max/lean4-htt
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions 2
30635 commits 1 branch 0 tags 5.3 GiB
Commit graph

1379 commits

Author SHA1 Message Date
Gabriel Ebner
d3c55ef249 perf: do not reset tc cache when adding local instances 2023-03-16 15:26:38 -07:00
Gabriel Ebner
0cc9d7a43d fix: do not reverse subgoals of local instances 2023-03-08 15:54:07 -08:00
Gabriel Ebner
2262579f9b fix: tc: filter out assigned subgoals at the correct place 2023-03-08 15:54:07 -08:00
Gabriel Ebner
3ab859553e fix: allow function coercion to assign universe mvars 2023-03-08 15:54:07 -08:00
Gabriel Ebner
1c641b569a chore: synthInstance trace message on cache hit 2023-03-08 15:54:07 -08:00
Gabriel Ebner
1f61633da7 fix: typo in trace class name 2023-03-08 15:54:07 -08:00
Gabriel Ebner
adcca17991 chore: add option to enable structure eta in tc search 2023-02-21 16:41:30 -08:00
Gabriel Ebner
75252d2b85 perf: whnf projections during defeq 2023-02-09 19:54:23 -08:00
Gabriel Ebner
ecc74c5a9d fix: defeq condition for projections 2023-02-09 19:54:23 -08:00
Gabriel Ebner
448f49ee91 Revert "fix: reenable structure eta during tc search" 
The fix is blocked by slow defeq checks for TC instances; see issues
1986 and 2055.  Enabling it right now causes lots of timeouts in
mathlib4.
https://leanprover.zulipchat.com/#narrow/stream/287929-mathlib4/topic/bump.20to.202023-02-06/near/326223768

This reverts commit 15a045ee66.
 2023-02-09 11:37:30 -08:00
Gabriel Ebner
15a045ee66 fix: reenable structure eta during tc search 
Fixes #2074.
 2023-02-05 11:41:00 -08:00
Gabriel Ebner
c21d2f29a2 perf: do not backtrack after eta-defeq 2023-01-09 16:12:02 -08:00
Leonardo de Moura
fedf235cba fix: fixes #2011 
In Lean 4, we have support for typing constraints of the form
```
(?m ...).1 =?= v
```
where the type of `?m ...` is a structure with a single field.
This kind of constraint is reduced to `?m ... =?= ⟨v⟩`

This feature is implemented by the function `isDefEqSingleton`.
As far as I remember, Lean 3 does not implement this feature.

This commit disables this feature if the structure is a class.
The goal is to avoid the generation of counterintuitive instances by
typing inference.

For example, in the example at issue #2011, the following weird
instance was being generated for `Zero (f x)`
```
(@Zero.mk (f x✝) ((@instZero I (fun i => f i) fun i => inst✝¹ i).1 x✝)
```
where `inst✝¹` is the local instance `[∀ i, Zero (f i)]`
Note that this instance is definitinally equal to the expected nicer
instance `inst✝¹ x✝`.
However, the nasty instance trigger nasty unification higher order
constraints later.

Note that a few tests broke because different error messages were
produced. The new error messages seem better. I do not expect this
change to affect Mathlib4 since Lean 3 does not have this feature.
 2023-01-05 17:33:45 -08:00
Leonardo de Moura
dd682bf1d5 feat: add support for HO projections at DiscrTree 
closes #1937

