azalea/kotlin-fork
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
60,697 Commits 1 Branch 0 Tags
437a26684d3529ee2cfdbe54e59d50f4a6f0a611
Commit Graph

595 Commits

Author SHA1 Message Date
Victor Petukhov 437a26684d NI: Prefer nullable lower bound to flexible one when substitution of type variable is performed and remember flexibility of type parameters based on flexibility of its upper bounds 
^KT-32435 Fixed
 2020-01-17 19:16:09 +03:00
Mikhail Zarechenskiy 4f74515508 [NI] Fix CST calculation for covariant type projections 2020-01-16 17:18:20 +03:00
Pavel Kirpichenkov fd1305c369 [NI] Refactor OnlyInputType annotation check 
Hold position consumed during incorporation in Constraint instead of ConstraintPosition.
 2020-01-16 12:01:07 +03:00
Pavel Kirpichenkov 0d393e3fd7 [NI] Update filtering of constraints with equal types 
Make check for synthetic nullability constraint status before skipping
new constraint with nonunique type.
 2020-01-16 12:01:06 +03:00
Pavel Kirpichenkov 2896642f94 [NI] Consider synthetic lower nullability constraint non-proper 2020-01-16 12:01:06 +03:00
Denis Zharkov 676c99b933 NI: Fix exception during callable references overload resolution 
^KT-35847 Fixed
 2020-01-16 10:49:49 +03:00
Dmitriy Novozhilov a3037a081e [NI] Fix resolve of lambdas with expected function type with receiver 
#KT-30245
 2020-01-15 22:12:29 +03:00
Dmitriy Novozhilov 04ce10b6c1 [NI] Improve completing callable references with type variable as expected type 
#KT-32462 Fixed
 2020-01-15 22:12:28 +03:00
Denis Zharkov c48539feb3 NI: Fix callable references resolution when LHS is generic nested class 
In case of null qualifier, we should not look into any static scope
NB: factory::createCallableProcessor returns NoExplicitReceiver processor
in case of null-receiver, that makes resolving the call in the test as
`property(::key)` that matches to the property itself, thus leading to
overload resolution ambiguity

^KT-35887 Fixed
 2020-01-15 16:10:34 +03:00
Dmitriy Novozhilov 3428a17759 [NI] Postpone calls with not enough information in builder inference 2020-01-15 14:01:25 +03:00
Denis Zharkov d6b01f1007 NI: In subtyping do not intersect arguments for non-covariant types 
It partially reverts 7898922066
because it's not obvious that it's a safe operation
for invariant/contravariant types.

Also, there's a necessary fix in prepareReceiverRegardingCaptureTypes
to make types order stable
Otherwise test bareTypesWithStarProjections becomes flaky.

Also, the changes in bareTypesWithStarProjections.kt are also expected
because the type of the expression `coneSymbol` after the second "if" is
FirVariableSymbol<*> & FirPropertySymbol & AbstractFirBasedSymbol<*>
thus we fix D in the call `coneSymbol.phasedFir()` to FirVariableSymbol<*>
because it's the first type in the list
(see the next line after the last changed in AbstractTypeChecker)
 2020-01-14 17:35:24 +03:00
Denis Zharkov 4202c9c1a4 NI: Fix regression for star-projections approximation 
See the test added

^KT-35703 Fixed
 2020-01-14 17:35:24 +03:00
Pavel Kirpichenkov 4a29de3716 [NI] discriminate Nothing for reified parameters 
Related issues: KT-32836, KT-35728
 2020-01-14 16:27:42 +03:00
Mikhail Zarechenskiy 6fe61c9056 [NI] Fix ambiguity when there are no applicable candidates 
#KT-35064 Fixed
 2020-01-14 14:30:01 +03:00
Ilya Chernikov caa677e6d2 [NI] Convert extension lambda to the non-extension one, if needed 
#KT-30245 fixed
 2020-01-09 08:08:45 +01:00
Pavel Kirpichenkov 2d21b82501 [NI] Remove hack for special functions 
Treating special functions for `if`, `when`, `try`, `?:` as not accepting `Nothing` result type is incorrect.
Making so leads to cases with uninferred `Nothing` result type for inner calls and lost data flow info.
 2019-12-27 18:40:42 +03:00
Dmitriy Novozhilov 5dfe100ae5 Allow use reference to reified type parameters in contracts since 1.4 2019-12-26 17:43:28 +03:00
Denis Zharkov 0b385cc532 NI: Approximate captured-for-star-projections to * 
^KT-35602 Fixed
 2019-12-26 12:35:32 +03:00
Dmitriy Novozhilov c8a1b876e8 [NI] Fix completion of error delegates with receiver with type parameters 2019-12-26 10:54:07 +03:00
Dmitriy Novozhilov 0c01499d98 [NI] Improve completing calls with multiple postponed arguments 
#KT-27999 Fixed
#KT-30244 Fixed
#KT-31102 Fixed
 2019-12-26 10:53:03 +03:00
Victor Petukhov 4309f53e11 NI: don't filter flexible types during CST calculation and force return as result the first of such types 
^KT-35658 Fixed
 2019-12-25 18:00:23 +03:00
Dmitriy Novozhilov 51424ec598 [NI] Add constraint K <: T from constraint K <: CapturedType(in T) 
#KT-33263 Fixed
 2019-12-25 16:19:19 +03:00
Dmitriy Novozhilov b54169d312 [NI] Fix substitution in completion of callable references in coroutine inference 2019-12-25 16:19:19 +03:00
Pavel Kirpichenkov 7ee71450bc [NI] Discriminate Nothing? result type from nullability constraints 
Nullability constraints should not be chosen when proper
argument constraints for variables in contravariant position present.

