DiffSK/configobj

Version 5.0.7 has disruptive changes

Closed this issue · 9 comments

romeroyonatan commented

Hi mainteiners,
first of all thanks for maintaining this library. Thanks for your time :)

I just realize that the last version changes the import of validate

Old

import validate

New

import configobj.validate

I updated my code but I think could be useful to note it in the changelog and, if it is not too much trouble, revert the change in a version 5.0.8 and re apply them in version 6.0.0 to maintain the semantic versioning

Thank you very much again

Have a nice day

triplethreatsolutions commented

Discovered the same issue this morning as well.

icct-galvarez commented

Same issue here!

robdennis commented

Thanks everyone, this is in a new milestone for 5.0.8 and will be addressed ASAP

icct-galvarez commented

Thanks so much for the support! Since I already modified our codebase to use the new import configobj.validate syntax, would it be possible to support both the old and new versions of the import statement in 5.0.8 and beyond so we don't have to go and revert those changes all over again?

robdennis commented

Certainly. There are new unit tests that assert each of the following forms work:

import validate

import configobj.validate

from configobj import validate
  • ❤️1
robdennis commented

I updated my code but I think could be useful to note it in the changelog and, if it is not too much trouble, revert the change in a version 5.0.8 and re apply them in version 6.0.0 to maintain the semantic versioning

Because, as you note, this change would violate semantic versioning, I've:

  • yanked it from pypi (with a note why)
  • prepping to release 5.0.8 with a changelog that notes this was shipped
robdennis commented

if there are new regressions, please open a new issue and I'll close ASAP

  • 👍2
romeroyonatan commented

Thank you :)