fix: allow minor and patch versions of minimatch

[G-Rath]

This allows downstream consumers to use later versions of minimatch, including v9.0.6 which addresses GHSA-3ppc-4f35-3m26

[G-Rath]

Running locally this passes the test suite:

> editorconfig@1.0.4 test
> npm run test:all


> editorconfig@1.0.4 test:all
> mocha && ctest . --preset Test


  [▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬▬]

  17 passing (50ms)

Test project /workspaces/editorconfig-core-js
196/196 Test #196: multiple_files_on_command_line
100% tests passed, 0 tests failed out of 196

Total Test time (real) =  16.85 sec

[G-Rath]

right so after further digging I see there is a test failure in the latest test suite, for some reason seemingly with the latest v9 version and v10 beyond the current pinned version - very odd

Relates to isaacs/minimatch#265

[hildjj]

Thanks for digging in to this. Please keep letting me know what actions I can take to move the ball forward, but as you can see, I only get a few windows a day to help.

[hildjj]

Let's rebase this, and show the tests failing for the correct reason.

[G-Rath]

@hildjj there's nothing to rebase since this is on the v1.0.x branch - also there's discussion going on over in minimatch and juliangruber/brace-expansion#88 has been approved, so I'm hopefully this will get resolved upstream soon.

I'm not sure what can be done about the macOS failure specifically other than skipping Node v14 on that OS, but I can look into that later

[hildjj]

Nod. Let's disable macos/node14, see what happens, then go from there.

[G-Rath]

@hildjj this should be good to go for v1 and v3: the latest minimatch v9 uses brace-expansion V2 which does not have the bug, and the latest brace-expansion has the bug fixed so should be good from all angles 🎉

Are you good to take it from here, or would you like my help?

[hildjj]

I can take over from here I think. The tricky part is going to be getting it released, since I think I'll need to backport the workflow changes, and we'll probably need to add "publishConfig": { "tag": "1.x" } to the package.json. I'll do this before landing #152 and publishing that patch in case I mess up the latest tag by mistake.

[hildjj]

I'm also fighting with cmake versions now.

[hildjj]

I'm going to need more help. The tests don't pass with cmake 4.2.3 which is what I have locally. They get close with cmake 3.13.4, but 65 - braces_closing_in_beginning still fails.

[G-Rath]

@hildjj I have very little experience with cmake so not sure if I'll be able to help much, but will give it a go.

I've been doing all my work in a GitHub Codespace, which looks to be using v3.28.3:

@G-Rath ➜ /workspaces/editorconfig-core-js (1.0.x) $ cmake --version
cmake version 3.28.3

CMake suite maintained and supported by Kitware (kitware.com/cmake).

Running npm run test on the 1.0.x branch and this relax-minimatch branch, all tests pass; if I update the submodule to the latest commit (https://github.com/editorconfig/editorconfig-core-test/tree/895b3a65d0d823dbd0acf2bc402376381995d1b1), then 174 - key_set_to_empty_string fails for both branches.

[G-Rath]

what would be really helpful is if you could show me how to find particular tests - searching for key_set_to_empty_string and similar is not returning anything...

found it, apparently the search just didn't want to find that particular value 🙄

[G-Rath]

They get close with cmake 3.13.4, but 65 - braces_closing_in_beginning still fails.

I have just found that tests seems to be excluded:

editorconfig-core-js/CMakePresets.json

Lines 17 to 21 in 3cdcd25

	"filter": {
	"exclude": {
	"name": "braces_closing_in_beginning"
	}
	},

So I would guess the issue is the config file is not being used correctly, rather than a new test failure?

[hildjj]

So I would guess the issue is the config file is not being used correctly, rather than a new test failure?

Yes, that's right.

All of tests pass for me with 3.28.3 as well. If we can get that version to run in CI, we can move forward.

[hildjj]

I've tested with node 14 on macos locally, so I'm fine with it not running on CI for this old version.

[G-Rath]

@hildjj CI is green, looks like things stop working after cmake v3.31 - let me know if you're happy with this, or want more work done

[hildjj]

LGTM. I'm going to take over, and tweak this to see if I can get it to release using OIDC.

[hildjj]

Great job @G-Rath! I realized right after I merged this that I really wish your name was in the contributors going forward, if you're ok with it. Can you send a patch against main for that, please?

[hildjj]

Oh F. I pushed the 1.05 patch to main. that's going to take a minute to unravel.

[G-Rath]

I've literally just shutdown my laptop 😅 but if you do the patch with me as a co-author I'm happy to approve it :)

[hildjj]

Looks like I didn't screw anything up too bad. Deep breath.