Commit graph

1051 commits

Author SHA1 Message Date
Leonardo de Moura
ffdeb0edc4 fix(frontends/lean/elaborator): unsolved metavariables, fix #329 2014-11-26 11:56:39 -08:00
Leonardo de Moura
ef75cac1c0 feat(kernel/expr): change the rules for inferring implicit arguments, closes #344 2014-11-25 12:54:07 -08:00
Leonardo de Moura
e81d9c9184 perf(kernel/level): apply two simple normalization rules at mk_max
They are variations of:
max l1 (max l1 l2) == (max l1 l2)
2014-11-16 18:16:57 -08:00
Leonardo de Moura
67de3b06f3 feat(kernel/level): improve universe level pretty printer
Example: produce `l+2` instead of `succ (succ l)`.
2014-11-14 14:51:03 -08:00
Leonardo de Moura
ffffabad95 feat(kernel/level): improve is_geq procedure for universe levels
Now, it also returns true for

          (succ^k1 a)  =?=  k2

where k1 >= k2
2014-11-14 14:20:35 -08:00
Leonardo de Moura
51719145f9 feat(library/unifier): solved universe constraints of the form succ^k1 a = max k2 ?m (when k1 >= k2) 2014-11-12 17:28:33 -08:00
Leonardo de Moura
50973bb4f3 feat(frontends/lean): default 'eval' command ignores opaque/irreducible annotations
To retrieve the previous behavior, we should use [strict] modifier
2014-11-10 12:46:04 -08:00
Leonardo de Moura
8d05238533 fix(kernel/error_msgs): ignore binder information when comparing type mismatches, fixes #304 2014-11-06 21:03:51 -08:00
Leonardo de Moura
af88e34588 fix(kernel/inductive): bug in eliminator for recursive datatypes in Prop 2014-11-06 14:03:28 -08:00
Leonardo de Moura
e499f8e20a feat(kernel/inductive): relax conditions for an inductive datatype in Prop to be able to eliminate into Type
The new relaxed version allows us to define the "accessability"
proposition and have an eliminator into Type.

