michael/lean2
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions
8723 commits 3 branches 0 tags 28 MiB
Commit graph

1909 commits

Author SHA1 Message Date
Leonardo de Moura
0268f92eb4 fix(frontends/lean/elaborator): fixes #965 2016-02-04 13:41:21 -08:00
Leonardo de Moura
42fbc63bb6 fix(library/tc_multigraph): avoid name collisions 
@avigad, @fpvandoorn, @rlewis1988, @dselsam

I changed how transitive instances are named.
The motivation is to avoid a naming collision problem found by Daniel.
Before this commit, we were getting an error on the following file
tests/lean/run/collision_bug.lean.

Now, transitive instances contain the prefix "_trans_".
It makes it clear this is an internal definition and it should not be used
by users.

This change also demonstrates (again) how the `rewrite` tactic is
fragile. The problem is that the matching procedure used by it has
very little support for solving matching constraints that involving type
class instances. Eventually, we will need to reimplement `rewrite`
using the new unification procedure used in blast.

In the meantime, the workaround is to use `krewrite` (as usual).
 2016-02-04 13:15:42 -08:00
Leonardo de Moura
587c263c28 feat(library/blast/forward/ematch): improve support for casts in the e-matcher 2016-01-29 11:38:43 -08:00
Leonardo de Moura
684995640a fix(library/blast/unit/unit_propagate): memory access violation 2016-01-27 15:22:34 -08:00
Leonardo de Moura
fb95b71a5e fix(library/blast/forward/pattern): bug in the pattern inference code 2016-01-27 13:39:19 -08:00
Leonardo de Moura
810ee9759c fix(library/blast/backward/backward_action): add missing normalize at backward_action, and remove incorrect fix at discrimination tree 2016-01-26 20:35:57 -08:00
Leonardo de Moura
4821af8685 feat(frontends/lean/scanner): disallow superscripts in identifiers 
See new test for motivating example.
 2016-01-26 18:46:40 -08:00
Leonardo de Moura
6f6672eaaa fix(library/blast/discr_tree): ignore annotations in discrimination trees 2016-01-26 18:46:40 -08:00
Leonardo de Moura
d12067193b fix(library/blast/forward/pattern): residue computation in the ematching module 2016-01-24 16:15:33 -08:00
Leonardo de Moura
38ab1cae9e feat(library/blast/unit/unit_propagate): basic support for "dependent lemmas" at unit propagate 
New test contains examples of "dependent lemmas"
 2016-01-24 16:02:08 -08:00
Leonardo de Moura
297ef10611 fix(library/blast/congruence_closure): subsingleton propagation in the congruence closure module 
We must normalize inferred type.
 2016-01-24 14:55:28 -08:00
Leonardo de Moura
3d0ea4c9d1 feat(library/type_context): improve find_unsynth_metavar 2016-01-24 13:39:25 -08:00
Leonardo de Moura
19bfbe2df8 fix(library/blast/congruence_closure): uselist initialization (aka add_occurrence) 
Make sure it matches the description in the paper.
 2016-01-16 19:53:36 -08:00
Leonardo de Moura
1d04160a9b fix(library/blast/congruence_closure): bug at eq_congr_key_cmp::operator()(eq_congr_key const & k1, eq_congr_key const & k2) 
It was not implementing the condition described in our paper.
 2016-01-16 19:33:24 -08:00
Daniel Selsam
a883101a3b feat(library/blast/forward/ematch): basic support for heq classes 2016-01-15 21:38:16 -08:00
Leonardo de Moura
60861d7dea chore(tests/lean/run/blast_cc_heq9): make sure that only congruence closure module is used 2016-01-13 22:13:04 -08:00
Leonardo de Moura
37fdac4d65 test(tests/lean/run/blast_cc_heq9): add new test 2016-01-13 22:10:15 -08:00
Daniel Selsam
7a005fb76b test(tests/lean/run/blast_vector_test): more tests 2016-01-13 22:03:47 -08:00
Leonardo de Moura
3f7122ce07 feat(library/blast/congruence_closure): more general congruence lemmas 2016-01-13 21:44:32 -08:00
Leonardo de Moura
912c5254de fix(tests/lean/run/cases_bug): adjust test to changes in the std lib 2016-01-13 16:37:33 -08:00
Leonardo de Moura
05c3ede683 feat(tests/lean/run/blast_vector_test): test ematch + cc at vectors 2016-01-13 16:35:21 -08:00
Leonardo de Moura
3643e79cb3 feat(library/blast/congruence_closure): improve the suppoer for subsingletons in the ematching procedure 2016-01-13 11:17:42 -08:00
Leonardo de Moura
8f7b533ca1 refactor(library): move 'cast' to init folder 2016-01-13 11:17:42 -08:00
Leonardo de Moura
081ad1212b test(tests/lean/run/blast_ematch_list): blast last_concat 2016-01-12 22:56:28 -08:00
Leonardo de Moura
12876ccc20 fix(library/blast/forward/ematch): ematch + subsingleton 2016-01-12 22:31:09 -08:00
Daniel Selsam
292f28e769 test(tests/lean/run/blast_tuple): more tests for blast + tuple 2016-01-11 17:37:44 -08:00
Leonardo de Moura
d736376a6f test(tests/lean/run/blast_cc_heq8): new congruence closure procedure can also handle subtypes without using a flattening preprocessing step 2016-01-11 16:50:03 -08:00
Leonardo de Moura
b40f0ffe8b fix(library/blast/forward/ematch): keep using ematching on implicit arguments 
Ematching module should only ignore type classes (i.e., instance
implicit) and subsingletons (which includes propositions).
