Re: Inconsistency in source package naming for python modules
On Jul 08, 2013, at 09:59 PM, Thomas Goirand wrote:
>Over the last months, I've seen lots of inconsistency in the source
>package naming scheme in the python module maintained in the team.
>Sometimes, module X will have its source package called python-X or just X.
>
>If we have a python module named X, then IMO, we should stick to call
>the source package python-X, and not just X. Why? Because AFAICT it
>seems that there's a consensus in Debian that, if a package is producing
>a single binary, then its source package should have the same name.
Hopefully, it will become more and more common to have at least python-X and
python3-X. With that in mind, many of our source packages that are producing
a single binary package today should hopefully be producing two or more binary
packages tomorrow.
>It isn't my intention to fix mistakes already made (IMO, too much work
>for not enough rewards), but I wanted to raise this topic to check if
>others have the same opinion, and to make sure we have this in the
>python policy (in one way or the other). Thoughts anyone?
In light of the above, I think it makes sense to name the source package after
the upstream package name if possible, e.g. as you might see it on PyPI. Of
course, there's a wild variation there too[1]. But I think that makes it
generally easier to find the path back to the original source[2].
-Barry
[1] e.g. mimeparse and python-mimeparse look quite similar except the latter
works in Python 3 instead of just claiming to. ;)
[2] Here's a dumb little script I've always wanted.
http://bazaar.launchpad.net/~barry/+junk/wid/view/head:/wid.py
Reply to: