Commit Graph

12866 Commits

Author SHA1 Message Date
Alexander Gorshenev 504bdfc283 testData for binary compatibility tests on klib library evolution 2021-01-26 03:42:15 +03:00
Steven Schäfer 3e820ff3b1 JVM: Take inline class mangling state into account for default methods 2021-01-25 18:12:01 +01:00
Dmitry Petrov 708e6914bd JVM JVM_IR hide sealed class constructors 2021-01-25 17:29:38 +03:00
Dmitriy Novozhilov f61a318c9d [Test] Migrate AbstractBytecodeTextTest to new test infrastructure 2021-01-25 17:11:21 +03:00
Dmitriy Novozhilov fd8795de99 [Test] Fix FirIdenticalChecker and update some FIR_IDENTICAL testdata 2021-01-25 17:08:32 +03:00
Dmitriy Novozhilov 98c22f9e59 [TD] Add forgotten bytecode dump 2021-01-25 17:08:32 +03:00
Dmitriy Novozhilov 5c0cfa87ba [Test] Support some jvm codegen specific configuration directives 2021-01-25 17:08:31 +03:00
Victor Petukhov c111c33950 Fix failing tests after 8bd78064be 2021-01-25 16:24:43 +03:00
Victor Petukhov f06a5321a6 Approximate definitely not-null types for type parameter's types if they are already not-null (has not-null upper bounds)
^KT-44440 Fixed
2021-01-25 14:17:58 +03:00
Simon Ogorodnik f2c1608c2b Update testData 2021-01-25 13:53:45 +03:00
Denis.Zharkov 8c30adc301 FIR2IR: Unwrap intersection overrides for intersection types 2021-01-25 12:29:31 +03:00
Denis.Zharkov 3a3d2ee3e9 FIR2IR: Fix mapping for intersection overrides of deserialized properties 2021-01-25 12:29:31 +03:00
Victor Petukhov fdf0934ade Don't generate txt dump for buildFlexibleEnhancement.kt diagnostic test 2021-01-25 12:16:13 +03:00
Victor Petukhov 8bd78064be Do substitution of a type enhancement, not only lower and upper bounds
^KT-44439 Fixed
2021-01-25 12:14:34 +03:00
Victor Petukhov 83c93aca2e Build type enhancement properly, by taking into account both bounds of the original flexible type
^KT-44420 Fixed
2021-01-25 11:36:49 +03:00
Mikhail Glukhikh 94e613dd01 FIR: support custom annotation-based type attributes 2021-01-23 10:56:42 +03:00
Mikhail Glukhikh 4cd6266bce FIR: fix suspend type serialization 2021-01-23 10:56:40 +03:00
Mikhail Glukhikh afe335e504 FIR: serialize ExtensionFunctionType attribute 2021-01-23 10:56:37 +03:00
Mikhail Glukhikh 9e5c9efadf FIR2IR: don't generate IrGetObject for anonymous objects 2021-01-23 10:56:36 +03:00
Mikhail Glukhikh 06ee768c6a FIR: in case of smart cast, use original type for FirThisRcvrExpression
Before this commit, we used type after smart cast both for original
FirThisReceiverExpression and for wrapping FirExpressionWithSmartCast.
However, this makes FIR2IR implicit cast generator work incorrectly
(it decides not to insert implicit cast because original type is the same).
After this commit, expressions have different types and implicit cast
generator works properly.
2021-01-23 10:56:34 +03:00
Mikhail Glukhikh 4fd4f504d0 FIR: make componentX functions operator 2021-01-22 18:11:57 +03:00
pyos 4a76ea6ecb JVM: regenerate objects if they have been regenerated in parent contexts
This is a hack to work around the fact that type mappings should not be
inherited by inlining contexts for lambdas called from anonymous
objects. As the lambda can call the inline function again, this could
produce a reference to the original object, which is remapped to a new
type in the parent context. Unfortunately, there are many redundant
`MethodRemapper`s between the lambda and the class file, so simply
editing `TypeRemapper` does not work. Hence, this hack. For now.

(Issue found by compiling IntelliJ IDEA BTW.)
2021-01-22 13:39:20 +01:00
Mads Ager ef36b81c67 [JVM_IR] Reduce the amount of super suffixes on accesibility bridges.
The super suffix was used for any static field/method that needed
an accessor. We should only use it when that field or method is
inherited.
2021-01-22 13:20:25 +01:00
Mikhail Zarechenskiy d2ce73853d Report more specific diagnostic for variable + invoke calls
Do not report same set of diagnostics for variable call if actual
 error was happened with a function candidate. Here the candidate is
 invoke function on DeepRecursiveFunction

