]>
matita.cs.unibo.it Git - helm.git/log
Enrico Tassi [Tue, 9 May 2006 13:40:24 +0000 (13:40 +0000)]
types2006 patch
Andrea Asperti [Fri, 5 May 2006 14:56:30 +0000 (14:56 +0000)]
New version of deep_subsumption
Enrico Tassi [Thu, 4 May 2006 15:15:47 +0000 (15:15 +0000)]
goal demodulated with new
Enrico Tassi [Thu, 4 May 2006 10:55:15 +0000 (10:55 +0000)]
new pp function for proofs
Enrico Tassi [Wed, 3 May 2006 11:26:21 +0000 (11:26 +0000)]
eq_chain
Enrico Tassi [Wed, 3 May 2006 11:25:45 +0000 (11:25 +0000)]
more transitivity on proofs
Andrea Asperti [Thu, 27 Apr 2006 10:07:47 +0000 (10:07 +0000)]
Equality chains.
-This line, and those below, will be ignored--
M acic_content/acic2content.ml
Andrea Asperti [Thu, 27 Apr 2006 10:06:54 +0000 (10:06 +0000)]
Added is_trans_eq_URI and is_sym_eq_URI
Andrea Asperti [Wed, 26 Apr 2006 14:53:03 +0000 (14:53 +0000)]
Build_proof_goal does not return the metasenv any more.
Andrea Asperti [Wed, 26 Apr 2006 13:23:55 +0000 (13:23 +0000)]
fixed demodulation_goal (used to return always false)
Andrea Asperti [Wed, 26 Apr 2006 13:23:29 +0000 (13:23 +0000)]
removed ocaml equality on equations
Enrico Tassi [Wed, 26 Apr 2006 10:05:28 +0000 (10:05 +0000)]
more cleanup
Enrico Tassi [Wed, 26 Apr 2006 09:59:07 +0000 (09:59 +0000)]
cleanup of saturate
Enrico Tassi [Wed, 26 Apr 2006 09:21:07 +0000 (09:21 +0000)]
added a new type for proofs.
Enrico Tassi [Wed, 26 Apr 2006 09:16:01 +0000 (09:16 +0000)]
added a whd (nodelta) in the carr function used by the refiner (in the eat_prods case)
to get the carrier of a coercion. this is useful in case of terms generated by paramodulation
(and we want to refine the proof letting the refiner add symmetry steps)
where we have ((\lambda x:A.P(x)) t) === P'(t)
and P/P' is M=N/N=M
Claudio Sacerdoti Coen [Sun, 16 Apr 2006 14:45:11 +0000 (14:45 +0000)]
Bug fixed: "paste as pattern" now pastes the full pattern (e.g.
"in H:(? %)"; it used to paste "(? %)" only)
Claudio Sacerdoti Coen [Sat, 15 Apr 2006 16:57:24 +0000 (16:57 +0000)]
Patch to avoid double execution of whelp queries reverted (since it seems
that queries are NOT executed twice without the patch (!?!) and with the
patch the query is sometimes not executed).
Enrico Tassi [Fri, 14 Apr 2006 11:36:27 +0000 (11:36 +0000)]
uft8 string length bug fixed (Ctr-Alt-. did not work properly any longer)
Enrico Tassi [Fri, 14 Apr 2006 10:15:48 +0000 (10:15 +0000)]
multi-instances aliases are compressed to single instances if possible
Enrico Tassi [Fri, 14 Apr 2006 10:14:43 +0000 (10:14 +0000)]
duplicate check for coercions when added to Db
Enrico Tassi [Fri, 14 Apr 2006 10:13:55 +0000 (10:13 +0000)]
alases instance not printed if 0
Enrico Tassi [Fri, 14 Apr 2006 10:13:07 +0000 (10:13 +0000)]
the same utf8 bug as before
Enrico Tassi [Fri, 14 Apr 2006 10:12:27 +0000 (10:12 +0000)]
fixed another utf8 string length bug
Enrico Tassi [Fri, 14 Apr 2006 10:11:32 +0000 (10:11 +0000)]
added minimal euristic for generic terms carrier comparison
Enrico Tassi [Thu, 13 Apr 2006 23:44:51 +0000 (23:44 +0000)]
to tired to write a message.
