You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Some of our tests were failing on Azure because of LF/CRLF weirdness once the .gitattributes file was deleted. In the interest of moving things along, I've disabled those tests (look for AZURE in gensim/test/test_*.py).
Ideally, we should make these tests run and pass.
I have 2 ideas
apply .gitattribute only for the concrete subset of files that fail azure (replace * with paths). This will definitely fix azure, but I'm not sure does this avoid effect with "modified files" that Gordon described
"re-checkout" repo in azure with generated .gitattribute in runtime, this definitely avoid Gordon issue, but not sure will azure work with it (.gitattributes works in a pretty strange way for me, looks like this applied on clone, not on local checkout, worth to investigate)
Gordon right: hacks for CI shouldn't affect contributors
Some of our tests were failing on Azure because of LF/CRLF weirdness once the .gitattributes file was deleted. In the interest of moving things along, I've disabled those tests (look for AZURE in gensim/test/test_*.py).
Ideally, we should make these tests run and pass.
I have 2 ideas
.gitattribute
only for the concrete subset of files that fail azure (replace*
with paths). This will definitely fix azure, but I'm not sure does this avoid effect with "modified files" that Gordon described.gitattribute
in runtime, this definitely avoid Gordon issue, but not sure will azure work with it (.gitattributes
works in a pretty strange way for me, looks like this applied onclone
, not on localcheckout
, worth to investigate)Gordon right: hacks for CI shouldn't affect contributors
Originally posted by @menshikh-iv in #2836 (comment)
The text was updated successfully, but these errors were encountered: