Github, generate report

Report generation failed on phase "make_report",
step "Generate report".
Link: https://github.com/checkstyle/checkstyle/actions/runs/5214184671

Github, generate report

Report generation failed on phase "make_report",
step "Generate report".
Link: https://github.com/checkstyle/checkstyle/actions/runs/5223346849

Github, generate report

Regression-1 :
https://checkstyle-diff-reports.s3.us-east-2.amazonaws.com/11f38c4_2023082524/reports/diff/index.html

Github, generate report

Report generation failed on phase "make_report",
step "Generate report".
Link: https://github.com/checkstyle/checkstyle/actions/runs/5375985181

Github, generate report

Report generation failed on phase "make_report",
step "Generate report".
Link: https://github.com/checkstyle/checkstyle/actions/runs/5430072516

Github, generate report

Regression-2:
https://checkstyle-diff-reports.s3.us-east-2.amazonaws.com/11f38c4_2023104934/reports/diff/index.html

Github, generate report

Regression-Aug-4-1:
https://checkstyle-diff-reports.s3.us-east-2.amazonaws.com/a8d1851_2023191323/reports/diff/index.html

Github, generate report

Regression-Aug-4-2:
https://checkstyle-diff-reports.s3.us-east-2.amazonaws.com/a8d1851_2023205715/reports/diff/index.html

@Kevin222004 what do we need to do to get this PR out of draft mode?

rebased and conflict is resolved we will move forward with removal of code as we see that diff report finds nothing.

I would be extremely careful removing this type of code.

We are removing items from the array on a leave token in this code. This code is the only time we remove items from the array. This means we are always adding to the array, so every method, every inner class is being collected together into this one array. We have seen weird issues before because of this type of behavior and code type. If you want my approval, I would want proof that this won't harm us.

I am not seeing a reset on this variable on a new file, so it seems we are even collecting items across multiple files basically making this a multi-file check it it is true.

CI job for no exceptions is failing with Caused by: java.lang.OutOfMemoryError: Java heap space: failed reallocation of scalar replaced objects that contribute to same reason as described above, as collection is growing.

Tests like testFinalLocalVariableSwitchAssignment leak data after the file is run.

prevScopeUninitializedVariables: [[res[92x15]], [], [], [], [], [], [], [], [], [], [], [], [], [], [], [], [], []]

It is the handling of the EXPR token and the difference between visit and leave which is leading to the uncleared data. They don't match up exactly. This isn't causing any issues right now most likely because we are dealing with ASTs and the code only currently does a == check on 2 separate instances of the AST, which will never match up unless they are from the same file. Different files, or even different AST generation runs, will have different AST instances.

I did a difference run with new code clearing the fields and it produced no differences.

In addition, we run Checkstyle twice during our tests. Once during verify(Checker checker, File[] processedFiles, Map<String, List<String>> expectedViolations) and the second during getActualViolationsForFile(Configuration config, String file) throws Exception.

@rdiachenko , please help to crack this heavy puzzle.

Resolved by #13814