Hide methods from PathManager which aren't supposed to be used from plugins and don't have external usages. Also, add javadoc to methods.
GitOrigin-RevId: e542995daa14eca297389465bfaa207b617a7dff
`Update`s should be decoupled from the calling code, otherwise the context of the caller would prevent the updates from merging.
GitOrigin-RevId: dae65fc3f0604d08f0dcb8427788e1886fe6605d
1. Delete `intellij.python.community.impl.xml` (move content to `PythonCore` plugin). It is necessary to fix incorrect dependencies: `PythonCore`, `Pythonid` include the same module.
2. Fixed dependencies: if some plugin/module needs Python core functionality, then it should depend on `PythonCore`.
Co-authored-by: Vladimir Koshelev <Vladimir.Koshelev@jetbrains.com>
Merge-request: IJ-MR-136158
Merged-by: Egor Eliseev <Egor.Eliseev@jetbrains.com>
GitOrigin-RevId: ce9ed4c17f601ee1ca2b6cf608f4e30fdac1d879
+ sys-property to customize indexing writer queue size/waiting time -- supposed to be used for benchmarks
GitOrigin-RevId: e48c3f74ac6d624053c7d267b44e89ec972e1e9d
1. Move WriteIntent wrapping back to application
2. Dispatch coroutines without Write Intent Lock and Java-style context propagation.
3. Don't check for WIL in wrapper, it will not be here in any case.
GitOrigin-RevId: 4ba88086589c79ce298b189ed7091d0394f34326
Java DFA can immediately tell the nullability of annotated types.
Previously it was lost likely because we don't preserve any annotations
on JKTree type arguments (KTIJ-21609).
KTIJ-29147
GitOrigin-RevId: dd8abacd2a521b8fcb2432ef3969dcf9b2a7904c
This change is needed for the upcoming generics nullability support.
1. Operating on `PsiType`s directly is not always correct
because a PsiType object can sometimes be recreated (for example, by PsiNewExpressionImpl.getType()),
and in that case we won't find it in our IdentityHashMap.
We should try to migrate this logic to PsiTypeElements, which are more persistent.
2. The changes in testdata related to varargs are expected:
In Kotlin, `vararg objs: Any?` means that `objs` is a not-null Array of `Any?` elements.
So, the nullability of the underlying JKJavaArrayType should be discarded,
like it was always done in JavaToJKTreeBuilder before the introduction
of J2KNullityInferrer. Since we don't infer nullability of vararg components,
it becomes nullable. The previous testdata was incorrect.
3. The change also revealed a problem with C-style array declarations,
which is mentioned in `PsiTypeElement#getType` JavaDoc. This corner case needs
to be supported for K2 nullability separately (a possible solution is to run
a Java pre-processing inspection to convert such declarations to Java style).
KTIJ-29147
GitOrigin-RevId: e32e9e069209f27675a68cb9613c23d710826cc1
Reading value with options as string leads to reading full option string - `[opt1|opt2|selected*|opt3]`, which is not what's expected, instead we need to read and set only a selected option
GitOrigin-RevId: 8849bff2be161a4e748efafd32877eaedcf37836
