feat: Gradle 2.x plugin SDK-based build

[sbouchet]

may close #761

[angelozerr]

@SCWells72 there are several tests with Java LS which fail, have you some idea?

[SCWells72]

@SCWells72 there are several tests with Java LS which fail, have you some idea?

Not sure. I believe tests passed 100% on my branch, but obviously lots of changes were needed to get it into proper shape.

I'll be happy to check out this branch tomorrow after I get my own release out and investigate if it's not resolved by then.

[angelozerr]

@sbouchet could you resolve conflicts please.

It seems CI build is failling. What is the status of this PR?

[sbouchet]

unit tests are failling but i don't have enough knowledge to debug/fix.
i can spot that for example formatting test when parameter is disabled does fails.
below the list of failures :

[sbouchet]

note that the preleminary work done by scott also failed, but with only one failure:
https://github.com/redhat-developer/lsp4ij/actions/runs/12918289698/job/36026492082

[angelozerr]

@SCWells72 could you help us to fix those tests please?

[SCWells72]

@SCWells72 could you help us to fix those tests please?

The issue with these Java-specific tests is that they register the extension *.java and create test files with that extension, but of course IntelliJ IDEA itself has first-class support for Java and is superseding LSP4IJ when those files. Evidently in 2023.2, that wasn't a problem, but in 2023.3 against which the new branch is running, the native Java language support is causing these failures.

I was able to work around this with some of them by changing the extension to .javax, but there are still some failures that I'm investigating now.

There's one other failure in BootstrapPropertiesCompletionTest that may be caused by a similar conflict with the IDE's first-class support for *.properties files, but I haven't confirmed that yet.

[angelozerr]

Ok a simple fix is to rename .java to .myjava

[SCWells72]

Ok a simple fix is to rename .java to .myjava

That doesn't quite work. That's pretty much what I did by changing the extension to javax, and it left some failures. I have a fix now that actually removes the file type association for *.java in setUp() and restores it in tearDown(). That fixes all but one folding test, at least on the Java side. But I should have fixes for most if not all of these shortly.

[SCWells72]

Okay, I've addressed the Java-specific failures. I confess that I'm not sure how to contribute changes to another user's PR branch, so here are the modified files:

pr_783_test_fixes.zip

Removing/restoring the file type association worked perfectly for the on-type formatting tests but not for the folding range tests, and vice-versa for the folding range tests, hence why there are two different solutions in these changes.

This leaves only remaining failure in BootstrapPropertiesCompletionTest.testCompletionOnPropertyValueWithValueAtEnd(). @angelozerr, it looks like you've been the author of that one and its associated base classes, so hopefully you can figure that one out which would get tests back to 100% on this branch.