 ^KT-40991 Fixed
 ^KT-41491 Fixed
 ^KT-40926 In Progress
2021-01-22 13:57:47 +03:00
Dmitriy Novozhilov c423750962 [TD] Fix incorrect module structure directives in test data file 2021-01-22 13:51:34 +03:00
Dmitriy Novozhilov b9e9620ab5 [TD] Unmute passing Fir2Ir text test 2021-01-22 13:51:32 +03:00
Dmitriy Novozhilov 842ed624a7 [TD] Remove some outdated dumps in FIR_IDENTICAL tests 2021-01-22 13:51:31 +03:00
Dmitriy Novozhilov 77115cea92 [TD] Union ir dumps of different test files 2021-01-22 13:51:30 +03:00
Dmitriy Novozhilov 4752924b13 [TD] Update FQN's in testdata of IR text tests 2021-01-22 13:51:29 +03:00
Dmitriy Novozhilov 8286a160bd [TD] Update order of directives 2021-01-22 13:51:27 +03:00
Dmitriy Novozhilov aba029237d [Test] Add handler for comparing pretty kt ir dumps 2021-01-22 13:51:25 +03:00
Dmitriy Novozhilov 6a7cd0c811 [Test] Add ability to specify applicability of diagnostic to module or file 2021-01-22 13:51:19 +03:00
Alexander Udalov e17153fab9 Minor, improve Ant test on includeRuntime
Also check that we _do_ include runtime if includeRuntime is true.
2021-01-21 18:29:19 +01:00
scaventz a13eb4c8e6 Ant task: do not include runtime by default if destination is a jar 2021-01-21 18:29:19 +01:00
Denis.Zharkov b0b7f39c75 FIR: Do not use return statement for type of a block expression
Type of a block is a kind of irrelevant for lambdas: their type is much
more complicated and defined via FirDataFlowAnalyzer#returnExpressionsOfAnonymousFunction
at at FirCallCompleter.LambdaAnalyzerImpl#analyzeAndGetLambdaReturnArguments
2021-01-21 18:53:03 +03:00
Denis.Zharkov 5afebb4e78 FIR: Transform synthetic calls children even without incomplete candidate
It might be necessary for cases like
run @l1{
   run {
      if (...) return@l1 1
   }
}

"if" is a synthetic call, but without candidate since there's just one branch
But return@l1 1 is incomplete because it's resolved in dependent context
and "1" is hang incomplete integer literal
2021-01-21 18:53:03 +03:00
Denis.Zharkov 5f0d00a83f FIR: Use proper tower data contexts for postponed callable references
E.g., it should contain local scopes of containing lambdas
2021-01-21 18:53:03 +03:00
Denis.Zharkov ad30c6c380 FIR: Fix callable references resolution when they're being returned from lambdas 2021-01-21 18:53:03 +03:00
Denis.Zharkov 0e368cc237 FIR: Analyze return statements for implicit lambda with independent context 2021-01-21 18:53:02 +03:00
Jinseong Jeon 2e660ef62a Raw FIR: use lambda parameter type if available 2021-01-21 16:06:23 +03:00
Jinseong Jeon fa0b933bc8 FIR checker: add diagnostics for missing/ambiguous component calls 2021-01-21 16:06:10 +03:00
Jinseong Jeon de592f4f67 FIR checker: introduce FunctionChecker alias
and use it to add support diagnostic FUNCTION_DECLARATION_WITH_NO_NAME
2021-01-21 16:06:09 +03:00
Jinseong Jeon f1d8a6e5d1 FIR checker: introduce DECLARATION_SIGNATURE_OR_DEFAULT positioning strategy
and fix CONFLICTING_OVERLOADS to use it
2021-01-21 16:06:09 +03:00
Igor Yakovlev 2e4daee1d4 [FIR] Fix invalid diagnostic fir node sites and improved invalid type parameters count diagnostic report 2021-01-21 15:20:30 +03:00
Ilmir Usmanov 704366e531 JVM: Mark suspend lambda receiver as used if callable reference
accesses it.
 #KT-44131 Fixed
2021-01-21 12:47:59 +01:00
Mikhael Bogdanov de00f72fa3 Generate linenumber for inline call site in case of implicit iterator/hasNext/next calls in for loop
Otherwise -1 is generated in smap
2021-01-21 12:04:09 +01:00
Ilmir Usmanov 952576e98f JVM_IR: Do not unbox Result parameter if it not only one inline class
parameter, since in this case, the compiler generates a bridge, where
the result is unboxed.
2021-01-20 18:30:00 +01:00
Ilmir Usmanov d48f92775b JVM_IR: Do not unbox Result parameter in invoke if there is a bridge
since the bridge unboxes it.

 #KT-44141 Fixed
2021-01-20 18:29:58 +01:00
Ilya Goncharov 0ddb603eaa [JS IR] Fix of test with method in abstract class calling extension on super type 2021-01-20 20:01:42 +03:00
Steven Schäfer 9c4f8f7e54 JVM IR: Fix signature mapping for inline classes using new mangling
When resolving inline class methods in binary dependencies we look for
methods matching both the new and the old mangling scheme. On the IR
backend the method signature has to be computed for the inline class
replacement, since the logic for signature mangling is not contained in
the method signature mapping, unlike in the old backend.
2021-01-20 17:43:41 +01:00