Handle an edge case with well-known ignores when applying exclusions

[daniilsapa]

Closes #138

This was a subtle one. The problem was that the folders that were disabled in config but still had diagnostics reported on them in my local env contained .DS_Store files. .DS_Store files prevented a folder from being excluded from VFS. The reason is that minimatch apparently has some registry of ignored files and it always returns false for them. It becomes a problem when we apply inverted glob groups (not(...)) as we invert the result when matching such a group and end up including those files/folders

Let me know if I failed to clearly articulate the problem and you need more explanations

[changeset-bot]

🦋 Changeset detected

Latest commit: 12301b8

The changes in this PR will be included in the next version bump.

This PR includes changesets to release 1 package

Name	Type
steiger	Patch

Not sure what this means? Click here to learn what changesets are.

Click here if you're a maintainer who wants to add another changeset to this PR

[illright]

Let's try to find another solution, because this solution might actually produce unexpected behavior for user-created hidden files (i.e. .eslintrc.js, contrived example, but still). Maybe we can even discuss it in the issue rather than in the PR

[illright]

note: I think what's happening here is that hidden files (those whose name starts with a dot) are not matched by globs by default. That's how terminal shells handle globs too.

issue: I don't think ignoring hidden files it's the right approach here, since there are legitimate reasons for why you might want to make a source code file hidden.

suggestion: I'd rather investigate why these files are picked up at all. I was under the impression that we automatically disregard the files that are in .gitignore. If we don't, then we should, I think.

[daniilsapa]

Oh, yeah, it was so obvious to me that it should have ignored only default hidden FS-specific files that I didn't even checked how it behaves for dot files as such. Yes, it turns out this happens with any dot files, not just a few specific ones. Given these new circumstances, yes, that's not the right approach at all.

Yes, I'll investigate why the files that are in .gitignore are not disregarded by default

[daniilsapa]

I rolled my previous changes back. It turned out you need to supply globby's isGitIgnored with a path to the root of a git repo. It doesn't search it recursively. Adding that function solved the problem

[illright]

Thanks for finding the issue and fixing it :)