programs that need exnMessage and want to use
3736    ml-build should either use General.exnMessage (strongly recommended) or
3737    refer to structure General at some other point so that CM sees a
3738    static dependency.]
3739    
3740    [Similar remarks go for "print" and "use":  If you want to use their
3741    functionality in stand-alone programs generated by ml-build, then use
3742    TextIO.output and Backend.Interact.useFile (from $smlnj/compiler.cm).]
3743    
3744    ----------------------------------------------------------------------
3745    Name: Matthias Blume
3746    Date: 2001/09/20 17:28:00 EDT
3747    Tag: blume-20010920-slowfp
3748    Description:
3749    
3750    Allen says that x86-fast-fp is not safe yet, so I turned it off again...
3751    
3752    ----------------------------------------------------------------------
3753    Name: Matthias Blume
3754    Date: 2001/09/20 17:20:00 EDT
3755    Tag: blume-20010920-canonicalpaths
3756    Description:
3757    
3758    0. Updated the BOOT file (something that I forgot to do earlier).
3759    
3760    1. Small internal change to CM so that it avoids "/../" in filenames
3761       as much as possible (but only where it is safe).
3762    
3763    2. Changed config/_run-sml (resulting in a changed bin/.run-sml) so
3764       that arguments that contain delimiters are passed through correctly.
3765       This change also means that all "special" arguments of the form
3766       @SMLxxx... must come first.
3767    
3768    3. Changed install script to put relative anchor names for tool commands
3769       into pathconfig.
3770    
3771    ----------------------------------------------------------------------
3772    Name: Matthias Blume
3773    Date: 2001/09/18 15:35:00 EDT
3774    Tag: blume-20010918-readme11036
3775    Description:
3776    
3777    Added README files.
3778    
3779    ----------------------------------------------------------------------
3780    Name: Matthias Blume
3781    Date: 2001/09/18 11:45:00 EDT
3782    Tag: Release_110_36 (retag)
3783    Description:
3784    
3785    Fixed mistake in config/preloads. Retagged as 110.36.
3786    
3787    ----------------------------------------------------------------------
3788    Name: Matthias Blume
3789    Date: 2001/09/18 09:40:00 EDT
3790    Tag: Release_110_36_orig (tag changed)
3791    Description:
3792    
3793    New version (110.36).  New bootfiles.
3794    
3795    ----------------------------------------------------------------------
3796    Name: Matthias Blume
3797    Date: 2001/09/14 16:15:00 EDT
3798    Tag: blume-20010914-x86fastfp
3799    Description:
3800    
3801    John committed some changes that Allen made, in particular a (hopefully)
3802    correctly working version of the x86-fp module.
3803    
3804    I changed the default setting of the Control.MLRISC.getFlag "x86-fast-fp"
3805    flag to "true".  Everything seems to compile to a fixpoint ok, and
3806    "mandelbrot" speeds up by about 15%.
3807    
3808    ----------------------------------------------------------------------
3809    Name: Matthias Blume
3810    Date: 2001/09/13 11:20:00 EDT
3811    Tag: blume-20010913-minimal
3812    Description:
3813    
3814    1. Stefan Monnier's patch to fix a miscompilation problem that
3815       was brought to light by John Reppy's work on Moby.
3816    
3817    2. Implemented a minimal "structure Compiler" that contains just
3818       "version" and "architecture".  The minimal version will be
3819       available when the full version is not.  This is for backward-
3820       compatibility with code that wants to test Compiler.version.
3821    
3822    ----------------------------------------------------------------------
3823    Name: Matthias Blume
3824    Date: 2001/08/28 14:03:00 EDT
3825    Tag: blume-20010828-ml-lex
3826    Description:
3827    
3828    Fix for bug 1581, received from Neophytos Michael.
3829    
3830    ----------------------------------------------------------------------
3831    Name: Matthias Blume
3832    Date: 2001/08/27 11:20:00 EDT
3833    Tag: blume-20010827-readme11035
3834    Description:
3835    
3836    Fleshed out the README file for 110.35.
3837    
3838    ----------------------------------------------------------------------
3839    Name: Matthias Blume
3840    Date: 2001/08/24 17:10:00 EDT
3841    Tag: Release_110_35
3842    Description:
3843    
3844    New version number (110.35).  New bootfiles.
3845    
3846    ----------------------------------------------------------------------
3847    Name: Lal George
3848    Date: 2001/08/24 13:47:18 EDT 2001
3849    Tag: george-20010824-MLRISC-graphs
3850    Description:
3851