- 28 3月, 2021 1 次提交
-
-
由 Mark Punzalan 提交于
in ExternalFunChecker. This allows RemoveModifierFix to provide a quickfix to remove it.
-
- 27 3月, 2021 10 次提交
-
-
由 Nikolay Krasko 提交于
Fix after 998a65d1
-
由 Tianyu Geng 提交于
It seems unnecessary to force all quickfixes to fit the `HLQuickfix` API, especially for those existing trivial fixes that simply modifies PSI tree. This change further loosen the API of `HLDiagnosticFixFactory` to allow it to create arbitrary `IntentionAction` objects. This also avoids code duplication (for example, specify the family name again in `ReplaceCallFixFactories`). `HLQuickfix` and the input/target paradighm can still be used where applicable.
-
由 Tianyu Geng 提交于
It looks like `project` can never be null
-
由 Tianyu Geng 提交于
This information is already determined by `DIAGNOSTIC` so there is no need to repeat it.
-
由 Tianyu Geng 提交于
The factory does not need to care about what target PSI and input a quick fix should need. We only need to ensure these two types match when an `HLQuickfix` is created. With this constraint loosened, now one factory can register multiple quickfixes applied on different targets and different input. This makes it much more flexible when implementing quickfixes.
-
由 Nikolay Krasko 提交于
-
由 Nikolay Krasko 提交于
-
由 Nikolay Krasko 提交于
-
由 Nikolay Krasko 提交于
-
由 Alexander Udalov 提交于
-
- 26 3月, 2021 29 次提交
-
-
由 Alexander Shabalin 提交于
Currently, it uses suboptimal storage for the freeze bit. This'll be revised later when we finalize stack objects handling in the new MM.
-
由 Mikhail Glukhikh 提交于
-
由 Mikhail Glukhikh 提交于
-
由 Mikhail Glukhikh 提交于
-
由 Aleksei.Cherepanov 提交于
Creation of KotlinTargetsIndex takes too long, even if Java project doesn't have any Kotlin files. Remove function hasJsStdLib, as it takes too much time because of recursively checking all dependencies (migrate to facets) Also fix tests: Add Kotlin JS facet, where its needed #KT-34351 Fixed
-
由 Jinseong Jeon 提交于
-
由 Jinseong Jeon 提交于
-
由 Tianyu Geng 提交于
-
由 Tianyu Geng 提交于
Finding type alias doesn't work if it's not top level because the class ID doesn't match. There doesn't seem to be a easy way to make it work so this commit simply makes the IDE tolerate it.
-
由 Ivan Gavrilovic 提交于
This is causing issues such as https://issuetracker.google.com/183423660. Annotation processor options that are provided by the Android Gradle plugin may contain references to files and file collections that are safe to evaluate only at execution time. This change avoids eagerly creating compiler plugin options for these options, as they are already a task input. Test: AbstractKotlinAndroidGradleTests.testAgpNestedArgsNotEvaluatedDuringConfiguration ^KT-39715 In Progress
-
由 Ilya Kirillov 提交于
-
由 Anton Yalyshev 提交于
-
由 sebastian.sellmair 提交于
-
由 sebastian.sellmair 提交于
-
由 sebastian.sellmair 提交于
This will prioritize classifiers from the following sources in the provided order 1) Classifiers from the target's modulesProvider 2) Classifiers from the traget's direct dependencies 3) Classifiers from the more common dependencies
-
由 sebastian.sellmair 提交于
-
由 sebastian.sellmair 提交于
^KT-45497
-
由 sebastian.sellmair 提交于
^KT-45497 Fixed
-
由 sebastian.sellmair 提交于
-
由 sebastian.sellmair 提交于
-
由 Ilya Chernikov 提交于
Before it, the wrong index lead to the validation error when repl script definition had c-tor parameters (see test)
-
由 Mikhail Glukhikh 提交于
-
由 Jinseong Jeon 提交于
-
由 Jinseong Jeon 提交于
-
由 Jinseong Jeon 提交于
-
由 Ilya Matveev 提交于
In a Gradle process, the user.dir property is set to the directory where the build was started. By default, if we start a child process via project.javaexec, Gradle sets its working directory to the directory of the current project. But passing Gradle's value of user.dir to that process overrides this setting. This makes tools started in a such way sensitive to directory the build is started from. This patch fixes this issue for K/N-related tools started out-of-process by the MPP Gradle plugin. But this issue is still relevant to in-process tool execution.
-
由 Ilya Matveev 提交于
In a Gradle process, the user.dir property is set to the directory where the build was started. By default, if we start a child process via project.javaexec, Gradle sets its working directory to the directory of the current project. But passing Gradle's value of user.dir to that process overrides this setting. This makes tools started in a such way sensitive to directory the build is started from. Thus a test using relative paths may fail if it is started from a wrong directory. This patch fixes this issue for Kotlin/Native tests.
-
由 Victor Petukhov 提交于
^KT-45674 Fixed
-
由 Victor Petukhov 提交于
-