azalea/kotlin-fork
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
69,619 Commits 1 Branch 0 Tags
bf35818438bf2da42576e5fba28229f817ce3bac
Commit Graph

5393 Commits

Author SHA1 Message Date
Mikhael Bogdanov bf35818438 Minor. Reformat 2020-10-02 11:47:27 +02:00
Dmitry Petrov 565c156ddb Minor: reformat and cleanup warnings in CollectionStubMethodGenerator.kt 2020-09-28 21:48:14 +03:00
Dmitriy Novozhilov 0d954f1190 Change module :core:compiler.backend.common.jvm to :compiler:backend.common.jvm 2020-09-24 16:51:06 +03:00
Dmitriy Novozhilov 7380abac13 Split AsmUtils to descriptors dependent and independent parts 2020-09-24 16:51:05 +03:00
Dmitriy Novozhilov 54a5a066e8 Introduce new module for common parts of backend utils 
Mainly it needed for abstract jvm type mapper for different type systems:
  IrTypes and ConeKotlinTypes
 2020-09-24 16:51:05 +03:00
Dmitriy Novozhilov c3a89e2507 Move common part of typeSignatureMapping.kt to :core:compiler.common.jvm 2020-09-24 16:51:05 +03:00
Dmitriy Novozhilov d215c413cc Move TypeMappingMode to :core:compiler.common.jvm 2020-09-24 16:51:05 +03:00
Ilmir Usmanov f22f10febb JVM_IR: Unbox inline classes in suspend functions 
if the function can be verified to safely returning unboxed
inline class.
Box the return value on resume path.
 2020-09-24 10:08:41 +02:00
Steven Schäfer 5e27d9b089 JVM IR: Make inlining in $default stubs compatible with the JVM BE 
The JVM BE inlines calls to the underlying function in a $default stub
verbatim, e.g., without renaming LVT entries or regenerating anonymous
objects. This commit introduces the same behavior in the JVM IR BE.

Fixes KT-36769.
 2020-09-24 07:53:31 +02:00
Mikhael Bogdanov 7d4a734791 Fix nullability of typePath parameter 
#KT-41484 Fixed
 2020-09-22 15:37:02 +02:00
Ilmir Usmanov b406022315 Mark Unit unspillable if it is merged with unspillable Unit 
#KT-42004: Fixed
 2020-09-17 23:23:14 +02:00
Ilmir Usmanov 9ecf5dc9af Support is/as operations on suspend callable reference conversion 2020-09-17 16:14:04 +02:00
Ilmir Usmanov f960201f52 Ignore type parameters of inline class 
when checking whether we can return unboxed inline class
from suspend function
 2020-09-17 11:20:48 +02:00
Ilmir Usmanov 58146c4452 Keep DebugMetadata annotation when regenerate lambda with state-machine 
When a suspend lambda does not capture crossinline lambda, it is
generated with as state-machine, since it does not inline anything.
However, when regenerating, the inliner used to remove all DebugMetadata
annotations to avoid duplication. This lead to missing annotation if
the lambda is regenerated, but state-machine is not regenerated.
This change fixes the blind spot by readding the annotation after
regeneration.
 #KT-41789 Fixed
 2020-09-15 19:57:02 +02:00
Igor Yakovlev c3810dc223 [ULC] Fixed UL backing field initializers 2020-09-11 12:49:00 +03:00
Ilmir Usmanov 52f9569d33 Generate CHECKCAST Object inside the markers 
otherwise, the unboxing interferes with bytecode analysis.
 2020-09-09 17:43:42 +02:00
Ilmir Usmanov e8a451072e Minor. Add tests with same JvmType in covariant override 2020-09-09 17:43:41 +02:00
Ilmir Usmanov 1c97eafea8 Check for COROUTINE_SUSPENDED inside callable reference 
#KT-41429 Fixed
 2020-09-09 17:43:37 +02:00
Ilmir Usmanov 4303e8126f Do not generate suspend markers inside callable reference 
#KT-41429
 #KT-25519 Fixed
 2020-09-09 17:43:36 +02:00
Ilmir Usmanov ccc5b7afe0 Box inline class in resume path of suspend call 
#KT-41429
 2020-09-09 17:43:35 +02:00
Ilmir Usmanov 023cdd7cd4 Support not boxing inline class of reference class 
even if the class is other than Any.
 #KT-41429 In progress
 2020-09-09 17:43:34 +02:00
Alexander Udalov 3d7619421f JVM IR: fix exception on star projection in type parameter upper bound 
#KT-41761 Fixed
 2020-09-09 17:07:52 +02:00
Alexander Udalov 669fda6b77 Do not add duplicate $completion when compiling JVM against JVM_IR 
This change fixes JvmOldAgainstIrBoxInlineTestGenerated.Suspend and
JvmIrAgainstOldBoxInlineTestGenerated.Suspend tests.
 2020-09-07 19:41:38 +02:00
Dmitry Petrov b00ce872ea JVM: record JVM signature for equals/hashCode/toString in inline classes 2020-09-07 12:25:50 +03:00
Dmitriy Novozhilov d1fd1da56f Rename Visibility to DescriptorVisibility 2020-09-04 11:07:42 +03:00
Dmitriy Novozhilov 64ec3fc42b Move BuiltinSpecialProperties and JvmAbi to :core:compiler.common.jvm 2020-09-04 11:07:41 +03:00
Dmitriy Novozhilov 10fd164b77 Extract FE 1.0 dependent part from JavaToKotlinClassMap 
Also move JavaToKotlinClassMap to :core:compiler.common.jvm
 2020-09-04 11:07:41 +03:00
Dmitriy Novozhilov 4d5f28d5c0 Move FunctionInvokeDescriptor.BIG_ARITY to :core:compiler.common 2020-09-04 11:07:41 +03:00
Dmitriy Novozhilov bf4bf30b4a Move CompanionObjectMapping to :core:compiler.common 2020-09-04 11:07:40 +03:00
Florian Kistner 9775a2148a 203: Fix compilation 2020-09-02 18:48:57 +02:00
Alexander Udalov 5e48be3d11 Add a fallback flag -Xno-unified-null-checks for KT-22275 
#KT-41482 Fixed
 2020-08-31 12:43:28 +02:00
Alexander Udalov 74c6d2b951 Do not generate non-standard compareTo as primitive comparison in all backends 
Previous changes related to this in the old JVM backend were in
582b1c5e66 and
0482f7e9c5, but they did not affect the
`ProperIeee754Comparisons` mode which became the default in 1.4.0. As a
result, we had a regression here.