See justification in the new comments at inductive.cpp
2014-11-06 09:36:54 -08:00
Leonardo de Moura
84b516994c fix(library/tactic): type check generalization result, fixes #273 2014-10-29 20:34:01 -07:00
Leonardo de Moura
777aa63660 fix(kernel/inductive): relax eliminator generation rules for empty types
This commit also removes the workaround false.rec_type. It is not needed anymore
2014-10-28 10:31:00 -07:00
Leonardo de Moura
7516fcad97 feat(kernel/type_checker): add is_stuck method, and improve ensure_pi method, closes #261 2014-10-27 13:16:50 -07:00
Leonardo de Moura
7240a1a640 feat(kernel/inductive): add get_num_minor_premises and get_num_type_formers APIs 2014-10-25 11:17:29 -07:00
Leonardo de Moura
2bc034da2c feat(kernel/inductive): expose 'get_elim_name' API 2014-10-25 10:47:12 -07:00
Leonardo de Moura
9e69a95b26 feat(kernel/inductive): add API for retrieving the number of indices in an inductive datatype 2014-10-25 10:42:05 -07:00
Leonardo de Moura
c30c0fa3b8 fix(kernel/metavar): avoid crash due to stack overflow, closes #253 2014-10-25 00:20:59 -07:00
Leonardo de Moura
40235c6af0 fix(kernel/type_checker): propagate 'memoize' flag to default_converter 2014-10-23 13:15:53 -07:00
Leonardo de Moura
33f18b9454 fix(kernel/converter): remove buggy eta-reduction and rely only on eta-expansion
The bug is exposed by new unit test
2014-10-21 16:54:25 -07:00
Leonardo de Moura
6a31a79265 refactor(frontends/lean/inductive_cmd): move auxiliary method to expr.h 2014-10-18 15:11:26 -07:00
Leonardo de Moura
2369388629 refactor(kernel/instantiate): cleanup beta-reduce 2014-10-17 20:19:51 -07:00
Leonardo de Moura
1ce3b83d79 fix(kernel/metavar): compilation error in some compilers 2014-10-17 17:22:25 -07:00
Leonardo de Moura
58aff5b4af perf(kernel/metavar): used thread local cache for instantiate_metavars 2014-10-17 17:08:35 -07:00
Leonardo de Moura
6d64da2981 perf(kernel/expr_eq_fn): use thread local cache, and avoid memory allocation/deallocation 2014-10-17 16:44:20 -07:00
Leonardo de Moura
7cc3dd0b4d chore(kernel/replace_fn): "name" constant 2014-10-17 14:20:35 -07:00
Leonardo de Moura
b6afbcb7f5 perf(kernel/replace_fn): use 'recursive' replace_fn for "small" terms 2014-10-17 14:15:09 -07:00
Leonardo de Moura
10b880ce3b perf(kernel/metavar): improve occurs_expr and occurs performance 2014-10-17 14:05:22 -07:00
Leonardo de Moura
d2cbd25985 refactor(kernel): replace_visitor doesn't need to be in the kernel anymore 2014-10-17 10:23:35 -07:00
Leonardo de Moura
14ebbda8cb perf(kernel/metavar): minor improvement 2014-10-16 21:43:54 -07:00
Leonardo de Moura
fe484b26f3 refactor(kernel/expr): remove dead code 2014-10-16 13:45:36 -07:00
Leonardo de Moura
3b6b23c921 refactor(kernel/expr): remove silly overloads 2014-10-16 13:37:55 -07:00
Leonardo de Moura
c4f02bd16a refactor(kernel/expr): remove dead code 2014-10-16 13:09:26 -07:00
Leonardo de Moura
97a7dae12a refactor(kernel/expr): remove code duplication 2014-10-16 12:48:49 -07:00
Leonardo de Moura
814778abb1 refactor(kernel/expr): tag expressions at "creation" time 2014-10-15 13:12:09 -07:00
Leonardo de Moura
de7c850782 feat(kernel/converter): relax is_def_eq test for local constants and meta-variables 2014-10-14 15:31:57 -07:00
Leonardo de Moura
940092a975 fix(kernel/constraint): bug in operator<< 2014-10-14 15:22:03 -07:00
Leonardo de Moura
a26618e0f2 feat(frontends/lean): add '[]' notation for marking arguments where class-instance resolution should be applied 2014-10-12 13:06:00 -07:00
Leonardo de Moura
235b8975d2 feat(kernel/inductive): K-like reduction in the kernel
Given (H_1 : a = a), we have that
      eq.rec H_2 H_1
reduces to H_2

This is not exclusive to equality.
It applies to any inductive datatype in Prop, containing only one
constructor with zero "arguments" (we say they are nullary).

BTW, the restriction to only one constructor is not needed, but it is
does not buy much to support multiple nullary constructors since Prop is
proof irrelevant.
2014-10-10 14:37:45 -07:00
Leonardo de Moura
5a71542aeb feat(kernel/inductive): track when K-like reduction can be used 2014-10-10 08:50:24 -07:00
Leonardo de Moura
93e0599a0f feat(kernel/level): add missing normalization rule 2014-10-08 22:23:19 -07:00
Leonardo de Moura
92c424936a refactor(kernel/macro_definition_cell): improve macro get_type API 2014-10-07 16:38:31 -07:00
Leonardo de Moura
bf081ed431 refactor(kernel): rename var_decl to constant_assumption
Motivation: it matches the notation used to declare it.
2014-10-02 17:55:34 -07:00
Leonardo de Moura
98e66586e9 feat(frontends/lean/elaborator): elaborator rejects 'Type' if the universe is explicit 2014-10-02 14:29:51 -07:00
Leonardo de Moura
c46ec6a548 fix(frontends/lean): missing type information for INFO, fixes #218 2014-10-01 14:29:07 -07:00
Leonardo de Moura
966366e18e feat(kernel/inductive): relaxed rules for defining datatypes with explicit universes, closes #217 2014-10-01 10:56:05 -07:00
Leonardo de Moura
a0c37b231f feat(kernel/expr): add hash_bi function that takes binder information into account 2014-09-29 16:44:55 -07:00
Leonardo de Moura
9f6a8827e0 refactor(*): use name_map 2014-09-28 10:23:11 -07:00
Leonardo de Moura
22e47430b5 feat(library/unifier): add 'on-demand' choice constraints, they are processed as soon as their type does not contain meta-variables anymore 2014-09-27 21:50:39 -07:00
Leonardo de Moura
8e7aac1eb4 fix(frontends/lean): add 'eval' command 2014-09-26 20:16:03 -07:00
Leonardo de Moura
03f71c73dc perf(kernel/instantiate): cache result of instantiate_type_univ_params
and instantiate_value_univ_params
2014-09-25 13:24:43 -07:00
Leonardo de Moura
28796593e3 fix(kernel/type_checker): bug in new is_opaque method 2014-09-25 12:24:14 -07:00
Leonardo de Moura
1d92097781 refactor(kernel/declaration): return reference to type/value/name 2014-09-25 12:17:04 -07:00
Leonardo de Moura
d4236e40b4 feat(kernel/type_checker): expose is_opaque 2014-09-25 11:19:54 -07:00
Leonardo de Moura
516c0c73b9 refactor(*): remove dependency to thread_local C++11 keyword, the
current compilers have several bugs associated with it

