Closed irl closed 7 years ago
The code you cite is part of the test framework, shunit2 not the tests themselves. So perhaps you can open an issue a ticket there and describe the problem.
As for
It would be possible to run the tests with the locales set to C, but this would really just be masking the problem.
Masking what problem? That some sets of tests fail because shunit2 is confused by certain locales? I would be much more concerned about a failure that say that this confusion causes zshdb to malfunction in debugging a zsh program. Is there any evidence it is?
I'm a bit busy right now. And given this rough assessment, I won't be able to get to it in the near future. If you can, would you please see if a more recent shunit2 has this problem? If a newer shunit2 is desirable, can it be slotted in here or will there need to be adjustments to the tests?
And as you hear a lot in open-source github projects: merge requests are always most welcome.
Thanks for the information @rocky. I shall investigate.
The shunit2 repo looks not very friendly, but the version there seems to be newer than the latest version in Debian.
This was fixed in Debian by forcing locale to C. You're entirely correct, the problem lies with shunit2.
Forcing a locale to C in tests seems to me to be perfectly reasonable thing to do for testing. Would pass along the change so that is used more pervasively?
~Nevermind. I see it it shunit2.~
Change that again. You said that this wouldn't work. Please send a patch.
This bug is forwarded from Debian #841342.
zshdb fails to build from source in unstable/amd64 under some locales (eg. LANG="fr_CH.UTF-8").
The code already tries to work around this:
… but this doesn't work if the "Binary file" message is in, say, French :)
Changing this:
… makes this appear to work, but then you get other errors:
etc. etc.
It would be possible to run the tests with the locales set to C, but this would really just be masking the problem.