Diff of /sml/trunk/HISTORY

Parent Directory | Revision Log | Patch

	revision 673, Wed Jun 21 08:57:07 2000 UTC	revision 683, Wed Jul 5 00:42:19 2000 UTC
#	Line 12	Line 12
12	Tag: <post-commit CVS tag>	Tag: <post-commit CVS tag>
13	Description:	Description:
14	----------------------------------------------------------------------	----------------------------------------------------------------------
15		Name: Allen Leung
16		Date: 2000/06/15 00:38:00
17		Tag: leunga-20000704-sparc-x86
18
19		1. x86 peephole improvement sp += k; sp -= k => nop  [from John]
20		2. fix to x86 RET bug [found by Dan Grossman]
21		3. sparc assembly bug fix for ticc instructions [found by Fermin]
22
23		Affects c-- and moby only
24
25		----------------------------------------------------------------------
26		Name: Matthias Blume
27		Date: 2000/07/04 15:26:00
28		Tag: blume-20000704-trigger
29		Description:
30
31		1. Improvements to CM manual.
32		2. SMLofNJ.Internals.BTrace.trigger reinstated as an alternative way
33		of getting a back-trace.  The function, when called, raises an
34		internal exception which explicitly carries the full back-trace history,
35		so it is unaffected by any intervening handle-raise pairs ("trivial"
36		or not).  The interactive loop will print that history once it arrives
37		at top level.
38		Short of having all exceptions implicitly carry the full history, the
39		recommended way of using this facility is:
40		- compile your program with instrumentation "on"
41		- run it, when it raises an exception, look at the history
42		- if the history is "cut off" because of some handler, go and modify
43		your program so that it explicitly calls BTrace.trigger
44		- recompile (still instrumented), and rerun; look at the full history
45
46		----------------------------------------------------------------------
47		Name: Matthias Blume
48		Date: 2000/07/03 15:36:00 JST
49		Tag: blume-20000702-manual
50		Description:
51
52		Small corrections and updates to CM manual.
53
54		----------------------------------------------------------------------
55		Name: Matthias Blume
56		Date: 2000/06/29 16:04:00 JST
57		Tag: blume-20000629-yacctool
58		Description:
59
60		Changes:
61
62		1. Class "mlyacc" now takes separate arguments to pass options to
63		generated .sml- and .sig-files independently.
64		2. Corresponding CM manual updates.
65		3. BTrace module now also reports call sites.  (However, for loop clusters
66		it only shows from where the cluster was entered.)  There are associated
67		modifications to core.sml, internals.{sig,sml}, btrace.sml, and btimp.sml.
68
69		----------------------------------------------------------------------
70		Name: Matthias Blume
71		Date: 2000/06/27 16:51:00 JST
72		Tag: blume-20000627-noweb
73		Description:
74
75		Changes:
76
77		1. Implemented "subdir" and "witness" options for noweb tool.
78		This caused some slight internal changes in CM's tool implementation.
79		2. Fixed bug in "tool plugin" mechanism.  This is essentially cleaning
80		some remaining issues from earlier path anchor changes.
81		3. Updated CM manual accordingly.
82
83		4. Changed implementation of back-tracing so that I now consider it
84		ready for prime-time.
85
86		In particular, you don't have to explicitly trigger the back-trace
87		anymore.  Instead, if you are running BTrace-instrumented code and
88		there is an uncaught exception (regardless of whether or not it was
89		raised in instrumented code), the top-level evalloop will print
90		the back-trace.
91
92		Features:
93
94		- Instrumented and uninstrumented code work together seemlessly.
95		(Of course, uninstrumented code is never mentioned in actual
96		back-traces.)
97
98		- Asymptotic time- and space-complexity of instrumented code is
99		equal to that of uninstrumented code.  (This means that
100		tail-recursion is preserved by the instrumentation phase.)
101
