Remove astropy<4.0 constraint in setup.py #188
Conversation
|
The reason is that 0.8.4 uses I'm in the process of dropping Python 2 support, which obviously removes any use of We can merge this PR to make sure it gets removed before next release. |
|
Ah, great! Don't care if you merge this PR or not, as you like. Note that if you modernise anyways, you could consider dropping astropy-helpers as well. In Gammapy we plan to go very modern and Python 3.6+ and modern Numpy / Astropy only very soon: https://github.com/gammapy/gammapy/blob/1536061257f60b56e6db95c412ec62bd461e9232/docs/development/pigs/pig-013.rst Let me know if there's anything I can help with in Naima ... |
|
Yeah, removing astropy-helpers would make quite a lot of sense. Right now, for pure python packages it does add enough value for the added complexity. I'll add an issue and try to tackle it after the Python2 PR. Your change is in #187, so I'll close this PR. |
|
@zblz - I still see this error: https://travis-ci.org/gammapy/gammapy/jobs/591549571#L1199 Did you release a Naima version that doesn't have the Thanks! cc @adonath |
@zblz - I see you added a
astropy<4.0constraint ininstall_requiresinsetup.pyhere:https://github.com/zblz/naima/pull/186/files#diff-2eeaed663bd0d25b7e608891384b7298R99
From the
naimamaintainer perspective it's understandable, you can't be sure if something will break with Astropy 4.0.But from the user perspective and for us in Gammapy from the "depends on both" perspective it's problematic. As soon as Astropy 4.0 is released and shipped with conda,
pip install naimawill fail or pick an oder version, because you forbid it here.In practice, it's very rare for Python packages to put constraints in this direction, and I think it's very unlikely that Astropy 4.0 will break things for Naima (or Gammapy).
So: OK to remove this constraint again?