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 993, Fri Nov 23 02:37:10 2001 UTC revision 1059, Mon Feb 11 17:05:33 2002 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  Name: Matthias Blume
17    Date: 2002/02/11 12:05:00 EST
18    Tag: blume-20020211-gensml
19    Description:
20    
21    A quick fix for a problem with GenSML (in the pgraph-util library):
22    Make generation of toplevel "local" optional.  (Strictly speaking,
23    signature definitions within "local" are not legal SML.)
24    
25    Other than that: updates to INSTALL and cm/TODO.
26    
27    ----------------------------------------------------------------------
28    Name: Matthias Blume
29    Date: 2002/02/08 15:00:00 EST
30    Tag: blume-20020208-uniquepid
31    Description:
32    
33    0. Version number has been bumped to 110.38.1.  NEW BOOTFILES!!!
34    
35    1. The installer (config/install.sh) has gotten smarter:
36    
37         - Configuration options are a bit easier to specify now
38           (in config/targets).
39         - Bug in recognizing .tar.bz2 files fixed.
40         - Installer automatically resolves dependencies between
41           configuration options (e.g., if you ask for eXene, you will
42           also get cml -- regardless whether you asked for it or not).
43         - Installer can run in "quieter mode" by setting the environment
44           variable INSTALL_QUIETLY to "true".  "Quieter" does not mean
45           "completely silent", though.
46         - Build HashCons library as part of smlnj-lib.
47    
48    2. A new scheme for assigning persistent identifiers to compilation
49       units (and, by extension, to types etc.) has been put into place.
50       This fixes a long-standing bug where types and even dynamic values
51       can get internally confused, thereby compromising type safety
52       (abstraction) and dynamic correctness.  See
53    
54         http://cm.bell-labs.com/cm/cs/who/blume/pid-confusion.tgz
55    
56       for an example of how things could go wrong until now.
57    
58       The downside of the new scheme is that pids are not quite as
59       persistent as they used to be: CM will generate a fresh pid
60       for every compilation unit that it thinks it sees for the first
61       time.  That means that if you compile starting from a clean, fresh
62       source tree at two different times, you end up with different
63       binaries.
64    
65       Cutoff recompilation, however, has not been compromised because
66       CM keeps pid information in special caches between runs.
67    
68    ----------------------------------------------------------------------
69    Name: Lal George
70    Date: 2002/02/07 15:34:13 EST 2002
71    Tag: <none>
72    Description:
73    
74    Compilers that generate assembly code may produce  global labels
75    whose value is resolved at link time. The various peephole optimization
76    modules did not take this in account.
77    
78    TODO. The Labels.addrOf function should really return an option
79    type so that clients are forced to deal with this issue, rather
80    than an exception being raised.
81    
82    ----------------------------------------------------------------------
83    Name: Lal George
84    Date: 2002/02/06 13:55:02 EST
85    Tag: george-20020206-ra-breakup
86    Description:
87    
88    1. A bug fix from Allen.
89    
90        A typo causes extra fstp %st(0)'s to be generated at compensation
91        edges, which might cause stack underflow traps at runtime.  This
92        occurs in fft where there are extraneous fstps right before the 'into'
93        trap instruction (in this case they are harmless since none of the
94        integers overflow.)
95    
96    2. Pulled out various utility modules that were embedded in the modules
97       of the register allocator. I need these modules for other purposes, but
98       they are not complete enough to put into a library (just yet).
99    ----------------------------------------------------------------------
100    Name: Matthias Blume
101    Date: 2002/01/31 16:05:00 EST
102    Tag: blume-20020131-sparc-ccalls
103    Description:
104    
105    1. C-calls on Sparc needlessly allocated a huge chunk (96 bytes)
106       of extra stack space by mistake.  Fixed.
107    
108    2. Bug in logic of handling of command-line options in ml-nlffigen fixed.
109    
110    ----------------------------------------------------------------------
111    Name: Allen Leung
112    Date: 2002/01/30
113    Tag: leunga-20020130-nowhere-bug-fix
114    Description:
115    
116       MLRISC bug fixes:
117       1. Fixed a bindings computation bug in the 'nowhere' program generator tool.
118       2. MachineInt.fromString was negating its value.
119    
120    ----------------------------------------------------------------------
121    Name: Matthias Blume
122    Date: 2002/01/29
123    Tag: blume-20020129-INSTALL
124    Description:
125    
126    - Added somewhat detailed installation instructions (file INSTALL).
