505 commits

Author	SHA1	Message	Date
Folkert	a37a895016	optimize tag name cache	2022-03-06 14:11:33 +01:00
Folkert	7ad55d67e2	Make type_to_variable manually tail-recursive ... by using a work stack and reserving variables. fun stuff	2022-03-06 13:15:50 +01:00
Folkert	4806dc378f	add comment and assert back	2022-03-05 23:49:21 +01:00
Folkert	0351b9fce7	improvements to adjust_rank	2022-03-05 23:45:07 +01:00
Folkert	97d9027271	improve pool_to_rank_table	2022-03-05 23:44:57 +01:00
Folkert	c7c9a90d65	restructure how def_types are stored in Constraints	2022-03-05 21:47:49 +01:00
hafiz	cad02d878c	Merge pull request #2645 from rtfeldman/solve-fully-tail-recursive ... make solve fully tail recursive	2022-03-05 14:35:39 -05:00
Folkert	1b66247303	renaming	2022-03-05 19:30:15 +01:00
Folkert	fc4212310f	Merge remote-tracking branch 'origin/trunk' into solve-fully-tail-recursive	2022-03-05 18:34:09 +01:00
Folkert	80956061dd	fix formatting bug in type pretty-print	2022-03-05 15:03:59 +01:00
Folkert	6370a80c62	make sure lambda sets within aliases are in IntroducedVariables	2022-03-05 14:29:34 +01:00
Folkert	d158702e1a	Merge remote-tracking branch 'origin/trunk' into take-scratchpad-bug	2022-03-05 02:15:46 +01:00
Folkert	8c321c1aa4	fix a bug with taking a scratchpad arena	2022-03-05 02:12:10 +01:00
Folkert	a79f6c6cdd	restructure and add comments	2022-03-05 01:29:21 +01:00
Folkert	c7d8ae6c79	make complex case tail-recursive	2022-03-05 01:02:27 +01:00
Folkert	3a1add6ce8	add LocalDefVarsVec::from_def_types	2022-03-05 00:53:27 +01:00
Folkert	41df04184e	make the simple case tail-recursive	2022-03-05 00:26:03 +01:00
Folkert	4ca9ea0b89	refactor After	2022-03-04 23:57:41 +01:00
Folkert	38d3d3169a	drop final suffixes	2022-03-02 21:30:38 +01:00
Folkert	546afc9661	drop soa suffix	2022-03-02 21:21:37 +01:00
Folkert	0eb98a4c59	move over constraint	2022-03-02 21:19:58 +01:00
Folkert	c52029c2d1	the debugging begins	2022-03-02 20:30:42 +01:00
Folkert	fe48bdf5b1	make solve that uses new constraint	2022-03-02 19:43:03 +01:00
Folkert	19d7f7ce09	make vars-by-symbol a vector	2022-03-01 21:51:25 +01:00
Folkert	8b457a56c5	reduce cloning of Env	2022-03-01 00:08:56 +01:00
Folkert	d3bbf6d504	struct of arrays!	2022-03-01 00:00:50 +01:00
Folkert	1ef9546391	use Vec instead of MutMap	2022-02-28 23:54:35 +01:00
Folkert	9d82f795b7	make it abstract	2022-02-28 23:43:58 +01:00
Folkert	751ae125a5	remove aliases from solve Env	2022-02-28 23:41:07 +01:00
Folkert	c18befeccf	short-circuit aliases	2022-02-28 23:37:33 +01:00
ayazhafiz	01da425851	Simplify solve worklist substantially	2022-02-27 16:41:31 -05:00
ayazhafiz	56d4dc3480	Transition solve to use a worklist rather than always recursing ... This should unblock some of @folkertdev's work on transitioning builtins to pure Roc code. It's not perfect yet (we still make some recursions), but it's a step, and it should avoid recursing on `Let` `ret_constraint`s, which caused stack overflows when they became really long before.	2022-02-27 15:51:19 -05:00
ayazhafiz	07b1829732	Improve error reporting for patterns not matching opaques	2022-02-27 00:11:11 -05:00
ayazhafiz	059c324268	Error reporting for type mismatches involving opaques	2022-02-27 00:10:12 -05:00
ayazhafiz	b6d7229525	Infer + checking tests for opaques	2022-02-27 00:10:12 -05:00
hafiz	ca9ecbea52	Merge pull request #2533 from rtfeldman/abilities! ... [WIP] Spike for abilities and friends	2022-02-21 23:18:40 -05:00
ayazhafiz	90de82e295	Validation of opaques during canonicalization	2022-02-21 18:25:19 -05:00
ayazhafiz	6b53692aac	Canonicalize opaque types	2022-02-20 13:47:01 -05:00
ayazhafiz	d90915a8cd	Implement Num.to* builtins ... Just wrap over Num.intCast	2022-02-19 11:28:41 -05:00
ayazhafiz	909fae5b6c	Generalize recursion variables properly ... Closes #2379 Closes #2481	2022-02-18 00:07:38 -05:00
Folkert	04adbe75ca	fix test compilation	2022-02-14 21:09:51 +01:00
ayazhafiz	8c0e39211d	Instantiate recursive aliases to their smallest closures ... Now, when we have two aliases like ``` T a : [ A, B (U a) ] U a : [ C, D (T a) ] ``` during the first pass, we simply canonicalize them but add neither to the scope. This means that `T` will not be instantiated in the definition of `U`. Only in the second pass, during correction, do we instantiate both aliases **independently**: ``` T a : [ A, B [ C, D (T a) ] ] U a : [ C, D [ A, B (U a) ] ] ``` and now we can mark each recursive, individually: ``` T a : [ A, B [ C, D <rec1> ] ] as <rec1> U a : [ C, D [ A, B <rec2> ] ] as <rec2> ``` This means that the surface types shown to users might be a bit larger, but it has the benefit that everything needed to understand a layout of a type in later passes is stored on the type directly, and we don't need to keep alias mappings. Since we sort by connected components, this should be complete. Closes #2458	2022-02-11 08:43:33 -05:00
ayazhafiz	0d241f3c3c	Only expand ErrorType::Range's type range when range unification fails	2022-02-06 15:04:13 -05:00
ayazhafiz	680bf8e0b7	Treat rigids as flex vars when checking number range types	2022-02-06 15:04:12 -05:00
ayazhafiz	c5d918e68c	Include floats in bounds for unspecified numbers	2022-02-06 15:04:12 -05:00
ayazhafiz	8dc92ccd97	Second pass	2022-02-06 15:04:12 -05:00
ayazhafiz	5e0d90ac53	First pass	2022-02-06 15:04:12 -05:00
ayazhafiz	df8113ce32	Typecheck numeric suffixes in patterns	2022-02-01 23:35:14 -05:00
ayazhafiz	e03592930f	Typecheck numeric literals with suffixes in expressions ... Part of #2350	2022-02-01 22:49:50 -05:00
ayazhafiz	d10eb0f9a3	Fix Apply usage	2022-01-31 23:00:37 -05:00