We use the simpler __thread (gcc and clang) and
__declspec(thread) (visual studio).
2014-09-24 12:51:04 -07:00
Leonardo de Moura
ca1b8ca80f refactor(util/memory_pool): simplify memory_pool, it is not a template anymore 2014-09-24 10:48:32 -07:00
Leonardo de Moura
9b61b18eaa perf(kernel/replace_visitor): use more expensive/precise cache
It does not use pointer equality, but structural equality
2014-09-24 10:12:29 -07:00
Leonardo de Moura
358074ae3d refactor(kernel/record): remove kernel extension for records, we will
implement it outside of the kernel on top of the inductive datatypes
2014-09-24 10:12:28 -07:00
Leonardo de Moura
da481c3274 refactor(kernel): explicit initialization/finalization
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2014-09-24 10:12:28 -07:00
Leonardo de Moura
531046626a refactor(*): explicit initialization/finalization for environment extensions 2014-09-22 17:30:29 -07:00
Leonardo de Moura
b6781711b1 refactor(*): explicit initialization/finalization for serialization
modules, expression annotations, and tactics
2014-09-22 15:26:41 -07:00
Leonardo de Moura
49d5af473d refactor(kernel): remove support for proof irrelevant classes
Motivation: we can use Prop
2014-09-19 07:32:07 -07:00
Leonardo de Moura
243f80231a chore(kernel): fix style and lua bindings 2014-09-17 18:30:28 -07:00
Leonardo de Moura
ea1bae0143 refactor(kernel/converter): replace extra_opaque_set with predicate
It gives us more flexibility.
2014-09-17 17:05:13 -07:00
Leonardo de Moura
80fd14b39e refactor(frontends/lean): replace collect_metavars with metavar_closure helper class 2014-09-11 14:49:35 -07:00
Leonardo de Moura
4a4de27a6c refactor(frontends/lean/elaborator): move local_context to separate file 2014-09-10 11:20:16 -07:00
Leonardo de Moura
38058450d4 fix(frontends/lean/elaborator): whnf may produce assertion violation if term contains free variables 2014-09-09 16:09:05 -07:00
Leonardo de Moura
3310eb3dfc feat(frontends/lean): remove restriction on implict arguments, add new test that demonstrates the new feature
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2014-09-07 12:29:32 -07:00
Leonardo de Moura
8743394627 refactor(kernel/inductive): replace recursor name, use '.rec' instead of '_rec'
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2014-09-04 15:04:57 -07:00
Leonardo de Moura
79acd3e1b7 fix(kernel/converter): missing case
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2014-08-28 15:21:57 -07:00
Leonardo de Moura
44c597724b fix(frontends/lean): fail if theorem type has metavariables after type elaboration (and before proof elaboration)
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2014-08-26 09:01:17 -07:00
Leonardo de Moura
04d9eb17d1 feat(kernel/conveter): improve support for proof irrelevant in the converter
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2014-08-25 11:19:18 -07:00
Leonardo de Moura
fbf13994d8 refactor(*): use + for concatenating format objects
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2014-08-24 09:35:25 -07:00
Leonardo de Moura
d4ac482d76 refactor(kernel): move annotation to library
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2014-08-22 10:38:10 -07:00
Leonardo de Moura
7d987df429 refactor(kernel/formatter): move simple_formatter to library
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2014-08-22 10:26:45 -07:00
Leonardo de Moura
937c465685 fix(library/unifier): too much reduction
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2014-08-21 21:37:51 -07:00
Leonardo de Moura
359c72b02f fix(frontends/lean/pp): bug when pretty printing binder information, fixes #73
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2014-08-21 09:28:07 -07:00
Leonardo de Moura
9588336c15 refactor(kernel/type_checker): remove "global" constraint buffer from type_checker, and use constraint_seq instead
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2014-08-20 16:46:19 -07:00
Leonardo de Moura
4cf3d32e0c chore(*): create alias for std::pair
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2014-08-20 16:46:19 -07:00
Leonardo de Moura
dc1613f535 feat(frontends/lean): annotate 'notation' subterms with 'noinfo' annotation (goal: improve typeinfo generation); fix initialization problem (with annotations)