127    - Fixed curl-detection bug in config/install.sh.
128    - It is now possible to select the URL getter using the URLGETTER
129      environment variable:
130    
131          not set / "unknown"      --> automatic detection (script tries wget,
132                                       curl, and lynx)
133          "wget" / "curl" / "lynx" --> use the specified program (script "knows"
134                                       how to properly invoke them)
135          other                    --> use $URLGETTER directly, it must take
136                                       precisely two command-line arguments
137                                       (source URL and destination file name)
138    
139    ----------------------------------------------------------------------
140    Name: Matthias Blume
141    Date: 2002/01/28
142    Tag: blume-20020128-sparc-ccalls
143    Description:
144    
145    - Fixed problem with calculation of "used" registers in sparc-c-calls.
146    - Make use of the allocParam argument in sparc-c-calls.
147    
148    ----------------------------------------------------------------------
149    Name: Matthias Blume
150    Date: 2002/01/28
151    Tag: blume-20020128-allocParam
152    Description:
153    
154    John Reppy:  Changes c-calls API to accept client-callback for
155    allocating extra stack space.
156    me: Corresponding changes to mlriscGen (using a dummy argument that
157        does not change the current behavior).
158    
159    ----------------------------------------------------------------------
160    Name: Matthias Blume
161    Date: 2002/01/28 12:00:00
162    Tag: Release_110_38
163    Description:
164    
165    This time for real!!!
166    
167    ----------------------------------------------------------------------
168    Name: Matthias Blume
169    Date: 2002/01/28 10:56:00 EST
170    Tag: blume-20020128-retraction
171    Description:
172    
173    0. Retracted earlier 110.38.  (The Release_110_38 tag has been replaced
174       with blume-Release_110_38-retracted.)
175    
176    1. Fixed a problem with incorrect rounding modes in real64.sml.
177       (Thanks to Andrew Mccreight <andrew.mccreight@yale.edu>.)
178    
179    2. A bug in ml-nlffigen related to the handling of unnamed structs, unions,
180       and enums fixed.  The naming of corresponding ML identifiers should
181       now be consistent again.
182    
183    ----------------------------------------------------------------------
184    Name: Allen Leung
185    Date: 2002/01/27
186    Tag: leunga-20020127-nowhere
187    Description:
188    
189       Added a target called nowhere in the configuration scripts.
190       Enabling this will build the MLRISC 'nowhere' tool (for translating
191       programs with where-clauses into legal SML code) during installation.
192    
193    ----------------------------------------------------------------------
194    Name: Matthias Blume
195    Date: 2002/01/25 21:27:00 EST
196    Tag: blume-Release_110_38-retracted
197    Description:
198    
199    Call it a (working) release!  Version is 110.38. Bootfiles are ready.
200    
201    README will be added later.
202    
203    !!! NOTE:  Re-tagged as blume-Release_110_38-retracted. Original tag
204    (Release_110_38) removed.  Reason: Last-minute bug fixes.
205    
206    ----------------------------------------------------------------------
207    Name: Matthias Blume
208    Date: 2002/01/25
209    Tag: blume-20020125-ffi
210    Description:
211    
212    A large number of tweaks and improvements to ml-nlffi-lib and
213    ml-nlffigen:
214    
215       - ML represenation types have been streamlined
216       - getter and setter functions work with concrete values, not abstract
217         ones where possible
218       - ml-nlffigen command line more flexible (see README file there)
219       - some bugs have been fixed (hopefully)
220    
221    ----------------------------------------------------------------------
222    Name: Lal George
223    Date: 2002/01/24
224    Tag: george-20020124-risc-ra-interface
225    Description:
226    
227       There is a dramatic simplification in the interface to the
228       register allocator for RISC architectures as a result of making
229       parallel copy instructions explicit.
230    
231    ----------------------------------------------------------------------
232    Name: Matthias Blume
233    Date: 2002/01/22
234    Tag: blume-20020122-x86-ccalls
235    Description:
236    
237    Bug fix for c-calls on x86 (having to do with how char- and
238    short-arguments are being handled).
239    
240    ----------------------------------------------------------------------
241    Name: Matthias Blume
242    Date: 2002/01/21
243    Tag: blume-20020121-ff
244    Description:
245    
246    Another day of fiddling with the FFI...
247    
248    1. Bug fix/workaround:  CKIT does not complain about negative array
249       dimensions, so ml-nlffigen has to guard itself against this possibility.