Since the `PRIMITIVE_NUMERIC_COMPARISON_INFO` slice is used in psi2ir to
determine how to generate the comparison, this fixes the regression both
in the old JVM backend, and in all IR backends.

 #KT-41426 Fixed
 2020-08-27 19:00:11 +02:00
Alexander Udalov a9ddf02556 Replace deprecated usages of max/min with maxOrNull/minOrNull 2020-08-25 20:33:59 +02:00
Dmitriy Novozhilov a764732020 Rename KotlinBuiltInsNames to StandardNames 2020-08-25 10:41:33 +03:00
Dmitriy Novozhilov 20a2ad8223 Move FunctionClassDescriptor.Kind to :core:descriptors.common module 2020-08-25 10:31:36 +03:00
Dmitriy Novozhilov 7a7fe77b8e Move static constants with builtin names to :core:descriptors.common module 2020-08-25 10:31:36 +03:00
Dmitriy Novozhilov 9d9f9c52c0 Extract some classes from descriptors module to :core:common 
This is needed to remove dependencies from fir modules to
  `:core:descriptors` module

What was extracted:
- Modality
- ClassKind
- org.jetbrains.kotlin.name package
 2020-08-25 10:31:35 +03:00
Dmitry Petrov a2dabe11c5 JVM KT-41150: Fix backward compatibility for inline vals in inline class 
In 1.3.x, for inline class member inline vals 'getFoo-impl' method was
generated in corresponding inline class.
Since 1.4.0, getters for properties returning inline class values are
mangled (so corresponding getters are named 'getFoo-<mangling_hash>'.
However, to maintain backward compatibility with libraries compiled with
1.3.x, inliner should be able to find 'getFoo-impl' method in the
bytecode.
 2020-08-24 16:17:12 +03:00
Alexander Udalov bf0bb9e946 JVM IR: remove obsolete code for duplicate signature diagnostics 2020-08-20 14:58:36 +02:00
Alexander Udalov a21f273570 Fix compiler warnings in compiler code 2020-08-17 21:18:20 +02:00
Ilmir Usmanov 31e4ddafd1 Minor. Add test with boolean parameter 2020-08-13 15:52:53 +02:00
Ilmir Usmanov bcbb050326 Use fields for spilled variables for lambda parameters as well 
In other words, do not generate p$ fields if we can use L$ fields,
which are being cleaned up.
 #KT-16222 Fixed
 2020-08-13 15:52:51 +02:00
Ilmir Usmanov 00bf07fc41 Force boxing kotlin.Result return type of suspend functions 
#KT-40843 Fixed
 2020-08-11 22:10:47 +02:00
Mikhael Bogdanov 607f99ed3c Don't generate implicit overrides delegating to Java defaults 
#KT-40920 Fixed
 2020-08-10 20:03:23 +02:00
Mikhail Zarechenskiy 2eb17df962 Do not compute SamType too eagerly as it can force member resolution 
Internally, during creation of `SamType` the compiler checks if a
 parameter is really can be used as a SAM, so this check involves
 lookup for abstract members, which can force additional computation of
 member scopes. This is totally fine for BE but it can be dangerous
 for light classes computation when not all scopes are processed.

 Note that this issue appeared quite recently after 16f175612e

 #KT-40869 Fixed
 2020-08-06 09:20:56 +03:00
Ilmir Usmanov 68342a1f72 Initialize fake inliner variables on resume path 
Otherwise, this will cause R8 to complain.
 2020-08-05 19:47:55 +02:00
Ilmir Usmanov bb5a99ec18 Do not put $completion to LVT if is dead 2020-08-05 19:47:54 +02:00
Victor Petukhov f45de9d8fb NI: approximate not top-level captured types during code generation 
^KT-40693 Fixed
 2020-08-05 10:36:36 +03:00
Ilmir Usmanov bbd4c21595 Cleanup spilled variables 
#KT-16222
 2020-08-04 15:42:37 +02:00
Ilmir Usmanov 999b41068b Minor. Refactor variables spilling 
Split information collection and method node mutation.
 2020-08-04 15:42:37 +02:00