 2016-01-11 15:40:51 -08:00
Leonardo de Moura
1c5418ac2e test(tests/lean/run): add example/test for blast + tuples 2016-01-11 12:17:14 -08:00
Leonardo de Moura
5edcccaeb0 feat(library/blast/forward/ematch): add support for the new hcongr lemmas in the ematching module 2016-01-11 11:56:36 -08:00
Leonardo de Moura
ddff37dd0f fix(library/blast/congruence_closure): bug when using blast.cc.heq and handling relation congruences 2016-01-10 19:28:55 -08:00
Leonardo de Moura
799317c43e fix(library/blast/congruence_closure): add missing eq => heq lifting 2016-01-10 18:03:35 -08:00
Leonardo de Moura
3a846a28a3 feat(library/blast/congruence_closure): support for subsingleton propagation 2016-01-10 17:24:12 -08:00
Leonardo de Moura
e9d24ec152 feat(library/blast/congruence_closure): create simpler congruence proofs when using blast.cc.heq 2016-01-10 15:11:31 -08:00
Leonardo de Moura
ea7da31bba feat(library/blast/congruence_closure): support for congruence lemmas that use heterogeneous equality 2016-01-10 13:45:40 -08:00
Leonardo de Moura
934f3b67ff feat(library/blast/congruence_closure): basic support for heterogeneous equality 
We still have to process the general congruence lemmas.
 2016-01-10 12:53:05 -08:00
Leonardo de Moura
b7be3ec6de test(tests/lean/run): add tests for heterogeneous congruence lemma generator 2016-01-09 16:18:39 -08:00
Leonardo de Moura
d3242a2068 refactor(library): rename heq.of_eq heq.to_eq auxiliary lemmas 2016-01-09 12:32:18 -08:00
Leonardo de Moura
27eea05da9 fix(library/blast/discr_tree): bug in the discrimination tree module 2016-01-06 17:30:44 -08:00
Leonardo de Moura
4e8ae94aba chore(tests/lean/run/blast_cc_noconfusion): make sure simp/subst are not used in the test 2016-01-06 17:30:31 -08:00
Leonardo de Moura
76cebb45f9 feat(library/blast/congruence_closure): add support for 'no_confusion' in the congruence closure module 2016-01-06 17:30:25 -08:00
Leonardo de Moura
cb02d1deae feat(library/blast/congruence_closure): add support for specialized congr lemmas in the congruence closure module 2016-01-06 17:30:20 -08:00
Leonardo de Moura
c9930d0a29 feat(library/blast/simplifier/simplifier): subsingleton normalization for application arguments and lambdas 2016-01-06 17:30:08 -08:00
Leonardo de Moura
f3b8aef24c feat(library/fun_info_manager,library/congr_lemma_manager,blast/simplifier): specialized congruence lemmas 
We still need a lot of polishing.
 2016-01-06 17:29:35 -08:00
Leonardo de Moura
3ca785b0e7 refactor(library/fun_info_manager): remove dead code 2016-01-06 17:29:02 -08:00
Rob Lewis
031831f101 feat(library/tactic): add replace tactic 2016-01-04 14:43:31 -05:00
Leonardo de Moura
ba392f504f feat(kernel/expr,library/blast/blast,frontends/lean/decl_cmds): add workaround for allowing users to use blast inside of recursive equations 2016-01-03 21:53:31 -08:00
Jeremy Avigad
a22346e411 fix(lean/tests/lean/run/new_obtain{3,4}): adapt tests to new notation for image 2016-01-03 18:52:25 -08:00
Leonardo de Moura
19ebedd480 feat(library/type_context): improve type_context get_level_core, add virtual method for checking types whenever a metavariable is assigned 
We add an example where app_builder fails without these new features.
That is, app_builder fails to solve the unification problem.
 2016-01-03 17:58:27 -08:00
Leonardo de Moura
66a722ff5a feat(library/unifier): remove "eager delta hack", use is_def_eq when delta-constraint does not have metavariables anymore 
The "eager-delta hack" was added to minimize problems in the interaction
between coercions and delta-constraints.
 2016-01-03 12:39:32 -08:00
Leonardo de Moura
d02ead320a feat(library/unifier): remove unifier.computation option 2016-01-03 11:00:16 -08:00
Leonardo de Moura
9935cbc3d7 feat(library/blast/blast): communicate assigned metavariables back to tactic framework 
We need this feature to be able to solve (input) goals containing
metavariables using blast.
See new test for example.
 2016-01-02 20:05:44 -08:00
Leonardo de Moura
56d9b6b0d3 fix(library/blast/blast): convert uref and mref back into tactic metavariables 2016-01-02 19:23:04 -08:00
Leonardo de Moura
5feef27c2b feat(frontends/lean/notation_cmd): relax restriction on user defined tokens 
Before this commit, Lean would forbid new tokens containing '(' or ')'.
We relax this restriction. Now, we just forbid new tokens starting with '(' or ending with ')'.
 2016-01-02 13:58:46 -08:00
Leonardo de Moura
befe41e8f5 fix(tests/lean): adjust tests 2016-01-02 12:51:20 -08:00
Leonardo de Moura
17a2f1fe47 chore(tests/lean/run/blast_grind1): use 'core_grind' 
Reason: we will change the semantics of 'grind'
 2016-01-01 16:46:15 -08:00
Leonardo de Moura
ac9d6c2021 refactor(library/data/bool): cleanup bool proofs and fix bxor definition 2016-01-01 13:52:42 -08:00
Leonardo de Moura
54f2c0f254 feat(library/blast/forward): inst_simp should use the left-hand-side as a pattern (if none is provided by the user) 
The motivation is to reduce the number of instances generated by ematching.

