This request is superseded by
request 827286
(Show diff)
Overview
Request 827139 superseded
Replace nose with pytest
- Created by jayvdb
- In state superseded
- Superseded by 827286
Loading...
Request History
jayvdb created request
Replace nose with pytest
scarabeus_iv declined request
This is nice, but we actually sent the conversions to upstreams to not have conversion hacks in specs, so we convert the package and then send it as a pr and get it merged.
Otherwise it would stay forever, only place where this makes sense is dead projects refusing new commits.
jayvdb reopened request
How rude. Upstream already uses pytest
https://github.com/python-social-auth/social-core/blob/master/tox.ini
Wouldn't it be easier then to bump to 3.4.0 that does not need the conversion? The convertor really makes sesnse for dead projects, but otherwise it might make more sense just to pick the commits or convince upstream to release a new version.
Because it isnt released yet https://github.com/python-social-auth/social-core/issues/485 And I havent reviewed the changes yet, but a minor release of social-auth-core needs to be regression tested against its dependencies because its dependencies often break on minor updates.
And when this package is updated, almost certainly this .spec will keep working as-is, and ideally the pytest plugin will warn when it is no longer needed (it already knows this, but doesnt report it in exit status or pytest warnings), or someone will notice and remove it from the spec. Normal maintenance activities.
The problem is the crust tend to stay, I can guarantee you that if you put something in the pkg it stays there for quite time before someone notices it is not really useful anymore. Here as it is in upstream I would just suggest adding comment explaining something along the lines "remove after updating to release x.y.z' and hope it would be enough to make people to notice.
You are not telling me something I do not know. However cleaning up old crap in spec files is normal. I also help doing it, as you well know. So why block my potential future crust, when you do not block all the other crust that ends up in the spec files.
And moreover, this specific crust will not stay when the tool errors when it doesn't do any conversion. How about we build that instead?