250       (Otherwise a negative dimension would send it into an infinite loop.)
251    
252    2. Some of the abstract types (light objects, light pointers, most "base"
253       types) in structure C are now eqtypes.
254    
255    3. Added constructors and test functions for NULL function pointers.
256    
257    ----------------------------------------------------------------------
258    Name: Matthias Blume
259    Date: 2002/01/18
260    Tag: blume-20020118-ready-for-new-release
261    Description:
262    
263    Made config/srcarchiveurl point to a new place.  (Will provide boot
264    files shortly.)
265    
266    Maybe we christen this to be 110.38?
267    
268    ----------------------------------------------------------------------
269    Name: Matthias Blume
270    Date: 2002/01/18
271    Tag: blume-20020118-more-ffifiddle
272    Description:
273    
274    Today's FFI fiddling:
275    
276      - Provided a structure CGetSet with "convenient" versions of C.Get.* and
277        C.Set.* that use concrete (MLRep.*) arguments and results instead
278        of abstract ones.
279    
280      - Provided word-style bit operations etc. for "int" representation
281        types in MLRep.S<Foo>Bitops where <Foo> ranges over Char, Int, Short,
282        and Long.
283    
284    ----------------------------------------------------------------------
285    Name: Matthias Blume
286    Date: 2002/01/18
287    Tag: blume-20020118-use-x86-fp
288    Description:
289    
290    Now that x86-fast-fp seems to be working, I turned it back on again
291    by default.  (Seems to work fine now, even with the FFI.)
292    
293    Other than that, I added some documentation about the FFI to
294    src/ml-nlffigen/README and updated the FFI test examples in
295    src/ml-nlffi-lib/Tests/*.
296    
297    ----------------------------------------------------------------------
298    Name: Allen Leung
299    Date: 2002/01/17
300    Tag: leunga-20020117-x86-fast-fp-call
301    Description:
302    
303       1. Fixed a problem with handling return fp values when x86's fast fp
304          mode is turned on.
305    
306       2. Minor pretty printing fix for cellset.  Print %st(0) as %st(0) instead
307          of %f32.
308    
309       3. Added a constructor INT32lit to the ast of MLRISC tools.
310    
311    ----------------------------------------------------------------------
312    Name: Matthias Blume
313    Date: 2002/01/16
314    Tag: blume-20020116-ffifiddle
315    Description:
316    
317    More fiddling with the FFI interface:
318    
319     - Make constness 'c instead of rw wherever possible.  This eliminates
320       the need for certain explicit coercions.  (However, due to ML's
321       value polymorphism, there will still be many cases where explicit
322       coercions are necessary.  Phantom types are not the whole answer
323       to modeling a subtyping relationship in ML.)
324    
325     - ro/rw coersions for pointers added.  (Avoids the detour through */&.)
326    
327     - "printf" test example added to src/ml-nlffi-lib/Tests.  (Demonstrates
328       clumsy workaround for varargs problem.)
329    
330    ----------------------------------------------------------------------
331    Name: Lal George
332    Date: 2002/01/15
333    Tag: <none>
334    Description:
335    
336    1. Since COPY instructions are no longer native to the architecture,
337       a generic functor can be used to implement the expandCopies function.
338    
339    2. Allowed EXPORT and IMPORT pseudo-op declarations to appear inside a
340       TEXT segment.
341    
342    ----------------------------------------------------------------------
343    Name: Matthias Blume
344    Date: 2002/01/15
345    Tag: blume-20020115-ffiupdates
346    Description:
347    
348    1. Fix for bug resulting in single-precision float values being returned
349       incorrectly from FFI calls.
350    
351    2. Small modifications to C FFI API:
352    
353        - memory-allocation routines return straight objects (no options)
354          and raise an exception in out-of-memory situations
355        - unsafe extensions to cast between function pointers and pointers
356          from/to ints
357        - added structure C_Debug as an alternative to structure C where
358          pointer-dereferencing (|*| and |*!) always check for null-pointers
359        - added open_lib' to DynLinkage;  open_lib' works like open_lib
360          but also takes a (possibly empty) list of existing library handles
361          that the current library depends on
362    
363    ----------------------------------------------------------------------
364    Name: Matthias Blume
365    Date: 2002/01/10
366    Tag: blume-20020110-newffigen
367    Description:
368    
369    1. Updates to portable graph code.