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2014-08-15 15:07:14 -07:00
Leonardo de Moura
562926e7ad refactor(kernel/instantiate): add functions instantiate_value_univ_params and instantiate_type_univ_params
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2014-08-12 17:40:30 -07:00
Leonardo de Moura
8ad0949269 fix(kernel/environment): initialization problem on OSX
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2014-08-09 16:14:35 -07:00
Leonardo de Moura
4bcde576b8 perf(kernel/abstract): improve mk_binding performance
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2014-08-09 14:14:14 -07:00
Leonardo de Moura
4986226e41 fix(kernel/converter): missing delay_check case
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2014-08-09 08:07:41 -07:00
Leonardo de Moura
24e8dca014 feat(library/explicit): allow 'as-is', 'explicit' and 'implicit' annotations to be saved in .olean files
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2014-08-07 19:13:48 -07:00
Leonardo de Moura
2486c483cf chore(kernel/error_msgs): change type mismatch error messages, closes #33
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2014-08-07 16:18:40 -07:00
Leonardo de Moura
9e6c5695be fix(library/unifier): make sure the imitation step is type correct, fix ensure_sufficient_args
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2014-08-07 16:09:45 -07:00
Leonardo de Moura
e6ffda0c51 feat(library/match): add basic match_plugin that just invokes whnf before failing
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2014-08-05 08:37:03 -07:00
Leonardo de Moura
0d5e346143 fix(library/expr_lt): make sure the builtin order is AC-compatible
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2014-08-04 15:51:10 -07:00
Leonardo de Moura
8bd36dabce refactor(kernel/pos_info_provider): get_pos_info return none if position is not available
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2014-08-01 20:17:26 -07:00
Leonardo de Moura
288831dc66 fix(kernel/formatter): fixes #21
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2014-08-01 15:07:01 -07:00
Leonardo de Moura
8e222e6244 fix(kernel/converter): missing constraints
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2014-07-31 12:25:35 -07:00
Leonardo de Moura
450131692a fix(library/converter): missing constraint on eta expansion
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2014-07-30 10:43:47 -07:00
Leonardo de Moura
936bb2744b fix(library/unifier): add a flag to sign that a choice constraint owns a metavariable ?m, that is, it has the right to assign ?m, and the unifier should postpone any other constraint that tries to assign ?m
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2014-07-29 17:32:55 -07:00
Leonardo de Moura
7df78ea503 feat(kernel): add combinator for combining normalizer_extensions
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2014-07-29 10:00:40 -07:00
Leonardo de Moura
33c77afc29 feat(frontends/lean/structure): add 'structure' command skeleton
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2014-07-28 19:59:38 -07:00
Leonardo de Moura
5f360cd8ec feat(kernel/error_msgs): improve application type mismatch error messages
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2014-07-28 07:08:12 -07:00
Leonardo de Moura
555425539d refactor(kernel/error_msgs): remove dead code
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2014-07-28 06:52:59 -07:00
Leonardo de Moura
faee08591f fix(*): make sure elaborator and type_checker use the same "rules" for treating opaque definitions
This is a big change because we have to store in constraints whether we can use the "relaxed" rules or not.
The "relaxed" case says that when type checking the value of an opaque definition we can treat other opaque definitions in the same module as transparent.

Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2014-07-27 12:12:54 -07:00
Leonardo de Moura
83d38674c9 feat(kernel/error_msgs): improve cryptic type mismatch error messages where the types may seem identical because key information is being suppressed
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2014-07-27 09:41:25 -07:00
Leonardo de Moura
8e402ae862 fix(kernel/type_checker): error message position information
Signed-off-by: Leonardo de Moura <leonardo@microsoft.com>
2014-07-26 20:33:44 -07:00