Enrico Tassi [Thu, 13 Apr 2006 14:24:03 +0000 (14:24 +0000)]
added include' to include everything but preferences (aka aliases)
Enrico Tassi [Thu, 13 Apr 2006 14:23:30 +0000 (14:23 +0000)]
partially fixed boxes in rewite
Enrico Tassi [Thu, 13 Apr 2006 14:21:12 +0000 (14:21 +0000)]
added keyword include'
Enrico Tassi [Thu, 13 Apr 2006 14:20:25 +0000 (14:20 +0000)]
-debug should work better
Enrico Tassi [Thu, 13 Apr 2006 13:06:25 +0000 (13:06 +0000)]
in eta_finxing: type_of_aux' not called on eta_fixed terms
in doubletypeInference calls to type_of_aux' wrapped because of eta fixing
Claudio Sacerdoti Coen [Thu, 13 Apr 2006 11:17:10 +0000 (11:17 +0000)]
Ugly solution to the "we proved T that is equivalent to T" problem:
we use pack_coercion (defined into cicRefine) inside doubleTypeInference
(that is well below in the module hierarchy :-(
This is also supposed to be quite expensive (since pack_coercion uses
the refiner and coercion packing is applied to every expected/synthesized
type).
Claudio Sacerdoti Coen [Thu, 13 Apr 2006 11:10:27 +0000 (11:10 +0000)]
Profiling code commented out.
Claudio Sacerdoti Coen [Thu, 13 Apr 2006 11:08:19 +0000 (11:08 +0000)]
Some times reduced in a second benchmark.
Claudio Sacerdoti Coen [Thu, 13 Apr 2006 11:06:29 +0000 (11:06 +0000)]
Utime + Systime used in place of gettimeofday.
Enrico Tassi [Wed, 12 Apr 2006 21:18:44 +0000 (21:18 +0000)]
added patch to allow agin "match sin ? = ?"
Enrico Tassi [Wed, 12 Apr 2006 21:17:18 +0000 (21:17 +0000)]
fixed new compilation order
Enrico Tassi [Wed, 12 Apr 2006 21:15:27 +0000 (21:15 +0000)]
exported pp function for terms
Enrico Tassi [Wed, 12 Apr 2006 16:22:09 +0000 (16:22 +0000)]
whelp locate now accepts * and ?
Enrico Tassi [Wed, 12 Apr 2006 15:50:57 +0000 (15:50 +0000)]
whelp macros have now () around args
Enrico Tassi [Wed, 12 Apr 2006 15:48:08 +0000 (15:48 +0000)]
some fixes for whelp macros (concerning pprint...)
Claudio Sacerdoti Coen [Tue, 11 Apr 2006 14:20:11 +0000 (14:20 +0000)]
Altri benchmarks.
Claudio Sacerdoti Coen [Tue, 11 Apr 2006 10:02:32 +0000 (10:02 +0000)]
CicEnvironment is emptied when a size treshold is reached.
Andrea Asperti [Mon, 10 Apr 2006 09:02:37 +0000 (09:02 +0000)]
Removed negative equations.
Enrico Tassi [Wed, 5 Apr 2006 15:00:49 +0000 (15:00 +0000)]
links to .opt are now generated in the world target
Enrico Tassi [Wed, 5 Apr 2006 14:47:33 +0000 (14:47 +0000)]
fix for distro
Enrico Tassi [Wed, 5 Apr 2006 13:24:35 +0000 (13:24 +0000)]
a bit of shareing
Enrico Tassi [Wed, 5 Apr 2006 13:23:39 +0000 (13:23 +0000)]
added estimate_size
Enrico Tassi [Wed, 5 Apr 2006 12:10:15 +0000 (12:10 +0000)]
create directory paramodulation for tests for paramodulation
Enrico Tassi [Wed, 5 Apr 2006 12:08:32 +0000 (12:08 +0000)]
added another interesting problem for paramodulation
Enrico Tassi [Wed, 5 Apr 2006 12:07:51 +0000 (12:07 +0000)]
the tactic now returns as open goals the open metas in the proof
subsumption is now used correctly in given_clause_fullred
Enrico Tassi [Wed, 5 Apr 2006 08:06:55 +0000 (08:06 +0000)]
subsumption fixed and called in given_clause_fullred.
Andrea Asperti [Tue, 4 Apr 2006 15:01:30 +0000 (15:01 +0000)]
Naif substitution. Removed local context in metas during relocation.