For example, given

   inv_inv:  forall a, (a⁻¹)⁻¹ = a

the new heuristic uses ((a⁻¹)⁻¹) as the pattern.
This matches the intuition that inv_inv should be used a simplification
rule.

The default pattern inference procedure would use (a⁻¹). This is bad
because it generates an infinite chain of instances whenever there is a
term (a⁻¹) in the proof state.
By using (a⁻¹), we get
   (a⁻¹)⁻¹ = a
Now that we have (a⁻¹)⁻¹, we can match again and generate
   ((a⁻¹)⁻¹)⁻¹ = a⁻¹
and so on
 2015-12-31 20:20:39 -08:00
Leonardo de Moura
b117a10f82 refactor(library/blast/simplifier): use priority_queue to store simp/congr lemmas, use name convention used at forward/backward lemmas, normalize lemmas when blast starts, cache get_simp_lemmas 2015-12-28 17:52:57 -08:00
Leonardo de Moura
c8b9c98eb6 refactor(library/blast/backward): use priority_queue, make sure head is normalized when building index 2015-12-28 12:26:06 -08:00
Leonardo de Moura
f177082c3b refactor(*): normalize metaclass names 
@avigad and @fpvandoorn, I changed the metaclasses names. They
were not uniform:
- The plural was used in some cases (e.g., [coercions]).
- In other cases a cryptic name was used (e.g., [brs]).

