Commit Graph

5406 Commits

Author SHA1 Message Date
Ilmir Usmanov 1376fed1d4 Support non-public inline class constructors
#KT-28056 Fixed
2020-11-03 12:10:21 +01:00
Igor Chevdar 94145f880c [IR] Inliner: supported recursion in default arguments 2020-11-03 11:39:00 +05:00
Mark Punzalan 89836a2c9a Minor, test-data-only: Remove unused "IGNORE_BACKEND_FIR" directives in
asmLike test-data.
2020-11-02 19:37:14 +01:00
Alexander Udalov dd33ed9297 Fix suspend function with inline class types in reflection
#KT-34024 Fixed
2020-11-02 19:33:23 +01:00
Ilmir Usmanov 7b315a8b52 JVM_IR: Do not box inline class in methods 2020-11-02 10:50:00 +01:00
Mikhael Bogdanov e20093d762 Support @JvmStatic and @JvmOverload in annotation companion 2020-11-02 08:27:49 +01:00
Mikhael Bogdanov 7ec2d036ae Don't generate final modifier on static interface methods produced by @JvmStatic+@JvmOverloads from interface companion
#KT-35716 Fixed
2020-11-02 08:27:49 +01:00
Dmitry Petrov e4ba787034 JVM_IR KT-43066 Generate 'clone' in class implementing kotlin.Cloneable 2020-10-30 19:03:31 +03:00
Shagen Ogandzhanian 2d0535a713 [JS_IR] Invoke companion init block while instantiating a class
KT-40740, squashed rr/shagen/KT-40740-static-init-js-ir
2020-10-30 16:19:13 +01:00
Ilya Chernikov 60246f2261 Ignore temporarily the kt22029.kts codegen test for JVM IR:
since it fails and I cannot fix it right now.
2020-10-30 12:57:42 +01:00
Ilmir Usmanov 28e148d35a Old JVM: support init blocks in inline classes
#KT-28055 Fixed
2020-10-29 20:38:03 +01:00
Ilmir Usmanov 999604541e JVM_IR: Support init blocks in inline classes
Put their content to constructor-impl, so they are called during
constructor call, but they are not called during boxing, because
box-impl calls <init> and not constructor-impl.

 #KT-28055 In progress
2020-10-29 20:38:01 +01:00
Mikhael Bogdanov 5804f73ebd JVM_IR. Deprecate public access to @JvmField/const fields in private companions
#KT-25009
2020-10-29 17:19:49 +01:00
Mikhael Bogdanov 57c9afc73a Deprecate public access to @JvmField/const fields in private companions
#KT-25009
2020-10-29 17:19:49 +01:00
Dmitry Petrov 08b761ae7a PSI2IR don't copy annotations for properties implemented by delegation
KT-43012
2020-10-29 16:23:02 +03:00
Alexander Udalov ad5b6da273 JVM IR: substitute generic type for inline class replacement function calls
The main change here is in `JvmInlineClassLowering.visitFunctionAccess`,
where we now store the substituted return type of the function as the
type of the call expression. Without it, the call could have a
meaningless type, e.g. some `T` which is inaccessible at that place, and
that could backfire in subsequent lowerings in codegen. For example, in
the `stringPlus.kt` test, it would prevent the code in
`FlattenStringConcatenationLowering.isStringPlusCall` from recognizing
and replacing the `String.plus` call, leading to a codegen exception.

Other changes are mostly cosmetics to make the code similar to
`visitFunctionReference`, and preventive optimizations for the case when
the substitution map is empty.
2020-10-28 16:46:28 +01:00
Mikhail Glukhikh 2dc6467b5d [FIR] Modify signatures also from ERASED_COLLECTION_PARAMETER_SIGNATURES
In this commit we change value parameter type of containsAll, removeAll,
retainAll from Java collections. Originally it's Collection<?>,
we change it to Collection<T>

#KT-42340 Fixed
2020-10-28 18:09:11 +03:00
Denis Zharkov 923bb3b674 FIR2IR: Mute failing bb test
^KT-42946 Open
2020-10-28 11:43:25 +03:00
Denis Zharkov 2bdb21793f FIR: Adjust test data 2020-10-28 11:43:25 +03:00
Mikhail Glukhikh 09043fb98d [FIR] JvmMappedScope: add mutable methods or not depending on a class 2020-10-27 10:27:06 +03:00
Mikhail Glukhikh 5c3269f489 [FIR] JvmMappedScope: don't add Java methods if Kotlin ones are here 2020-10-27 10:26:55 +03:00
Mikhail Glukhikh caafe21e84 [FIR] Add CheckLowPriorityStage to callable reference resolve 2020-10-26 20:07:37 +03:00
Ilmir Usmanov a775fa195b Unbox inline class parameter of lambda if underlying type is Any or Any?
The inline class is boxed when we pass it as lambda argument, now we
unbox it. If the underlying type is not Any or Any?, bridge method does
the unboxing.