Removed apply_theorems in saturate, replaced by a local check for
identity.
Claudio Sacerdoti Coen [Mon, 3 Apr 2006 13:12:21 +0000 (13:12 +0000)]
Useless code simplified out.
Claudio Sacerdoti Coen [Fri, 31 Mar 2006 11:02:29 +0000 (11:02 +0000)]
New benchmark after removal of some profiling code.
Claudio Sacerdoti Coen [Thu, 30 Mar 2006 16:40:50 +0000 (16:40 +0000)]
Less profiling.
Claudio Sacerdoti Coen [Thu, 30 Mar 2006 16:21:45 +0000 (16:21 +0000)]
Profiling code for merge_ugraphs commented out (since profiling is much more
expensive than the profiled code).
Claudio Sacerdoti Coen [Thu, 30 Mar 2006 14:28:28 +0000 (14:28 +0000)]
A few benchmarks on the library of Coq committed.
Claudio Sacerdoti Coen [Thu, 30 Mar 2006 11:40:40 +0000 (11:40 +0000)]
Added deadline (now 30s) to each test.
Claudio Sacerdoti Coen [Thu, 30 Mar 2006 11:15:52 +0000 (11:15 +0000)]
Sys.Break used to be captured.
Claudio Sacerdoti Coen [Thu, 30 Mar 2006 11:12:53 +0000 (11:12 +0000)]
Bug fixed: terms with a Cast used to raise assert false when whd was avoided
by the conversion strategy.
Claudio Sacerdoti Coen [Wed, 29 Mar 2006 16:26:32 +0000 (16:26 +0000)]
Huge speed-up in conversion: the old conversion strategy is set back.
1. try to "convert" the two terms recursively without performing reduction
2. if it fails, do a whd and try again
The overall result on the library of Coq is really remarkable.
Claudio Sacerdoti Coen [Wed, 29 Mar 2006 14:44:00 +0000 (14:44 +0000)]
#### EXPERIMENTAL COMMIT ####
Added a potentially dangerous ~expand_beta_redexes flag to
CicSubstitution.subst. If used carefully, it enormously speeds up
the type-checking of many theorems (since a very frequent conversion test
happens between a beta-redex and its contractum).
Claudio Sacerdoti Coen [Wed, 29 Mar 2006 13:34:55 +0000 (13:34 +0000)]
Debugging code added.
Enrico Tassi [Wed, 29 Mar 2006 12:18:56 +0000 (12:18 +0000)]
removed unif_ty ref
Enrico Tassi [Wed, 29 Mar 2006 12:13:16 +0000 (12:13 +0000)]
reverted the addition of _ to mistyped names
Enrico Tassi [Wed, 29 Mar 2006 12:07:55 +0000 (12:07 +0000)]
few bits to debug the benchmark system
Enrico Tassi [Tue, 28 Mar 2006 08:41:01 +0000 (08:41 +0000)]
deals with colors
Enrico Tassi [Tue, 28 Mar 2006 08:22:51 +0000 (08:22 +0000)]
more profiling and fixes for paramod
Andrea Asperti [Mon, 27 Mar 2006 15:02:32 +0000 (15:02 +0000)]
args removed from equalities.
New selection strategy. De morgan: 253 s.
Claudio Sacerdoti Coen [Mon, 27 Mar 2006 14:14:52 +0000 (14:14 +0000)]
Debugging code commented out.
Claudio Sacerdoti Coen [Mon, 27 Mar 2006 14:10:47 +0000 (14:10 +0000)]
* trust = true
* better pretty-printing of type-checking exceptions
Claudio Sacerdoti Coen [Mon, 27 Mar 2006 14:01:26 +0000 (14:01 +0000)]
"Performance bug" fixed: I removed a whd in the does_not_occur function.
Consequences:
1. incredible speed-up in the type-checking of certain inductive types
2. "dummy" occurrences (i.e. occurrences that are doomed to disappear because
of reduction) are now rejected (they used to be accepted). Coq has the
same behaviour.
Claudio Sacerdoti Coen [Mon, 27 Mar 2006 12:21:48 +0000 (12:21 +0000)]
Several "try ... with _ -> " specialized.
Claudio Sacerdoti Coen [Mon, 27 Mar 2006 11:28:06 +0000 (11:28 +0000)]
More robust handling of Control-C.
