330 commits

Author	SHA1	Message	Date
Ayaz Hafiz	b0c602eb8d	Clippy	2022-06-02 09:44:38 -05:00
Ayaz Hafiz	bde107da49	Remove TagName in favor of explicit TagName or Closure ... Note that we still need a `TagOrClosure` enum for phases past mono, but at least we get rid of the branches in the front end.	2022-06-02 09:44:34 -05:00
Ayaz Hafiz	a8453b0fca	Fix macro	2022-06-01 13:27:24 -05:00
Ayaz Hafiz	c66f989467	Recursive closure layout working again in effect test	2022-06-01 13:18:54 -05:00
Ayaz Hafiz	2e70c9aba4	Getting closer. Checking in for a break.	2022-06-01 13:18:53 -05:00
Ayaz Hafiz	790eb8e20c	Lambda sets don't need extension vars	2022-06-01 13:18:03 -05:00
Ayaz Hafiz	c2a2ce690c	Lift lambda sets as their own type ... This prepares for unspecialized lambda set in the type system in general.	2022-06-01 13:17:22 -05:00
Folkert	9540aa5b70	make deep_copy_var non-recursive	2022-06-01 17:59:35 +02:00
Richard Feldman	4eec34becf	Update to use new square brace formatting	2022-05-22 23:15:54 -04:00
Folkert	98ce9f4d09	remove the snapshots	2022-05-21 20:42:19 +02:00
Folkert de Vries	feaf53371f	Merge pull request #3113 from rtfeldman/new-utable-copy ... New utable variable copy	2022-05-21 18:17:56 +02:00
Richard Feldman	3cdaa52a30	Merge pull request #3110 from rtfeldman/handcrafted-unification-table ... Handcrafted unification table	2022-05-21 11:39:58 -04:00
Folkert	722ae9da6b	pass along the specific pool we're adding to	2022-05-21 14:48:45 +02:00
Folkert	1814a69c30	remove closure that is used once	2022-05-21 14:42:48 +02:00
Folkert	8c44661bc2	use set_content_unchecked; replace modify with a single set	2022-05-21 14:06:43 +02:00
Folkert	f1e1c45f2f	some more unchecked operations	2022-05-21 13:52:44 +02:00
Folkert	a1894195d3	refactoring trivial copy detection	2022-05-21 13:48:28 +02:00
Ayaz Hafiz	ababfa4129	Use two caches instead of a unified one for obligation checking	2022-05-20 17:39:09 -04:00
Folkert	a705d6a6e8	unchecked in making the rank table	2022-05-20 23:28:56 +02:00
Folkert	e20ac3f38f	use unchecked operations in generalization	2022-05-20 23:18:27 +02:00
Folkert	d291a87945	more efficient use of Subs api	2022-05-20 22:55:00 +02:00
Ayaz Hafiz	a4c122d5db	Validate derives clauses after solving	2022-05-20 15:56:55 -04:00
Folkert	4550f3fa88	update deep_copy_var_help	2022-05-20 21:44:46 +02:00
Folkert	be104d25eb	update instantiate rigid	2022-05-20 21:36:05 +02:00
Folkert	f2deb37a30	update instantiate rigid	2022-05-20 21:18:52 +02:00
Folkert	dc155ea95a	update generalize	2022-05-20 21:05:17 +02:00
Folkert	4489a308e4	use modify over getting a mutable reference	2022-05-20 21:02:53 +02:00
Ayaz	eb706d2863	Merge pull request #3100 from rtfeldman/solve-copy-sequence ... use macro in copy_var to clean up logic	2022-05-20 08:50:42 -04:00
Folkert	9d21343227	use macro to clean logic up a bit	2022-05-20 10:29:41 +02:00
Ayaz Hafiz	cf56b5015a	Fix existing reporting tests	2022-05-19 18:23:02 -04:00
Ayaz Hafiz	2c87214b44	Check derivability of abilities for structural types	2022-05-19 18:23:02 -04:00
Ayaz Hafiz	4dede6e570	Simplify instantiate aliases	2022-05-19 18:21:45 -04:00
Ayaz Hafiz	312cdd0b70	Correct imports of Encode and abilities across modules	2022-05-19 18:21:40 -04:00
Richard Feldman	9bc1e320dc	Fix --release warnings	2022-05-15 21:50:49 -04:00
Folkert	7face320eb	move back to one loc_symbols vector	2022-05-14 14:44:37 +02:00
Ayaz Hafiz	710a10a29c	Check for invalid cycles after type solving recursive defs ... Disallow cycles that pass through a non-function value. Since we evaluate eagerly, having one such cycle means there is at least one path in the program that (likely) has unbounded recursion. Of course we can't be certain (halting problem), but it's very likely, and avoids stuff like #1926. Also, mono (as it's done today) won't work if things in a cycle aren't functions. Closes #1926	2022-05-11 14:57:01 -04:00
Ayaz Hafiz	da00c47102	Opportunistically resolve specializations during solving	2022-05-11 08:53:15 -04:00
Richard Feldman	11e2faa87a	Merge pull request #3050 from rtfeldman/bindgen-tag-payloads ... Bindgen single-tag unions	2022-05-10 19:27:25 -04:00
Ayaz Hafiz	3de35f7aa2	Loosen recursion var rank restriction in presence of errors	2022-05-10 11:15:21 -04:00
Richard Feldman	2db527888b	Drop unnecessary parens	2022-05-09 22:31:29 -04:00
Richard Feldman	df7df4ccf8	Merge branch 'trunk' into mfonism/rename-num-float-type-to-frac	2022-05-08 20:54:33 -04:00
Ayaz Hafiz	2cacda00eb	Add some internal error assertions	2022-05-08 15:08:19 -04:00
Ayaz Hafiz	b3845c21bd	Print able type variables correctly	2022-05-08 15:08:19 -04:00
Ayaz Hafiz	154b17d9f0	Correct instantiation of delayed opaques ... I didn't realize that delayed instantiation of aliases instantiates their real var, not the whole alias type. So prior to this, we were wrapping all opaques in a structural `Alias(..)`. Which is okay from a typechecking perspective because Aliases are transparent, but is wasteful.	2022-05-08 15:08:19 -04:00
Ayaz Hafiz	c1b1f60630	Coupon cutting	2022-05-08 15:08:16 -04:00
Ayaz Hafiz	a242a90416	Stale comment	2022-05-08 15:08:16 -04:00
Ayaz Hafiz	2e57bf0b6a	Permit able variables in aliases, and thread them through	2022-05-08 15:08:14 -04:00
Mfon Eti-mfon	85c622b169	Update symbols definition	2022-05-08 15:54:34 +01:00
Ayaz Hafiz	af3a677f12	Correct instantiation of delayed opaques ... I didn't realize that delayed instantiation of aliases instantiates their real var, not the whole alias type. So prior to this, we were wrapping all opaques in a structural `Alias(..)`. Which is okay from a typechecking perspective because Aliases are transparent, but is wasteful.	2022-05-07 16:50:14 -04:00
Folkert	fa3eff6515	re-use a vector that already has a good capacity	2022-05-07 13:21:23 +02:00