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
As discussed in #229, when determining the latest tag, it would be ideal to use a general, unambiguous test for whether or not to ignore a given SCM tag.
For instance, a commit marked with tag v2.1.0a4 should, of course, be considered version 2.1.0a4. A commit tagged only with my.awesome.tag should be ignored (and definitely NOT parsed as major='my', minor='awesome', patch='tag').
After dealing with #229, I'll add a more complete spec here for how this test should work. My preference is to delegate to an existing tool that normalizes version strings according to PEP440.
The text was updated successfully, but these errors were encountered:
I've recently started getting errors like AssertionError: cant parse version default/1.2.6.dev1
in several repos.
(I believe this tag was generated by hg-git (unconfirmed). In any case, I did make the 1.2.6.dev1 tag, but not with the default/ prefix.)
Regardless of the cause, it seems that setuptools_scm does not deal gracefully with tags that look like version numbers but aren't. Currently, it appears that all tags with a . are deemed to be version tags (in hg.py:get_latest_normalizable_tag()), which is clearly too permissive.
As discussed in #229, when determining the latest tag, it would be ideal to use a general, unambiguous test for whether or not to ignore a given SCM tag.
For instance, a commit marked with tag
v2.1.0a4
should, of course, be considered version 2.1.0a4. A commit tagged only withmy.awesome.tag
should be ignored (and definitely NOT parsed asmajor='my', minor='awesome', patch='tag'
).After dealing with #229, I'll add a more complete spec here for how this test should work. My preference is to delegate to an existing tool that normalizes version strings according to PEP440.
The text was updated successfully, but these errors were encountered: