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

Parent Directory | Revision Log | Patch

	revision 890, Thu Jul 19 20:38:56 2001 UTC	revision 993, Fri Nov 23 02:37:10 2001 UTC
#	Line 13	Line 13
13	Description:	Description:
14
15	----------------------------------------------------------------------	----------------------------------------------------------------------
16		Name: Matthias Blume
17		Date: 2001/11/22 21:40:00 EST
18		Tag: Release_110_37
19		Description:
20
21		Release 110.37.  This time for real.
22
23		----------------------------------------------------------------------
24		Name: Matthias Blume
25		Date: 2001/11/21 16:35:00 EST
26		Tag: blume-20011121-foot-in-mouth
27		Description:
28
29		Removed the "Release_110_37" tag because of a serious bug.
30		This will be re-tagged once the bug is fixed.
31
32		----------------------------------------------------------------------
33		Name: Matthias Blume
34		Date: 2001/11/21 16:14:00 EST
35		Tag: blume-20011121-forgottenfile
36		Description:
37
38		Forgot to add a file.  (Just a .tex-file -- part of
39		the CM manual source.)
40
41		----------------------------------------------------------------------
42		Name: Matthias Blume
43		Date: 2001/11/21 16:10:00 EST
44		Tag: blume-20011121-invalid_110_37
45		Description:
46
47		Note: I removed the original tag "Release_110_37" from this commit
48		because we found a serious bug in all non-x86 backends.
49		- Matthias
50
51		1. Modifications to the SML/NJ code generator and to the runtime system
52		so that code object name strings are directly inserted into code
53		objects at code generation time.  The only business the runtime system
54		has with this is now to read the name strings on occasions.
55		(The encoding of the name string has also changed somewhat.)
56
57		2. CM now implements a simple "set calculus" for specifying export lists.
58		In particular, it is now possible to refer to the export lists of
59		other libraries/groups/sources and form unions as well as differences.
60		See the latest CM manual for details.
61
62		3. An separate notion of "proxy" libraries has again be eliminated from
63		CM's model.  (Proxy libraries are now simply a special case of using
64		the export list calculus.)
65
66		4. Some of the existing libraries now take advantage of the new set
67		calculus.
68		(Notice that not all libraries have been converted because some
69		of the existing .cm-files are supposed to be backward compatible
70		with 110.0.x.)
71
72		5. Some cleanup in stand-alone programs.  (Don't use "exnMessage" -- use
73		"General.exnMessage"!  The former relies on a certain hook to be
74		initialized, and that often does not happen in the stand-alone case.)
75
76		----------------------------------------------------------------------
77		Name: Lal George
78		Date: 2001/11/21  13:56:18 EST
79		Tag: george-2001121-pseudo-ops
80		Description:
81
82		Implemented a complete redesign of MLRISC pseudo-ops. Now there
83		ought to never be any question of incompatabilities with
84		pseudo-op syntax expected by host assemblers.
85
86		For now, only modules supporting GAS syntax are implemented
87		but more should follow, such as MASM, and vendor assembler
88		syntax, e.g. IBM as, Sun as, etc.
89
90		----------------------------------------------------------------------
91		Name: Matthias Blume
92		Date: 2001/11/14 11:52:00 EST
93		Tag: blume-20011114-srcname
94		Description:
95
96		1. Routed the name of the current source file to mlriscgen where it
97		should be directly emitted into the code object.  (This last part
98		is yet to be done.)
99
100		2. Some cleanup of the pgraph code to make it match the proposal that
101		I put out the other day.  (The proposal notwithstanding, things are
102		still in flux here.)
103
104		----------------------------------------------------------------------
105		Name: Lal George
106		Date: 2001/11/14 09:44:04 EST
107		Tag:
108		Description:
109
110		Fix for a backpatching bug reported by Allen.
111
112		Because the boundary between short and long span-dependent
113		instructions is +/- 128, there are an astounding number of
114		span-dependent instructions whose size is over estimated.
115
116		Allen came up with the idea of letting the size of span
117		dependent instructions be non-monotonic, for a maxIter
118		number of times, after which the size must be monotonically
119		increasing.
120
121		This table shows the number of span-dependent instructions
122		whose size was over-estimated as a function of maxIter, for the
123		file Parse/parse/ml.grm.sml:
124
125		maxIter            # of instructions:
126		10                      687
127		20                      438
128		30                      198
129		40                        0
130
131		In compiling the compiler, there is no significant difference in
132		compilation speed between maxIter=10 and maxIter=40. Actually,
133		my measurements showed that maxIter=40 was a tad faster than
134		maxIter=10! Also 96% of the  files in the compiler reach a fix
135		point within 13 iterations, so fixing maxIter at 40, while high,
136		is okay.
137
138		----------------------------------------------------------------------
139		Name: Matthias Blume
140		Date: 2001/10/31 15:25:00 EST
141		Tag: blume-20011031-pgraph
142		Description:
143
144		CKIT:
145		* Changed the "Function" constructor of type Ast.ctype to carry optional
146		argument identifiers.
147		* Changed the return type of TypeUtil.getFunction accordingly.
148		* Type equality ignores the argument names.
149		* TypeUtil.composite tries to preserve argument names but gives up quickly
150		if there is a mismatch.
151
152		installation script:
153		* attempts to use "curl" if available (unless "wget" is available as well)
154
155		CM:
156		* has an experimental implementation of "portable graphs" which I will
157		soon propose as an implementation-independent library format
158		* there are also new libraries $/pgraph.cm and $/pgraph-util.cm
159
160		NLFFI-LIB:
161		* some cleanup (all cosmetic)
162
163		NLFFIGEN:
164		* temporarily disabled the mechanism that suppresses ML output for
165		C definitions whose identifiers start with an underscore character
166		* generate val bindings for enum constants
167		* user can request that only one style (light or heavy) is being used;
168		default is to use both (command-line arguments: -heavy and -light)
169		* fixed bug in handling of function types involving incomplete pointers
170		* generate ML entry points that take record arguments (i.e., using
171		named arguments) for C functions that have a prototype with named
172		arguments
173		(see changes to CKIT)
174
175		----------------------------------------------------------------------
176		Name: Allen Leung
177		Date: 2001/10/27 20:34:00 EDT
178		Tag: leunga-20011027-x86-fast-fp-call
179		Description:
180
181		Fixed the bug described in blume-20010920-slowfp.
182
183		The fix involves
184		1. generating FCOPYs in FSTP in ia32-svid
185		2. marking a CALL with the appropriate annotation
186
187		----------------------------------------------------------------------
188		Name: Matthias Blume
189		Date: 2001/10/16 11:32:00 EDT
190		Tag: blume-20011016-netbsd
191		Description:
192
193		Underscore patch from Chris Richards (fixing problem with compiling
194		runtime system under recent NetBSD).
195
196		----------------------------------------------------------------------
197		Name: Allen Leung
198		Date: 2001/10/12 17:18:32 EDT 2001
199		Tag: leung-20011012-x86-printflowgraph
200		Description:
201
202		X86RA now uses a valid (instead of dummy) PrintFlowgraph module.
203
204		----------------------------------------------------------------------
205		Name: Lal George
206		Date: 2001/10/11 23:51:34 EDT
207		Tag: george-20011011-too-many-instrs
208		Description:
209
210		The representation of a program point never expected to see more
211		than 65536 instructions in a basic block!
212
213		----------------------------------------------------------------------
214		Name: Lal George
215		Date: 2001/10/09 09:41:37 EDT
216		Tag: george-20011008-mlrisc-labels
217		Description:
218
219		Changed the machine description files to support printing of
220		local and global labels in assembly code, based on host assembler
221		conventions.
222
223		----------------------------------------------------------------------
224		Name: Matthias Blume
225		Date: 2001/09/25 15:25:00 EDT
226		Tag: blume-20010925-exninfo
227		Description:
228
229		I provided a non-hook implementation of exnName (at the toplevel) and
230		made the "dummy" implementation of exnMessage (at the toplevel) more
231		useful: if nothing gets "hooked in", then at least you are going to
232		see the exception name and a message indicating why you don't see more.
233
234		[For the time being, programs that need exnMessage and want to use
235		ml-build should either use General.exnMessage (strongly recommended) or
236		refer to structure General at some other point so that CM sees a
237		static dependency.]
238
239		[Similar remarks go for "print" and "use":  If you want to use their
240		functionality in stand-alone programs generated by ml-build, then use
241		TextIO.output and Backend.Interact.useFile (from $smlnj/compiler.cm).]
242
243		----------------------------------------------------------------------
244		Name: Matthias Blume
245		Date: 2001/09/20 17:28:00 EDT
246		Tag: blume-20010920-slowfp
247		Description:
248
249		Allen says that x86-fast-fp is not safe yet, so I turned it off again...
250
251		----------------------------------------------------------------------
252		Name: Matthias Blume
253		Date: 2001/09/20 17:20:00 EDT
254		Tag: blume-20010920-canonicalpaths
255		Description:
256
257		0. Updated the BOOT file (something that I forgot to do earlier).
258
259		1. Small internal change to CM so that it avoids "/../" in filenames
260		as much as possible (but only where it is safe).
261
262		2. Changed config/_run-sml (resulting in a changed bin/.run-sml) so
263		that arguments that contain delimiters are passed through correctly.
264		This change also means that all "special" arguments of the form
265		@SMLxxx... must come first.
266
267		3. Changed install script to put relative anchor names for tool commands
268		into pathconfig.
269
270		----------------------------------------------------------------------
271		Name: Matthias Blume
272		>>>>>>> 1.169
273		Date: 2001/09/18 15:35:00 EDT
274		Tag: blume-20010918-readme11036
275		Description:
276
277		Added README files.
278
279		----------------------------------------------------------------------
280		Name: Matthias Blume
281		Date: 2001/09/18 11:45:00 EDT
282		Tag: Release_110_36 (retag)
283		Description:
284
285		Fixed mistake in config/preloads. Retagged as 110.36.
286
287		----------------------------------------------------------------------
288		Name: Matthias Blume
289		Date: 2001/09/18 09:40:00 EDT
290		Tag: Release_110_36_orig (tag changed)
291		Description:
292
293		New version (110.36).  New bootfiles.
294
295		----------------------------------------------------------------------
296		Name: Matthias Blume
297		Date: 2001/09/14 16:15:00 EDT
298		Tag: blume-20010914-x86fastfp
299		Description:
300
301		John committed some changes that Allen made, in particular a (hopefully)
302		correctly working version of the x86-fp module.
303
304		I changed the default setting of the Control.MLRISC.getFlag "x86-fast-fp"
305		flag to "true".  Everything seems to compile to a fixpoint ok, and
306		"mandelbrot" speeds up by about 15%.
307
308		----------------------------------------------------------------------
309		Name: Matthias Blume
310		Date: 2001/09/13 11:20:00 EDT
311		Tag: blume-20010913-minimal
312		Description:
313
314		1. Stefan Monnier's patch to fix a miscompilation problem that
315		was brought to light by John Reppy's work on Moby.
316
317		2. Implemented a minimal "structure Compiler" that contains just
318		"version" and "architecture".  The minimal version will be
319		available when the full version is not.  This is for backward-
320		compatibility with code that wants to test Compiler.version.
321
322		----------------------------------------------------------------------
323		Name: Matthias Blume
324		Date: 2001/08/28 14:03:00 EDT
325		Tag: blume-20010828-ml-lex
326		Description:
327
328		Fix for bug 1581, received from Neophytos Michael.
329
330		----------------------------------------------------------------------
331		Name: Matthias Blume
332		Date: 2001/08/27 11:20:00 EDT
333		Tag: blume-20010827-readme11035
334		Description:
335
336		Fleshed out the README file for 110.35.
337
338		----------------------------------------------------------------------
339		Name: Matthias Blume
340		Date: 2001/08/24 17:10:00 EDT
341		Tag: Release_110_35
342		Description:
343
344		New version number (110.35).  New bootfiles.
345
346		----------------------------------------------------------------------
347		Name: Lal George
348		Date: 2001/08/24 13:47:18 EDT 2001
349		Tag: george-20010824-MLRISC-graphs
350		Description:
351
352		removed clusters from MLRISC completely and replaced with graphs.
353
354		----------------------------------------------------------------------
355		Name: Matthias Blume
356		Date: 2001/08/23 17:50:00 EDT
357		Tag: blume-20010823-toplevel
358		Description:
359
360		- some reorganization of the code that implements various kinds of
361		environments in the compiler (static, dynamic, symbolic, combined)
362		- re-implemented the EnvRef module so that evalStream works properly
363		(if the stream contains references to "use", "CM.make", etc.)
364		- cleaned up evalloop.sml and interact.sml (but they need more cleaning)
365
366		----------------------------------------------------------------------
367		Name: Matthias Blume
368		Date: 2001/08/20 15:50 EDT
369		Tag: blume20010820-slipup
370		Description:
371
372		I forgot to commit a few files.  Here they are...
373
374		----------------------------------------------------------------------
375		Name: Matthias Blume
376		Date: 2001/08/20 15:35:00 EDT
377		Tag: blume-20010820-debugprof
378		Description:
379
380		!!!! NEW BOOTFILES !!!!
381
382		This is another round of reorganizing the compiler sources.  This
383		time the main goal was to factor out all the "instrumentation"
384		passes (for profiling and backtracing) into their own library.
385		The difficulty was to do it in such a way that it does not depend
386		on elaborate.cm but only on elabdata.cm.
387
388		Therefore there have been further changes to both elaborate.cm and
389		elabdata.cm -- more "generic" things have been moved from the former
390		to the latter.  As a result, I was forced to split the assignment
391		of numbers indicating "primtyc"s into two portions: SML-generic and
392		SML/NJ-specific.  Since it would have been awkward to maintain,
393		I bit the bullet and actually _changed_ the mapping between these
394		numbers and primtycs.  The bottom line of this is that you need
395		a new set of bin- and bootfiles.
396
397		I have built new bootfiles for all architectures, so doing a fresh
398		checkout and config/install.sh should be all you need.
399
400		The newly created library's name is
401
402		$smlnj/viscomp/debugprof.cm
403
404		and its sources live under
405
406		src/compiler/DebugProf
407
408		----------------------------------------------------------------------
409		Name: Matthias Blume
410		Date: 2001/08/15 17:15:00 EDT
411		Tag: blume-20010815-compreorg
412		Description:
413
414		This is a first cut at reorganizing the CM libraries that make up the
415		core of the compiler.  The idea is to separate out pieces that could
416		be used independently by tools, e.g., the parser, the typechecker, etc.
417
418		The current status is a step in this direction, but it is not quite
419		satisfactory yet.  Expect more changes in the future.
420
421		Here is the current (new) organization...
422
423		What used to be $smlnj/viscomp/core.cm is now divided into
424		six CM libraries:
425
426		$smlnj/viscomp/basics.cm
427		/parser.cm
428		/elabdata.cm
429		/elaborate.cm
430		/execute.cm
431		/core.cm
432
433		The CM files for these libraries live under src/system/smlnj/viscomp.
434		All these libraries are proxy libraries that contain precisely
435		one CM library component.  Here are the locations of the components
436		(all within the src/compiler tree):
437
438		Basics/basics.cm
439		Parse/parser.cm
440		ElabData/elabdata.cm
441		Elaborator/elaborate.cm
442		Execution/execute.cm
443		core.cm
444
445		[This organization is the same that has been used already
446		for a while for the architecture-specific parts of the visible
447		compiler and for the old version of core.cm.]
448
449		As you will notice, many source files have been moved from their
450		respective original locations to a new home in one of the above
451		subtrees.