Now, I tried to use the attribute name as the metaclass name whenever
possible. For example, we write

   definition foo [coercion] ...
   definition bla [forward] ...

and

  open [coercion] nat
  open [forward] nat

It is easier to remember and is uniform.
 2015-12-28 10:39:15 -08:00
Leonardo de Moura
96bec8b4f9 fix(frontends/lean/builtin_cmds): allow token and metaclass to have the same name 2015-12-28 09:57:45 -08:00
Leonardo de Moura
fe66e2aa4a fix(tests/lean): subtype notation is not in the top-level anymore 2015-12-28 09:04:11 -08:00
Leonardo de Moura
657b508cdf fix(tests/lean/simplifier_light): adjust test 2015-12-17 22:42:31 -08:00
Leonardo de Moura
d81b2d0f29 feat(library/attribute_manager, frontends/lean/builtin_cmds): use attribute manager information when pretty print definitions 2015-12-17 21:16:31 -08:00
Leonardo de Moura
2502039a5c fix(frontends/lean/parser): tactic notation that may take numerical parameters 2015-12-17 11:27:31 -08:00
Leonardo de Moura
73b28c91a6 fix(library/type_context): local constant management bug 2015-12-15 18:49:26 -08:00
Leonardo de Moura
1387cdfa0f feat(library/type_context): add eta-expansion to type_context 2015-12-15 17:24:53 -08:00
Leonardo de Moura
10273bf176 feat(library/blast/forward/pattern): improve pattern inference heuristic 2015-12-15 12:55:43 -08:00
Sebastian Ullrich
2185ee7e95 feat(library/tactic): make let tactic transparent, introduce new opaque note tactic 
The new let tactic is semantically equivalent to let terms, while `note`
preserves its old opaque behavior.
 2015-12-14 10:14:02 -08:00
Leonardo de Moura
193a9d8cde refactor(library/norm_num): avoid manual constant name initialization 
@rlewis1988 We group all Lean constants used in the C++ code at
src/library/constants.txt

Jeremy and Floris check this file before renaming constants in the
library. So, they can quickly decide whether C++ code will be affected
or not.

We also have a python script for initializing the C++ name objects.
To use the script:
   - go to src/library
   - execute
       python ../../script/gen_constants_cpp.py constants.txt

It will create the boring initialization and finalization code, and
declare a procedure get_<id>_name() for each constant in the file constants.txt.

I also move the norm_num1.lean to the set of unit tests that are
executed whenever we push a commit to the main branch.

I found an assertion violation at line 606. Could you take a look?

Best,
Leo
 2015-12-13 21:38:59 -08:00
Leonardo de Moura
e3a35ba4fd feat(frontends/lean): add 'with_attributes' tactical 
closes #494
 2015-12-13 18:27:44 -08:00
Leonardo de Moura
727a4f5a3a feat(library/tactic/intros_tactic): use '_' to say that some names are irrelevant in the intro tactic 
See #695
 2015-12-13 16:47:31 -08:00
Leonardo de Moura
d4e49a8434 feat(library/tactic/intros_tactic): intro without argument should introduce a single variable 
see #695
 2015-12-13 16:28:39 -08:00
Leonardo de Moura
999f23cbc0 feat(kernel/expr_eq_fn): take names into account when CompareBinderInfo is true 
This is the correct fix for the id declaration pretty printing
discrepancy reported by Daniel.

