As per the title, the current version of the repo fails to clone on Windows due to the filesystem limitations: it crashes midway with an error message about a filename being too long. The result is a repo clone with half of the files missing (which can be confirmed by git status), and using the command git restore --staged * creates all of the missing files except for the one that caused the error.
The file in question is the cassette for google network handler test; it includes the entire URL (with some characters replaced by dashes, but still), and thus doesn't fit within the Windows filepath limits (270 chars of 260 allowed).
P.S. …Actually, what's up with the cassette files organisation? Some are placed in tests/vcr_cassettes/, others in tests/vcr_cassettes/tests.test_bukuDb/, and others yet in tests/cassettes/test_buku/. Shouldn't they either be in one folder, or one per testfile but named accordingly (and gathered in one place)?
As per the title, the current version of the repo fails to clone on Windows due to the filesystem limitations: it crashes midway with an error message about a filename being too long. The result is a repo clone with half of the files missing (which can be confirmed by
git status), and using the commandgit restore --staged *creates all of the missing files except for the one that caused the error.The file in question is the cassette for google network handler test; it includes the entire URL (with some characters replaced by dashes, but still), and thus doesn't fit within the Windows filepath limits (270 chars of 260 allowed).
Proposed fix: custom test case IDs
P.S. …Actually, what's up with the cassette files organisation? Some are placed in
tests/vcr_cassettes/, others intests/vcr_cassettes/tests.test_bukuDb/, and others yet intests/cassettes/test_buku/. Shouldn't they either be in one folder, or one per testfile but named accordingly (and gathered in one place)?