102		- Modules whose code has been instrumented in different sessions
103		work together without problem.
104
105		- There is no penalty whatsoever on uninstrumented code.
106
107		- There is no penalty on "raise" expressions, even in
108		instrumented code.
109
110		A potential bug (or perhaps it is a feature, too):
111
112		A back-trace reaches no further than the outermost instrumented
113		non-trivial "raise".  Here, a "trivial" raise is one that is the
114		sole RHS of a "handle" rule.  Thus, back-traces reach trough
115
116		<exp> handle e => raise e
117
118		and even
119
120		<exp> handle Foo => raise Bar
121
122		and, of course, through
123
124		<exp> handle Foo => ...
125
126		if the exception was not Foo.
127
128		Back-traces always reach right through any un-instrumented code
129		including any of its "handle" expressions, trivial or not.
130
131		To try this out, do the following:
132
133		- Erase all existing binfiles for your program.
134		(You may keep binfiles for those modules where you think you
135		definitely don't need back-tracing.)
136		- Turn on back-trace instrumentation:
137		SMLofNJ.Internals.BTrace.mode (SOME true);
138		- Recompile your program.  (I.e., run "CM.make" or "use".)
139		- You may now turn instrumentation off again (if you want):
140		SMLofNJ.Internals.BTrace.mode (SOME false);
141		- Run your program as usual.  If it raises an exception that
142		reaches the interactive toplevel, then a back-trace will
143		automatically be printed.  After that, the toplevel loop
144		will print the exception history as usual.
145
146		----------------------------------------------------------------------
147		Name: Matthias Blume
148		Date: 2000/06/26 09:56:46 JST
149		Tag: blume-20000626-setup
150		Description:
151
152		CM: - setup-parameter to "sml" added; this can be used to run arbitrary
153		ML code before and after compiling a file (e.g., to set compiler
154		flags)
155
156		Compiler: - improved btrace API (in core.sml, internals.{sig,sml})
157		- associated changes to btrace.sml (BTrace instrumentation pass)
158		- cleaner implementation of btimp.sml (BTrace tracing and report
159		module)
160
161		CM manual: * new path encoding documented
162		* description of setup-parameter to "sml" added
163
164		The biggest user-visible change to back-tracing is that it is no
165		longer necessary to compile all traced modules within the same
166		session.  (This was a real limitation.)
167
168		----------------------------------------------------------------------
169		Name: Matthias Blume
170		Date: 2000/06/24 12:40:00 JST
171		Tag: blume-20000624-startup
172		Description:
173
174		Fixes startup slowdown problem.  (I was calling SrcPath.sync a _tad_
175		bit too often -- to put it mildly. :)
176
177		----------------------------------------------------------------------
178		Name: Matthias Blume
179		Date: 2000/06/23 18:20:00 JST
180		Tag: blume-20000623-btrace
181		Description:
182
183		This updates adds a backtrace facility to aid programmers in debugging
184		their programs.  This involves the following changes:
185
186		1. Module system/smlnj/init/core.sml (structure _Core) now has hooks for
187		keeping track of the current call stack.  When programs are compiled
188		in a special mode, the compiler will insert calls to these hooks
189		into the user program.
190		"Hook" means that it is possible for different implementations of
191		back-tracing to register themselves (at different times).
192
193		2. compiler/MiscUtil/profile/btrace.sml implements the annotation phase
194		as an Absyn.dec->Absyn.dec rewrite.  Normally this phase is turned off.
195		It can be turned on using this call:
196		SMLofNJ.Internals.BTrace.mode (SOME true);
197		Turning it off again:
198		SMLofNJ.Internals.BTrace.mode (SOME false);
199		Querying the current status:
200		SMLofNJ.Internals.BTrace.mode NONE;