 #KT-32450 Fixed
 #KT-39923 Fixed
 #KT-32228 Fixed
 #KT-40282 Fixed
2020-10-26 17:34:37 +01:00
Dmitry Petrov b1629cc5f4 JVM KT-22465 don't generate accessor to private setter in other class 2020-10-26 14:56:07 +03:00
Igor Chevdar 9b2aa0951b [box-tests] Turned on tests on typeof for K/N 2020-10-23 18:27:10 +05:00
Igor Chevdar 23d12a717e [box-tests] Added a test 2020-10-23 14:22:48 +05:00
pyos e19ecdfb3d [Tests] Suspend Codegen Tests Improvements
This commit enables the execution of suspend box tests in a separate
test. It's a QoL improvement in the existing bb tests but, motivating
these changes, enables the new debugger stepping tests to step
coroutine code.
2020-10-23 11:19:58 +02:00
Ilya Goncharov 4479bf0933 [JS_IR] Enum constructor copy parameters with mapping by index
Previously mapping was by name, and there was clash in case of user
parameters named 'name' or 'ordinal'

^KT-39186 fixed
2020-10-23 11:42:43 +03:00
Mikhail Glukhikh acb03cb28c [FIR2IR] Expand type during super qualifier search 2020-10-23 08:12:16 +03:00
Mikhail Glukhikh c50aa5f2ec [FIR TEST] Bad test data changes due to Java signature transformation
Both changes are related to getOrDefault. We should either make Java
version low priority or remove it at all.
2020-10-23 08:12:16 +03:00
Mikhail Glukhikh d3e85dbce0 [FIR] Implement replacing Object with type parameters for specials builtins
#KT-42601 Fixed
2020-10-23 08:12:15 +03:00
Alexander Gorshenev 7815529eed [Private fake overrides] Tests for private fake overrides construction 2020-10-22 13:23:29 +03:00
Mikhail Glukhikh 23e7468e57 [FIR2IR] Cache Java field-based properties more correctly #KT-42805 Fixed
Before this commit, we cached such IR properties by FIR property
which was created by Java field each time when we referenced it.
This led to signature clashes. Now we cache such IR properties
directly by associated FIR field.
2020-10-22 13:05:00 +03:00
Juan Chen 620a5d404d [FIR2IR] Keep redundant cast on 'this' for local anonymous function
#KT-42517 Fixed
2020-10-22 10:51:23 +03:00
Jinseong Jeon 46cc01602e FIR2IR: add implicit NOT_NULL cast if needed 2020-10-22 10:51:20 +03:00
Dmitry Petrov b1b87becc8 PSI2IR more JVM-like exhaustive when behavior KT-36840 2020-10-21 20:07:11 +03:00
Ilya Goncharov 14d9aa1660 Add test on break in when without label
^KT-42262 fixed
2020-10-21 16:59:24 +03:00
pyos a4b67f007f JVM_IR: treat suspend-converted references as lambdas for inlining
Again, this is consistent with adapted function references, since
they're pretty much the same anyway.
2020-10-21 15:21:01 +02:00
pyos 12bec7cca2 JVM_IR: when convering references to lambdas, bind the receiver
instead of storing it into a captured temporary.

This makes the resulting lambdas equivalend to adapted function
references produced by PSI2IR or FIR2IR.

 #KT-42492 Fixed
2020-10-21 15:21:01 +02:00
Pavel Punegov ce8d983b2a Fix test after the fix for KT-17691 2020-10-21 13:58:53 +03:00
Dmitry Petrov 0b34526321 JVM_IR remove lambda hack required for odd JVM+OI behavior in KT-35849 2020-10-21 13:47:01 +03:00
Dmitry Petrov 3979cde738 Add box and PSI2IR tests for special cases listed in KT-35849 2020-10-20 19:31:40 +03:00
Dmitry Petrov 47c784f023 JVM_IR fix Java wildcard types translation
Types such as 'Collection<? extends CharSequence>' are treated as
'(kotlin.collections.MutableCollection<out kotlin.CharSequence!>
  ..kotlin.collections.Collection<kotlin.CharSequence!>?)'
by the front-end.
When generating generic signatures, JVM BE takes lower bound (which is
'kotlin.collections.MutableCollection<out kotlin.CharSequence!>').
Emulate this behavior in TypeTranslator.
2020-10-19 20:52:53 +03:00
Ilmir Usmanov 819d64a2ef Box inline class returned from suspend lambda non-locally in JVM_IR
#KT-41194 Fixed
2020-10-19 14:28:39 +02:00
Ilmir Usmanov 70a4ed3ebc Box inline class returned from suspend lambda non-locally
All inline classes should be boxed coming in and out of lambdas,
however, if the inline class was returned non-locally, it was not boxed.
This change fixes the issue in Old JVM BE.
 #KT-41194 In progress
2020-10-19 14:28:36 +02:00
Mads Ager 0505bd958a [JVM] Fix the nop removal optimization.
Never remove a nop if that would cause a line number to move
across a local lifetime boundary.

This fixes KT-42725.
2020-10-16 19:21:08 +03:00
Roman Artemev d7db643422 Add box test for KT-33992 2020-10-16 16:02:02 +03:00
Roman Artemev 023a62395a Add box test for KT-36897 2020-10-16 16:02:01 +03:00
pyos a9b53adc50 JVM_IR: make primitive == object slightly less lazy.
Discarding the value used to leave an unused-but-never-destroyed
temporary variable. It's best to not separate calls to `enterTemp`
and `leaveTemp`.

Not sure what kind of test to add though, since this is minor -- if the
result of the comparison is discarded, then the entire statement is more
or less pointless.

 #KT-42251 Fixed
2020-10-16 12:25:55 +02:00
Dmitry Petrov ee5edf4caa JVM_IR fix 'remove' in inline class implementing MutableCollection 2020-10-16 00:02:12 +03:00