452
453		The division of labor between the new libraries is the following:
454
455		basics.cm:
456		- Simple, basic definitions that pertain to many (or all) of
457		the other libraries.
458		parser.cm:
459		- The SML parser, producing output of type Ast.dec.
460		- The type family for Ast is also defined and exported here.
461		elabdata.cm:
462		- The datatypes that describe input and output of the elaborator.
463		This includes types, absyn, and static environments.
464		elaborator.cm:
465		- The SML/NJ type checker and elaborator.
466		This maps an Ast.dec (with a given static environment) to
467		an Absyn.dec (with a new static environment).
468		- This libraries implements certain modules that used to be
469		structures as functors (to remove dependencies on FLINT).
470		execute.cm:
471		- Everything having to do with executing binary code objects.
472		- Dynamic environments.
473		core.cm:
474		- SML/NJ-specific instantiations of the elaborator and MLRISC.
475		- Top-level modules.
476		- FLINT (this should eventually become its own library)
477
478		Notes:
479
480		I am not 100% happy with the way I separated the elaborator (and its
481		data structures) from FLINT.  Two instances of the same problem:
482
483		1. Data structures contain certain fields that carry FLINT-specific
484		information.  I hacked around this using exn and the property list
485		module from smlnj-lib.  But the fact that there are middle-end
486		specific fields around at all is a bit annoying.
487
488		2. The elaborator calculates certain FLINT-related information.  I tried
489		to make this as abstract as I could using functorization, but, again,
490		the fact that the elaborator has to perform calculations on behalf
491		of the middle-end at all is not nice.
492
493		3. Having to used exn and property lists is unfortunate because it
494		weakens type checking.  The other alternative (parameterizing
495		nearly *everything*) is not appealing, though.
496
497		I removed the "rebinding =" warning hack because due to the new organization
498		it was awkward to maintain it.  As a result, the compiler now issues some of
499		these warnings when compiling init.cmi during bootstrap compilation. On
500		the plus side, you also get a warning when you do, for example:
501		val op = = Int32.+
502		which was not the case up to now.
503
504		I placed "assign" and "deref" into the _Core structure so that the
505		code that deals with the "lazy" keyword can find them there.  This
506		removes the need for having access to the primitive environment
507		during elaboration.
508
509		----------------------------------------------------------------------
510		Name: Matthias Blume
511		Date: 2001/08/13
512		Tag: blume-20010813-closures
513		Description:
514
515		This fix was sent to us by Zhong Shao.  It is supposed to improve the
516		performance of certain loops by avoiding needless closure allocation.
517
518		----------------------------------------------------------------------
519		Name: Lal George
520		Date: 2001/07/31 10:03:23 EDT 2001
521		Tag: george-20010731-x86-fmalloc
522		Description: Fixed bug in x86 calls
523
524		There was a bug where call instructions would mysteriously
525		vanish. The call instruction had to be one that returned
526		a floating point value.
527
528		----------------------------------------------------------------------
529	Name: Lal George	Name: Lal George
530	Date: 2001/07/19 16:36:29 EDT 2001	Date: 2001/07/19 16:36:29 EDT 2001
531	Tag: george-20010719-simple-cells	Tag: george-20010719-simple-cells
#	Line 2733	Line 3246
3246	elaborator). There were a lot of changes during my "linkpath" trials	elaborator). There were a lot of changes during my "linkpath" trials
3247	that could have been reverted to their original state but weren't.	that could have been reverted to their original state but weren't.
3248	Please, don't be too harsh on me for messing with this code a bit more	Please, don't be too harsh on me for messing with this code a bit more
3249	than what was strictly necessary...  (I _did_ resist the tempation	than what was strictly necessary...  (I _did_ resist the temptation
3250	of doing any "global reformatting" to avoid an untimely death at	of doing any "global reformatting" to avoid an untimely death at
3251	Dave's hands. :)	Dave's hands. :)
3252

---

Colored Diff Long Colored Diff Unidiff Context Diff Side by Side

Legend: Removed from v.890   changed lines   Added in v.993

---

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