201		Annotated programs are about twice as big as normal ones, and they
202		run a factor of 2 to 4 slower with a dummy back-trace plugin (one
203		where all hooks do nothing).  The slowdown with a plugin that is
204		actually useful (such as the one supplied by default) is even greater,
205		but in the case of the default plugin it is still only an constant
206		factor (amortized).
207
208		3. system/Basis/Implementation/NJ/internals.{sig,sml} have been augmented
209		with a sub-structure BTrace for controlling back-tracing.  In particular,
210		the above-mentioned function "mode" controls whether the annotation
211		phase is invoked by the compiler.  Another important function is
212		"trigger": when called it aborts the current execution and causes
213		the top-level loop to print a full back-trace.
214
215		4. compiler/MiscUtil/profile/btimp.sml is the current default plugin
216		for back-tracing.  It keeps track of the dynamic call stack and in
217		addition to that it keeps a partial history at each "level" of that
218		stack.  For example, if a tail-calls b, b tail-calls c, and c tail-calls
219		d and b (at separate times, dynamically), then the report will show:
220
221		GOTO   d
222		/c
223		GOTO  \b
224		CALL   a
225
226		This shows that there was an initial non-tail call of a, then a
227		tail-call to b or c, looping behavior in a cluster of functions that
228		consist of b and c, and then a goto from that cluster (i.e., either from
229		b or from c) to d.
230
231		Note that (depending on the user program) the amount of information
232		that the back-trace module has to keep track of at each level is bounded
233		by a constant.  Thus, the whole implementation has the same asymptotical
234		complexity as the original program (both in space and in time).
235
236		5. compiler/TopLevel/interact/evalloop.sml has been modified to
237		handle the special exception SMLofNJ.Internals.BTrace.BTrace
238		which is raised by the "trigger" function mentioned above.
239
240		Notes on usage:
241
242		- Annotated code works well together with unannotated code:
243		Unannotated calls simply do not show up at all in the backtrace.
244
245		- It is not a good idea to let modules that were annotated during
246		different sessions run at the same time.  This is because the compiler
247		chooses small integers to identify individual functions, and there
248		will be clashes if different modules were compiled in separate sessions.
249		(Nothing will crash, and you will even be told about the clashes, but
250		back-trace information will in general not be useful.)
251
252		- Back-tracing can be confused by callcc and capture.
253
254		- The only way of getting a back-trace right now is to explicitly
255		invoke the "trigger" function from your user program.  Eventually, we
256		should make every exception carry back-trace information (if
257		available).  But since this creates more overhead at "raise"-time
258		(similar to the current exnHistory overhead), I have not yet
259		implemented this.  (The implementation will be rather easy.)  With
260		exceptions carrying back-trace information, this facility will be even
261		more useful because users don't need to modify their programs...
262
263		- While it is possible to compile the compiler with back-trace
264		annotations turned on (I did it to get some confidence in
265		correctness), you must make absolutely sure that core.sml and
266		btimp.sml are compiled WITHOUT annotation!  (core.sml cannot actually
267		be compiled with annotation because there is no core access yet, but
268		if you compile btimp.sml with annotation, then the system will go into
269		an infinite recursion and crash.)
270		Since CM currently does not know about BTrace, the only way to turn
271		annotations on and off for different modules of the compiler is to
272		interrupt CMB.make, change the settings, and re-invoke it.  Of course,
