Each compiler-based LL FIR test can:
- fail in both non-reversed and reversed configurations, muted with `// MUTE_LL_FIR`
- fail in non-reversed test only, muted with `// IGNORE_NON_REVERSED_RESOLVE`
- fail in reversed test only, muted with `// IGNORE_REVERSED_RESOLVE`
`KotlinAnnotationsResolver` must see everything that is resolvable from
the current module; otherwise it will not work properly.
The previous scope was the module itself + its binary dependencies
(jars).
Now the scope will contain the module itself and ALL its dependencies,
including other modules.
^KT-63096 Fixed
The existing reference shortener does not use import alias when it
shortens a symbol. Instead, it adds a new import directive for the
symbol that is already imported. This commit updates reference shortener
to let it reuse the existing import alias rather than adding a new one:
1. When shortening a symbol, check whether the symbol is already
imported.
2. If it is already imported by an import alias, keep the symbol
reference expression and the import alias as a string together in
`ShortenCommand`.
The actual PSI update (shortening) based on the ShortenCommand is done
by IntelliJ.
^KTIJ-27205
- While `collectionInitializationInfo` unwrapped a script's top-level
declarations, `check` forgot to do it, so a script effectively had no
top-level properties in the mind of the checker.
^KT-63286 fixed
- The test data is missing initialization checks and will be fixed with
the subsequent commit.
- The test data is originally taken from
`backingField/TopLevelPropertyInitialization` and has been simplified.
- I didn't add `const val` properties to the test data (compared to the
original) due to KT-20110.
^KT-63286
This change reorders declared fields such that LLVM can pack each class
space optimally, minimizing the padding required. The guarantee is that
an objects body will take up space equal to the sum of the sizes of its
fields (inherited and declared) rounded up to a multiple of 8 bytes.
This also implies that no object will contain more than 7 bytes of
padding.
Previous changes uncovered a problem with lacking of stdlib in
dependencies, which was quite hard to debug and determine.
So this check is supposed to discover similar issues in the future
much more easily.
Setting `linkViaSignatures = false` leads to the situation when fir2ir
provides no guarantees about how SymbolTable is filled, so if some
backend relies on symbol table content (JS and Native does) then
it's required to set this flag to `true`
The issue was uncovered by previous commits because previously fir2ir
created symbols with signatures for some declarations even with
disabled flag, which was fixed by those commits
This is needed to be able to implement creation of unbound symbols
for references of corresponding declarations (KT-62856)
There was an exception from FIR2IR that was fixed with this change,
so fir2ir test SuperClass started to pass along with IrActualizer,
which reported some new errors
This is needed to be able to implement creation of unbound symbols
for references of corresponding declarations (KT-62856)
There was an exception from FIR2IR that was fixed with this change,
so fir2ir test expectDeclarationWithWeakIncompatibilities started to
pass along with IrActualizer, which reported some new errors
Types based on type parameters which have both hasEnhancedNullability
and hasFlexibleNullability were incorrectly converted to IR: only
`@EnhancedNullability` annotation was added to the IR type. This means
that the nullability of the type started to matter, which is incorrect
because the type is supposed to be flexible.
This backfired in the IR fake override builder mode, where nullability
of methods inherited from both Kotlin maps and Java maps clashed, which
resulted in the exception from FakeOverrideRebuilder.
If the type is both flexible and enhanced, we're now adding
`@FlexibleNullability` to the IR type. Note that `@EnhancedNullability`
is not needed because it only affects primitive types which need to be
handled as wrapper types on JVM, but `@FlexibleNullability` already has
that effect.
IR still slightly differs from K1 for collection subclasses, which can
be seen for example in the changed IR text tests. This is to be
investigated later.
The change in tests is needed because the `kt48495*.kt` tests are now
passing on K2 in the IR fake override builder mode, while failing on the
vanilla K2.
This is related to KT-61370 but does not entirely fix it, although
curiously it fixes all remaining tests which were muted because of it.
Ilya Kirillov