Superchupu

from my testing, the array of two elements produced from globbing in that test is not alphabetically sorted, and it is in tinyglobby, every single time i've ran these tests....

that array of two elements in particular seems to be reverse sorted, not because fast-glob returns globs reverse sorted but because an array of two can only ever be either...

it seems like the config in that test in particular breaks tinyglobby, and it'd be complex and hacky enough to add. thankfully it's the only remaining failing test https://github.com/vitest-dev/vitest/blob/eae3bf4a72c1ab0b0b8ea0e13a3ed8a415ebfc66/test/reporters/reportTest2/custom-reporter-path.vitest.config.ts it...

after thinking a lot about the problem, it is something important to have regardless of its use in vitest. i have a not that hacky fix almost ready, but it...

opened a fix SuperchupuDev/tinyglobby#18, will wait until you have the meeting you mentioned though

after refactoring out the `../` in that one test tests seem to pass now, except one in macos that's failing in other prs too

ended up adding `../` support in `0.2.3` anyways, some use cases from an unrelated package were breaking without it

support for symlinks should be very straightforward, but unfortunately it needs thecodrr/fdir#114 to be merged first. support should arrive in a new tinyglobby version almost immediately once that gets released....

yep, hopefully they make a new fdir release soon 🙏

hi @yannbf, i've just released a new `tinyglobby` version that comes with symlinks resolution by default. this should fix your issue as soon as vitest upgrades to it and releases...