273		this is awkward and clumsy.
274
275		Sample sessions:
276
277		Standard ML of New Jersey v110.28.1 [FLINT v1.5], June 5, 2000
278		- SMLofNJ.Internals.BTrace.mode (SOME true);
279		[autoloading]
280		[autoloading done]
281		val it = false : bool
282		- structure X = struct
283		-     fun main n = let
284		-         fun a (x, 0) = d x
285		-           | a (x, n) = b (x, n - 1)
286		-         and b (x, n) = c (x, n)
287		-         and c (x, n) = a (x, n)
288		-         and d x = e (x, 3)
289		-         and e (x, 0) = f x
290		-           | e (x, n) = e (x, n - 1)
291		-         and f 0 = SMLofNJ.Internals.BTrace.trigger ()
292		-           | f n = n * g (n - 1)
293		-         and g n = a (n, 3)
294		-     in
295		-         f n
296		-     end
297		- end;
298		structure X : sig val main : int -> int end
299		- X.main 3;
300		*** BACK-TRACE ***
301		GOTO   stdIn:4.2-13.20: X.main[2].f
302		GOTO-( stdIn:4.2-13.20: X.main[2].e
303		GOTO   stdIn:4.2-13.20: X.main[2].d
304		/ stdIn:4.2-13.20: X.main[2].a
305		| stdIn:4.2-13.20: X.main[2].b
306		GOTO-\ stdIn:4.2-13.20: X.main[2].c
307		CALL   stdIn:4.2-13.20: X.main[2].g
308		GOTO   stdIn:4.2-13.20: X.main[2].f
309		GOTO-( stdIn:4.2-13.20: X.main[2].e
310		GOTO   stdIn:4.2-13.20: X.main[2].d
311		/ stdIn:4.2-13.20: X.main[2].a
312		| stdIn:4.2-13.20: X.main[2].b
313		GOTO-\ stdIn:4.2-13.20: X.main[2].c
314		CALL   stdIn:4.2-13.20: X.main[2].g
315		GOTO   stdIn:4.2-13.20: X.main[2].f
316		GOTO-( stdIn:4.2-13.20: X.main[2].e
317		GOTO   stdIn:4.2-13.20: X.main[2].d
318		/ stdIn:4.2-13.20: X.main[2].a
319		| stdIn:4.2-13.20: X.main[2].b
320		GOTO-\ stdIn:4.2-13.20: X.main[2].c
321		CALL   stdIn:4.2-13.20: X.main[2].g
322		GOTO   stdIn:4.2-13.20: X.main[2].f
323		CALL   stdIn:2.15-17.4: X.main[2]
324		-
325
326		(Note that because of a FLINt bug the above code currently does not
327		compile without BTrace turned on.)
328
329		Here is another example, using my modified Tiger compiler:
330
331		Standard ML of New Jersey v110.28.1 [FLINT v1.5], June 5, 2000
332		- SMLofNJ.Internals.BTrace.mode (SOME true);
333		[autoloading]
334		[autoloading done]
335		val it = false : bool
336		- CM.make "sources.cm";
337		[autoloading]
338		...
339		[autoloading done]
340		[scanning sources.cm]
341		[parsing (sources.cm):parse.sml]
342		[creating directory CM/SKEL ...]
343		[parsing (sources.cm):tiger.lex.sml]
344		...
345		[wrote CM/sparc-unix/semant.sml]
346		[compiling (sources.cm):main.sml]
347		[wrote CM/sparc-unix/main.sml]
348		[New bindings added.]
349		val it = true : bool
350		- Main.compile ("../testcases/merge.tig", "foo.out");
351		*** BACK-TRACE ***
352		CALL   lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trvar
353		CALL   lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
354		CALL   lib/semant.sml:289.3-295.22: SemantFun[2].transExp.trexp.check[2]
355		GOTO   lib/semant.sml:289.3-295.22: SemantFun[2].transExp.trexp.check[2]
356		CALL   lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
357		CALL   lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
358		CALL   lib/semant.sml:488.3-505.6: SemantFun[2].transDec.trdec[2].transBody[2]
359		/ lib/semant.sml:411.65-543.8: SemantFun[2].transDec
360		CALL-\ lib/semant.sml:413.2-540.9: SemantFun[2].transDec.trdec[2]
361		CALL   lib/semant.sml:99.2-396.21: SemantFun[2].transExp.trexp
362		CALL   lib/semant.sml:8.52-558.4: SemantFun[2].transProg[2]
363		CALL   main.sml:1.18-118.4: Main.compile[2]
364		-
365
366		----------------------------------------------------------------------
367	Name: Matthias Blumen	Name: Matthias Blumen
368	Date: 2000/06/21 18:00:00 JST	Date: 2000/06/21 18:00:00 JST
369	Tag: blume-20000621-manual	Tag: blume-20000621-manual

---

Colored Diff Long Colored Diff Unidiff Context Diff Side by Side

Legend: Removed from v.673   changed lines   Added in v.683

---

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