Write a new naming guideline for python/python3 modules?

[Chen Lei]

2010/9/16 Toshio Kuratomi <a.badger at gmail.com>:
> When we've talked about this before, in the Packaging Committee we
> haven't really cared to stipulate one proper way that maintainers must
> follow since there's several possible ways which all seem equally
> valid.  Inconsistency by itself is not a problem.  If it's causing an
> issue then it would be something the packaging committee would
> address.
>
> Note that the packaging committee last discussed this several years
> ago so people might be more amenable to some parts of your proposal --
> In particular, whether to mandate python-pyfoo or allow pyfoo was
> contentious before and new things have emerged (namely, that python3
> packages must prefix with python3-).  However, we'd likely grandfather
> existing pyfoo packages in so they wouldn't have to rename.  I
> personally don't htink the other parts of your proposal make the
> guidelines more clear but the packaging committee as a whole would
> decide this so you're welcome to propose it.
>
I think we can only rename pyfoo packages which are already ported to
python3 (e.g. PyQt4) currently, very few packages are affected by this
proposal IMHO. If a exsited pyfoo/foopy packages want to add support
to python3, then we can rename them one by one to keep a consistent
name with their corresponding python3 modules. However, we can mandate
all new pyfoo/foopy packages add python- prefix ( it seems you also
like this idea that append python- prefix to all python2 modules?[1]
).  If FPC don't like this proposal, at least we should not allow
foopy packages to apply exception rules to pkgname, I think
python-scipy is more consistence than scipy.


>From Fedora naming guideline - "They should take into account the
upstream name of the python module. This makes a package name format
of python-$NAME. When in doubt, use the name of the module that you
type to import it in a script. "  So currently package submitters are
free to choose either upstream name or module name, I'd like an
opinion from FPC that which name is preferred or highly recommended
for a python module - module name which we type to import it in a
script or upstream name which is mostly considered as tarball name,
the naming guideline is ambigous in this point. In Debian, they always
use module names unless the package ships multiple python modules, it
seems dmalcolm and tomspur also like this naming convention.
Personally, I also like this proposal, howerer I can also acceptable
use upstream name as a perfer as long as FPC detemines which naming
convetion are preferred. e.g. python-zmq/python-PySide(module name) or
python-pyzmq/python-pyside(upstream name) looks good for me,
python-zmq/python-pyside is considered as inconsistence, there are
more exsited example in fedora repos.

I'm not a native English speaker, maybe my statement is not clear
enough. But I think ambigous statement is not trivial to Fedora naming
guideline (the name of a particular package is unimportant though) ,
after all the purpose of naming guildeline is providing a consistent
naming convetion for the whole Fedora distribution. I hope some
volunteers can help me to write a formal guideline draft.


[1]http://lists.fedoraproject.org/pipermail/python-devel/2009-October/000193.html


Regards,
Chen Lei