370    
371    2. Major update to ml-nlffigen and ml-nlffi-lib.  Things are much
372       more scalable now so that even huge interfaces such as the one
373       for GTK compile in finite time and space. :-)
374       See src/ml-nlffigen/README for details on what's new.
375    
376    ----------------------------------------------------------------------
377    Name: Lal George
378    Date: 2001/01/09 14:31:35 EST 2002
379    Tag: george-20011206-rm-native-copy
380    Description:
381    
382            Removed the native COPY and FCOPY instructions
383            from all the architectures and replaced it with the
384            explicit COPY instruction from the previous commit.
385    
386            It is now possible to simplify many of the optimizations
387            modules that manipulate copies. This has not been
388            done in this change.
389    
390    ----------------------------------------------------------------------
391    Name: Lal George
392    Date: 2001/12/06 16:50:13 EST 2001
393    Tag: george-20011206-mlrisc-instruction
394    Description:
395    
396    Changed the representation of instructions from being fully abstract
397    to being partially concrete. That is to say:
398    
399      from
400            type instruction
401    
402      to
403            type instr                              (* machine instruction *)
404    
405            datatype instruction =
406                LIVE of {regs: C.cellset, spilled: C.cellset}
407              | KILL of {regs: C.cellset, spilled: C.cellset}
408              | COPYXXX of {k: CB.cellkind, dst: CB.cell list, src: CB.cell list}
409              | ANNOTATION of {i: instruction, a: Annotations.annotation}
410              | INSTR of instr
411    
412    This makes the handling of certain special instructions that appear on
413    all architectures easier and uniform.
414    
415    LIVE and KILL say that a list of registers are live or killed at the
416    program point where they appear. No spill code is generated when an
417    element of the 'regs' field is spilled, but the register is moved to
418    the 'spilled' (which is present, more for debugging than anything else).
419    
420    LIVE replaces the (now deprecated) DEFFREG instruction on the alpha.
421    We used to generate:
422    
423            DEFFREG f1
424            f1 := f2 + f3
425            trapb
426    
427    but now generate:
428    
429            f1 := f2 + f3
430            trapb
431            LIVE {regs=[f1,f2,f3], spilled=[]}
432    
433    Furthermore, the DEFFREG (hack) required that all floating point instruction
434    use all registers mentioned in the instruction. Therefore f1 := f2 + f3,
435    defines f1 and uses [f1,f2,f3]! This hack is no longer required resulting
436    in a cleaner alpha implementation. (Hopefully, intel will not get rid of
437    this architecture).
438    
439    COPYXXX is intended to replace the parallel COPY and FCOPY  available on
440    all the architectures. This will result in further simplification of the
441    register allocator that must be aware of them for coalescing purposes, and
442    will also simplify certain aspects of the machine description that provides
443    callbacks related to parallel copies.
444    
445    ANNOTATION should be obvious, and now INSTR represents the honest to God
446    machine instruction set!
447    
448    The <arch>/instructions/<arch>Instr.sml files define certain utility
449    functions for making porting easier -- essentially converting upper case
450    to lower case. All machine instructions (of type instr) are in upper case,
451    and the lower case form generates an MLRISC instruction. For example on
452    the alpha we have:
453    
454      datatype instr =
455         LDA of {r:cell, b:cell, d:operand}
456       | ...
457    
458      val lda : {r:cell, b:cell, d:operand} -> instruction
459        ...
460    
461    where lda is just (INSTR o LDA), etc.
462    
463    ----------------------------------------------------------------------
464    Name: Matthias Blume
465  Date: 2001/11/22 21:40:00 EST  Date: 2001/11/22 21:40:00 EST
466  Tag: Release_110_37  Tag: Release_110_37
467  Description:  Description:
# Line 269  Line 717 
717    
718  ----------------------------------------------------------------------  ----------------------------------------------------------------------
719  Name: Matthias Blume  Name: Matthias Blume
 >>>>>>> 1.169  
720  Date: 2001/09/18 15:35:00 EDT  Date: 2001/09/18 15:35:00 EDT
721  Tag: blume-20010918-readme11036  Tag: blume-20010918-readme11036
722  Description:  Description:

Legend:
Removed from v.993  
changed lines
  Added in v.1059

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