Folkert
9f7c7b56a1
use ReferenceMatrix in DefOrdering
2022-04-22 15:49:32 +02:00
Folkert
0c10fa31f5
rename
2022-04-22 15:41:21 +02:00
Folkert
e40e4ae846
rename
2022-04-22 15:17:43 +02:00
Folkert
8837eb2c19
remove old def sorting code
2022-04-22 15:09:45 +02:00
Folkert
c763d51551
use new sccs in error case too
2022-04-22 15:04:16 +02:00
Folkert
887acb7519
be smarter about when a function is recursive
2022-04-22 14:41:42 +02:00
Folkert
8c08c63151
faster strongly-connected components
2022-04-22 14:10:46 +02:00
Folkert
45567a591c
rename
2022-04-22 13:42:57 +02:00
Folkert
141f88365d
use extra bitvec to spot faulty recursion
2022-04-22 12:56:50 +02:00
Folkert
b6ccd9c8fb
use bitvec-based topological sort
2022-04-22 11:35:08 +02:00
Folkert de Vries
2cc8e95198
Merge pull request #2901 from rtfeldman/vec-map
...
VecMap
2022-04-21 09:16:38 +02:00
Ayaz Hafiz
4bbc1d3a2b
Clippy
2022-04-20 17:46:19 -04:00
Ayaz Hafiz
b9f79fdd31
Able variables through different functions compile
2022-04-20 17:43:17 -04:00
Ayaz Hafiz
a07323fb40
Typecheck annotations with able variables outside ability members
2022-04-20 17:43:16 -04:00
Folkert
e87ca7e4b7
create and union fewer Output's
2022-04-20 21:20:59 +02:00
Folkert
9d17a075d9
halve the number of lookups into References
2022-04-20 20:22:52 +02:00
Folkert
ec43d7d770
clippy
2022-04-20 18:05:06 +02:00
Folkert
b92c28b237
make calls private
2022-04-20 17:43:18 +02:00
Folkert
c531191e49
make value_lookups private
2022-04-20 17:34:08 +02:00
Folkert
e740bbe529
make some of References' fields private
2022-04-20 17:25:22 +02:00
Folkert
8b144c446d
remove PartialEq for a bunch of types that we should not compare
2022-04-20 16:58:30 +02:00
Folkert
b59d33a1d5
refactor roc_collections
2022-04-19 23:00:05 +02:00
Ayaz Hafiz
f129777115
Explicitly disallow ability definitions in nested scopes
...
Abilities can only be defined on the toplevel of a module. There is a
technical reason to this, which is that during type solving we must
introduce all abilities at the very beginning, and we need to make sure
ranks are correct. But there is a practical reason as well, which is
that nested ability definitions don't seem to be very useful.
Note that specializations can be nested, and are allowed to be. Also, we
can revisit this in the future. I just don't want experiments to break
right now because someone uses an ability in a nested scope where we
don't expect.
Closes #2878
2022-04-18 18:04:46 -04:00
Folkert
7f3ca4458a
use VecSet in IntroducedVariables
2022-04-17 20:50:41 +02:00
Folkert
dda4f46e67
use VecSet in References
2022-04-17 19:37:10 +02:00
Ayaz Hafiz
6b4294307f
Start saving ability signature on designated variable again
2022-04-14 16:22:54 -04:00
Ayaz Hafiz
0792ccbbe8
Calling ability with non-specialized type is an error
2022-04-13 11:49:33 -04:00
Ayaz Hafiz
d110fbde6c
Annotation specializes ability
2022-04-13 10:44:34 -04:00
Ayaz Hafiz
25b3aa7ca7
Typed body specializations solve
2022-04-13 10:42:46 -04:00
Ayaz Hafiz
389c46edcf
Remove unneeded var
2022-04-13 10:12:45 -04:00
Ayaz Hafiz
5e1ab8225e
Report when ability member binds >1 variable to parent
2022-04-13 10:11:24 -04:00
Ayaz Hafiz
cbbbb8c855
Remove stray dbg
s
2022-04-12 16:32:50 -04:00
Ayaz Hafiz
15a040ec87
Basic type inference and solving for abilities
...
Note that is still pretty limited. We only permit opaque types to
implement abilities, abilities cannot have type arguments, and also no
other functions may depend on abilities
2022-04-12 16:18:07 -04:00
ayazhafiz
94a5cd3559
Improve shadow errors
2022-04-07 22:38:20 -04:00
Ayaz Hafiz
73bfff699f
Handle symbols that shadow ability member definitions
...
Just add the shadowing symbol for now. We'll handle checking that a
specialization's type matches the member's type definition in a later
pass, during typechecking.
2022-04-07 17:21:17 -04:00
Ayaz Hafiz
66ec1b4a84
Basic canonicalization and error checking for abilities
2022-04-07 17:21:17 -04:00
Ayaz Hafiz
d3b848fa46
Clippy
2022-04-06 22:24:14 -04:00
Ayaz Hafiz
a3ac68a41f
Split up Defs into TypeDef and ValueDef
...
Just a refactoring PR. This is useful because during canonicalization
we always process type defs first, then value defs. With abilities this
distinction continues to grow; in that case, we have patterns associated
with types that we want to process before patterns from values.
2022-04-06 22:18:57 -04:00
Folkert
0dd2cec09a
update with new clippy lints
2022-03-31 19:34:51 +02:00
Folkert
7029717cd0
fix missing import
2022-03-21 23:46:25 +01:00
Folkert
725eb217d8
more vecs, less sets
2022-03-21 23:20:57 +01:00
Folkert
14b53c0ccf
simplify local_successors
2022-03-21 23:16:07 +01:00
Folkert
cee1a787c9
and remove everything because it has no effect
2022-03-21 23:07:27 +01:00
Folkert
1b1a7b0385
more cleaned up
2022-03-21 22:54:24 +01:00
Folkert
a170f461e0
cleanup
2022-03-21 22:44:14 +01:00
Folkert
cc9873eb6d
exploration
2022-03-21 22:41:32 +01:00
hafiz
3f07afe3b5
Merge pull request #2750 from rtfeldman/delay-instantiating-aliases-fix-limitations
...
Delay instantiating aliases fix limitations
2022-03-19 20:41:14 -04:00
Folkert
4e315797b0
expose ImMap entry
2022-03-19 00:00:56 +01:00
Folkert
9b63e6a3a9
simplify IntroducedVariables
2022-03-18 23:47:39 +01:00
Folkert
d6f0005fb3
bubble up lambda sets used by inner aliases
2022-03-18 20:04:07 +01:00