Requires update stage0
 2023-01-05 13:33:43 -08:00
Leonardo de Moura
30c9f58e6a chore: remove leftover 2023-01-03 15:05:31 -08:00
Gabriel Ebner
70a6c06eef fix: erase *dependent* local instances 2023-01-03 11:39:46 -08:00
Gabriel Ebner
a2f5959118 chore: use deriving Nonempty 2022-12-22 03:48:15 +01:00
Gabriel Ebner
7736c051ff fix: assigned tc mvar check 2022-12-22 02:02:55 +01:00
Gabriel Ebner
6083b01c86 fix: remove maxCoeSize option 2022-12-22 02:02:55 +01:00
Gabriel Ebner
f798507bbf chore: tc: only normalize level mvars at current depth 2022-12-22 02:02:55 +01:00
Gabriel Ebner
572ffe77e3 fix: implement assertAfter using revert 2022-12-21 21:42:07 +01:00
Gabriel Ebner
eeab2af7ae fix: remove Inhabited Environment instance 2022-12-21 20:08:08 +01:00
Gabriel Ebner
e59ddb0c16 refactor: replace hardcoded list of coercions by attribute 2022-12-21 04:24:39 +01:00
Gabriel Ebner
e386d5941e refactor: replace ignoreLevelMVarDepth by levelAssignDepth 2022-12-19 20:14:17 +01:00
Sebastian Ullrich
6169435259 refactor: consolidate MessageData constructors into lazy formatting with infos 2022-12-07 19:16:25 +01:00
Leonardo de Moura
0dda3a8c02 fix: include instance implicits that depend on outParams at outParamsPos 
This fixes the fix for #1852
 2022-12-01 06:11:48 -08:00
Leonardo de Moura
0a031fc9bb chore: replace Expr.forEach with Expr.forEachWhere 2022-12-01 05:19:32 -08:00
Leonardo de Moura
8fc3d77a0b feat: add trace.Meta.Tactic.simp.numSteps and trace.Meta.Tactic.simp.heads 2022-11-30 07:07:07 -08:00
Leonardo de Moura
069873d8e5 fix: fixes #1891 2022-11-29 08:59:46 -08:00
Leonardo de Moura
c510d16ef5 fix: fixes #1808 2022-11-28 07:48:54 -08:00
Leonardo de Moura
4be7543adb feat: add APIs for issue #1878 
We need `update-stage0` because this commit affects the .olean format.
 2022-11-24 12:55:41 -08:00
Leonardo de Moura
9d8b324f8d fix: fixes #1869 
Better support for simplifying class projections.
 2022-11-24 11:56:36 -08:00
Leonardo de Moura
71b7562c2f fix: class projection at DiscrTree 2022-11-24 11:56:36 -08:00
Leonardo de Moura
0694731af8 fix: fixes #1870 2022-11-23 05:49:19 -08:00
Leonardo de Moura
9c561b593a feat: add withTraceNodeBefore and use it at ExprDefEq 
`withTraceNode` uses the metavariable context after executing
`k`. This is bad when debugging `isDefEq`.
 2022-11-22 14:43:28 -08:00
Leonardo de Moura
dfaf9c6ebd chore: register missing trace options, and fix inherited parameter 
The current setting was bad for debugging `isDefEq` issues.
 2022-11-22 14:43:28 -08:00
Leonardo de Moura
556b6706ee fix: fixes #1856 2022-11-18 19:34:32 -08:00
Leonardo de Moura
d2a5ea137d fix: fixes #1842 2022-11-16 17:29:41 -08:00
Leonardo de Moura
edadd8c034 fix: fixes #1841 
This commit adds the configuration option
`ApplyConfig.approx` (available in Lean 3), and sets it to true by
default like in Lean 3.
 2022-11-16 14:46:05 -08:00
Leonardo de Moura
98922b878a fix: fixes #1815 2022-11-15 17:08:54 -08:00
Leonardo de Moura
8225be2f0e feat: ensure projections are not reducing at DiscrTree V (simpleReduce := true) 
Now, the `simp` discrimination tree does not perform `iota` nor reduce
projections.
 2022-11-15 16:47:12 -08:00
Leonardo de Moura
1b0c2f7157 feat: parameterize DiscrTree indicating whether non trivial reductions are allowed or not when indexing/retrieving terms 2022-11-15 16:47:12 -08:00
Leonardo de Moura
81c84bf045 feat: do not perform iota reduction at the discrimination tree module 2022-11-15 16:47:12 -08:00
Gabriel Ebner
75aa732af5 doc: mention ignoreLevelMVarDepth 2022-11-15 11:29:26 -08:00
Leonardo de Moura
1cc58e60ef fix: fixes #1829 2022-11-15 08:31:36 -08:00
Leonardo de Moura
c3d86001c4 chore: simplify dependencies at MatchEqs 2022-11-11 05:51:05 -08:00
Leonardo de Moura
8030aed56e fix: fixes #1814 
Missing `hasAssignableMVar` checks.
 2022-11-10 20:42:02 -08:00
Mario Carneiro
999b61007c
feat: tweak behavior of congrN to match lean 3 (#1798) 2022-11-04 06:55:13 -07:00
Mario Carneiro
ad1c23f172 fix: bug in replaceLocalDeclDefEq 
fixes #1615
 2022-11-01 19:18:25 -07:00
Leonardo de Moura
5f38a483f2 fix: congr tactic 
`MVarId.congr?` and `MVarId.hcongr?` should return `none` if an
exception is thrown while applying congruence theorem.

`MVarId.hcongr?` should try `eq_of_heq` before trying to apply
`hcongr` theorem.

closes #1787

BTW: Lean 4 `congr` tactic is applying `assumption`. Lean 3 version does not.
 2022-10-28 08:00:04 -07:00