TODO: decide whether we need another eq-mode where names are ignored.
For example, in blast, it makes sense to increase sharing by ignoring
binder names.
 2015-12-13 14:47:11 -08:00
Leonardo de Moura
ce622e9179 feat(frontends/lean): add auto-include for structures and inductive decls 2015-12-13 13:39:34 -08:00
Leonardo de Moura
20de22a8ad feat(frontends/lean): automatically include anonymous instance implicit variables/parameters (whenever their parameters have been included) 2015-12-13 13:20:54 -08:00
Leonardo de Moura
b6bac2e542 chore(tests/lean/run): adjust tests 2015-12-13 12:23:44 -08:00
Leonardo de Moura
ef546c5c5b refactor(library): use anonymous instance implicit arguments 2015-12-13 11:46:48 -08:00
Leonardo de Moura
b7de10a6d2 feat(library/trace): allow user to disable subclasses of a trace class 
Example:

set_option trace.blast true         -- enables trace.blast class and all subclasses
set_option trace.blast.action false -- disables the given subclass

Result: all blast classes are traced but blast.action
 2015-12-11 11:03:16 -08:00
Leonardo de Moura
7f1800962a feat(frontends/lean/pp): allow user to override pp.all setting 
see #922
 2015-12-11 10:40:48 -08:00
Leonardo de Moura
a35fd06a33 test(tests/lean/run/blast_simp4): add simp test 2015-12-11 10:23:58 -08:00
Leonardo de Moura
ac0bd539b0 feat(frontends/lean/notation_cmd): allow 'abstract ... end' to be used in notation declarations 
helps #825
 2015-12-11 09:55:27 -08:00
Daniel Selsam
2b1e7e7759 feat(frontends/lean/pp): all disables numerals 2015-12-10 23:31:40 -08:00
Leonardo de Moura
3057dde885 fix(library/class,library/coercion): fixes #852 2015-12-10 22:52:02 -08:00
Leonardo de Moura
cf61adc5d5 feat(frontends/lean): identifiers starting with '_' are now reserved 2015-12-10 22:32:03 -08:00
Leonardo de Moura
13f9d8dd6c fix(tests/lean): move hott tests 2015-12-10 10:55:27 -08:00
Leonardo de Moura
7b29ee1666 fix(library/tactic/induction_tactic): fixes #892 2015-12-10 10:52:57 -08:00
Leonardo de Moura
8094ca1c70 fix(library/blast/backward): crash when pretty printing backward rule set 
We make sure it doesn't depend on the blast state.
 2015-12-10 10:38:53 -08:00
Leonardo de Moura
1abaa9eb71 fix(frontends/lean/parser): fixes #858 2015-12-10 10:31:14 -08:00
Leonardo de Moura
8b3cbb8fdd fix(library/tactic/induction_tactic): apply substitution to hypothesis type (it may contain metavars) 
closes #876
 2015-12-10 10:11:55 -08:00
Leonardo de Moura
a507ac8594 test(tests/lean/interactive/findp): remove test that breaks whenever we revise init/logic.lean 
closes #888
 2015-12-10 10:11:55 -08:00
Leonardo de Moura
c9ff175cf4 fix(library/tactic/induction_tactic): fixes #893 2015-12-10 10:11:55 -08:00
Floris van Doorn
f495fa04c8 fix(tests): fix tests after port 2015-12-09 12:36:11 -08:00
Leonardo de Moura
50323e5b14 fix(tests/lean/run/simplifier1): option name 2015-12-09 12:30:20 -08:00
Leonardo de Moura
bed03272f5 fix(tests/lean/run/blast_cc1): option name 2015-12-09 12:23:00 -08:00
Leonardo de Moura
3f9549485f feat(frontends/lean/parser): restore config options in the end of sections/namespaces 2015-12-09 11:24:37 -08:00
Leonardo de Moura
273f6fc04c fix(tests/lean/run): new option names 2015-12-08 20:06:43 -08:00
Leonardo de Moura
b145a7332a feat(library/blast/congruence_closure): use new tracing infrastructure 2015-12-08 15:18:21 -08:00
Leonardo de Moura
7da64a768f refactor(library/type_context): with the new tracing infrastructure, type_context doesn't need an io_state 2015-12-08 14:58:08 -08:00