Claudio Sacerdoti Coen [Mon, 27 Mar 2006 08:33:25 +0000 (08:33 +0000)]
Flush stdout added in proper position.
Stefano Zacchiroli [Fri, 24 Mar 2006 20:45:13 +0000 (20:45 +0000)]
ignoring test_library{,.opt}
Claudio Sacerdoti Coen [Fri, 24 Mar 2006 18:33:30 +0000 (18:33 +0000)]
Recently introduced bug fixed in the kernel: a stack was "forgot" during
reduction of a MutCase.
Claudio Sacerdoti Coen [Fri, 24 Mar 2006 18:02:18 +0000 (18:02 +0000)]
Unable to parse my own output. Fixed.
Claudio Sacerdoti Coen [Fri, 24 Mar 2006 17:57:58 +0000 (17:57 +0000)]
% forgot in the output
Claudio Sacerdoti Coen [Fri, 24 Mar 2006 17:56:35 +0000 (17:56 +0000)]
Serious test for the Coq library added (name test_library).
Input: a file that holds a list of URIs.
Output: the benchmark; the output file is also a valid input.
In this case the output also prints the differences w.r.t. the
last execution.
Ctrl-break can be used to interrupt the type-checking of a single URI.
An interactive question is posed to the user to decide if proceeding with
the next URIs or not.
Claudio Sacerdoti Coen [Fri, 24 Mar 2006 17:54:21 +0000 (17:54 +0000)]
more error messages were on stdout :-(
Claudio Sacerdoti Coen [Fri, 24 Mar 2006 17:52:05 +0000 (17:52 +0000)]
error message was printed on stdout
Claudio Sacerdoti Coen [Fri, 24 Mar 2006 16:07:10 +0000 (16:07 +0000)]
Colors are back! :-)
Andrea Asperti [Fri, 24 Mar 2006 16:01:08 +0000 (16:01 +0000)]
New unification and new matching.
Claudio Sacerdoti Coen [Fri, 24 Mar 2006 15:36:03 +0000 (15:36 +0000)]
Legend for profiling printed iff something is profiled.
Enrico Tassi [Thu, 23 Mar 2006 11:12:15 +0000 (11:12 +0000)]
fix
Enrico Tassi [Thu, 23 Mar 2006 10:49:30 +0000 (10:49 +0000)]
fix
Andrea Asperti [Thu, 23 Mar 2006 10:33:41 +0000 (10:33 +0000)]
Best parameter setting for de morgan.
-- This line, and those below, will be ignored--
M tactics/paramodulation/saturation.ml
Enrico Tassi [Thu, 23 Mar 2006 09:06:06 +0000 (09:06 +0000)]
fix
Enrico Tassi [Wed, 22 Mar 2006 14:29:21 +0000 (14:29 +0000)]
if DISTRIBUTED all targets require a depend-stamp
Andrea Asperti [Wed, 22 Mar 2006 13:47:39 +0000 (13:47 +0000)]
:
This line, and those below, will be ignored--
M tactics/paramodulation/utils.ml
Enrico Tassi [Wed, 22 Mar 2006 12:27:01 +0000 (12:27 +0000)]
profiler on steroids. added -profile-only to specify a regex to choose the profilings to print
Enrico Tassi [Wed, 22 Mar 2006 09:18:47 +0000 (09:18 +0000)]
removed mysql_escape that should be ok.... but adds some \ that make mysql
complain
Enrico Tassi [Tue, 21 Mar 2006 17:40:53 +0000 (17:40 +0000)]
added calls number
Stefano Zacchiroli [Tue, 21 Mar 2006 16:50:37 +0000 (16:50 +0000)]
removed mention of the "library" target, no longer needed for the installation
Enrico Tassi [Tue, 21 Mar 2006 16:18:45 +0000 (16:18 +0000)]
fixed timestamp issue on tactics.mli
Stefano Zacchiroli [Tue, 21 Mar 2006 15:21:28 +0000 (15:21 +0000)]
mock-up code for tactics contextual menu in the gui
Andrea Asperti [Tue, 21 Mar 2006 15:11:34 +0000 (15:11 +0000)]
Changed the type of compute-equality_weight that now takes also
in input the ordering
Enrico Tassi [Tue, 21 Mar 2006 14:01:53 +0000 (14:01 +0000)]
go
Enrico Tassi [Tue, 21 Mar 2006 13:58:58 +0000 (13:58 +0000)]
go