^KT-32106 Fixed
^KT-33166 Fixed
 2019-12-25 14:59:06 +03:00
Pavel Kirpichenkov a1ad680edb [NI] Disallow full completion based on lower Nothing constraint 2019-12-25 14:59:06 +03:00
Pavel Kirpichenkov 384bd858e9 [NI] Support fixation to subtype in completion mode calculator 
Allow full compltion mode for return type depending on type parameters in
contravariant position only if they have proper equality constraint.
 2019-12-25 14:59:05 +03:00
Pavel Kirpichenkov e69e45e2a6 [NI] Remove direcion calculator from variable fixation 
Force constraint system completer to always fix variable with unknown position.
This makes inferred type for in parameters less specific, but OI infers to
subtype in such cases, so NI won't infer to less specific types than before.
 2019-12-25 14:59:05 +03:00
Pavel Kirpichenkov e9941f8c12 [NI] Implement algorithm for completion mode selection 
Current selection of completion mode for call is not always correct in case of full mode,
and sometimes too conservative in case of partial mode. Updated algorithm checks constraints
wrt position of type variables in return type and in other related constraints.
Full completion happens if proper constraint requirements are satisfied for variables.
 2019-12-25 14:59:05 +03:00
Pavel Kirpichenkov a9391c8dfb [NI] Remove Nothing result type restriction in most cases 
Make Nothing as result type not suitable only for if, when, try and ?: special functions.
 2019-12-25 14:59:05 +03:00
Mikhail Zarechenskiy ae1630f376 [NI] Don't discriminate Nothing-type from resulting ones 
^KT-32106 Fixed
 2019-12-25 14:59:04 +03:00
Dmitriy Novozhilov e730965bc5 [NI] Approximate intersection type in type argument to star if it's necessary 
#KT-32196 Fixed
 2019-12-20 11:33:06 +03:00
Dmitriy Novozhilov e466fd5196 [NI] Infer type variable to Nothing if all upper constraints are from upper bounds 
#KT-32196 Fixed
 2019-12-20 11:17:47 +03:00
Ilya Chernikov 45e881f03f [NI] Fix inference if inner system has only by "uninteresting" constraint 
#KT-32250 fixed
 2019-12-19 11:20:10 +01:00
Denis Zharkov e8131d6e30 Choose Java field during overload resolution with a pure Kotlin property 
^KT-31244 Fixed
 2019-12-19 10:53:31 +03:00
Dmitriy Novozhilov ee36fb903f Allow use expression of array type as named argument for vararg 2019-12-19 10:22:44 +03:00
Denis Zharkov c486fa6189 NI: Report incompatible receiver of callable reference 
^KT-35535 Fixed
 2019-12-18 19:33:55 +03:00
Denis Zharkov dd56c3d918 NI: Fix property references overload ambiguity 
^KT-35213 Fixed
 2019-12-18 19:33:55 +03:00
Alexander Gorshenev dc8240c24e Expect/actual support in klibs under -Xklib-mpp 2019-12-18 19:29:56 +03:00
Pavel Kirpichenkov 86dc0925b1 [NI] Preserve constraint position for OnlyInputType during incorporation 
Variable with complex dependency on other variable loses its
original constraint position when variable it depends on is fixed.
To prevent that, original input position is saved in incorporation
position.
 2019-12-18 16:19:33 +03:00
Denis Zharkov 2aa1c40de4 Avoid checking trivial incorporated constraints 2019-12-17 15:06:17 +03:00
Denis Zharkov b53c00cf69 FIR: Optimize simple things in inference 2019-12-17 15:06:17 +03:00
Denis Zharkov 0064429339 Avoid using TypeSystemInferenceExtensionContextDelegate in FIR 
It's only needed in old FE to avoid clashes when initializing DI

The main idea is gettind rid of intermediate interfaces because
each of them adds another intermediate DefaultImpls call
 2019-12-13 14:39:06 +03:00
Victor Petukhov 07269661b4 NI: Allow to resolve to functions with SAM conversion and passing array without spread as vararg (with warning) 
^KT-35224 Fixed
 2019-12-12 23:46:16 +03:00
victor.petukhov 71cb65c064 NI: Add flexible Nothing to trivial constraints to filter it 2019-12-12 18:45:11 +03:00
victor.petukhov 2d5a05466d NI: Fix some unwanted exclusions constraints with remained corresponding optimization 2019-12-12 18:45:07 +03:00
victor.petukhov a92f58b1ac Revert "NI: Prefer nullable lower bound to flexible one when substitution of type variable is performed ^KT-32435 Fixed" 
This reverts commit 09c2b92b
 2019-12-11 15:32:46 +03:00
Victor Petukhov 09c2b92b1d NI: Prefer nullable lower bound to flexible one when substitution of type variable is performed 
^KT-32435 Fixed
 2019-12-11 12:15:02 +03:00
Dmitriy Novozhilov b76c984b26 [FIR] Support IntegerLiteralTypes 2019-12-09 17:24:26 +03:00
Denis Zharkov 5b4e8170ae Fix variables leaking when adding other system under transaction 2019-12-04 10:52:19 +03:00
Denis Zharkov 692442d7a2 Do not look into captured star projection as types 2019-12-04 10:52:19 +03:00