FS#54356 - [python-magic] Replace python-magic from darwinsys with ahupp implementation
Attached to Project:
Community Packages
Opened by Oscar Garcia (ogarcia) - Thursday, 08 June 2017, 07:25 GMT
Last edited by Toolybird (Toolybird) - Tuesday, 04 April 2023, 02:38 GMT
Opened by Oscar Garcia (ogarcia) - Thursday, 08 June 2017, 07:25 GMT
Last edited by Toolybird (Toolybird) - Tuesday, 04 April 2023, 02:38 GMT
|
Details
Hi.
Current in pacman we have python-magic from http://darwinsys.com/file/ with an unknown custom license. We have other implementation with MIT license made by ahupp [1]. The second one is available in pypi and is more advanced that the first one. I have been checking the code of dependencies of python-magic and I can see that deepin-movie and diffoscope supports well ahupp version. Why not replace python-magic with diffoscope version, as can see in AUR [2] this version is used for much more packages. Greetings. [1]: https://github.com/ahupp/python-magic [2]: https://aur.archlinux.org/packages/python-magic-ahupp/ |
This task depends upon
Closed by Toolybird (Toolybird)
Tuesday, 04 April 2023, 02:38 GMT
Reason for closing: Fixed
Additional comments about closing: python-magic 1:0.4.27-1
Tuesday, 04 April 2023, 02:38 GMT
Reason for closing: Fixed
Additional comments about closing: python-magic 1:0.4.27-1
There's no guarantee the other 34 actually work with it. Your initial report said that you tested 1 of the 5 packages currently in the repos which works with the ahupp alternative (plus a 6th one which is currently not written in python, I didn't check if it used to be).
> as can see in AUR [2] this version is used for much more packages.
This is a false statement, as I said I cannot see it. Please clarify *why* you think you can possibly see it, and how you obtained this hidden information.
I agree it should be done with lots of care and testing of our repo packages, but to be honest this is a sane request.
However, the actual ticket which was opened is:
- confusingly mixing multiple issues (licensing, assertions of quality)
- asserting something, claiming it can be "proved" to be as asserted, then not actually providing the claimed data.
If you have additional information which was not yet provided in this ticket, which shows that the request is sane, I urge you to share it... the fact that no useful data or rationale was provided to date is the probable cause for this ticket languishing for over 3 years now.
It's unlikely any ticket is going to magically be answered by "you don't make a good point, but just for the record I thought the same thing and I actually have my own good reasons to do so, so I'll do it anyway". The purpose of this ticket is supposed to be to share knowledge and help the maintainer make a decision they wouldn't otherwise know to make.
That means not assuming the maintainer is secretly privy to the knowledge of why this ahupp implementation is so good.
(I cannot be of help here as I know nothing of either one. But I would like to encourage various potential ways of reaching a resolution. Whether that be switching implementations or consolidating them into a unified API.)
That is all, this was a suggestion for package maintairner, he could adopt it or close it.
For me the best solution (at least, at this moment) is that both projects merge in a well maintained one.
All have ahupp implementation ([python-magic in Pypi][1]) in requeriments (or similar file):
- [alot][2]
- [diffoscope][3]
- [python-mocket][4]
- [s3cmd][5]
- [weechat-matrix][6]
[1]: https://pypi.org/project/python-magic/
[2]: https://github.com/pazz/alot/blob/master/setup.py
[3]: https://salsa.debian.org/reproducible-builds/diffoscope/-/blob/master/setup.py
[4]: https://github.com/mindflayer/python-mocket/blob/master/Pipfile
[5]: https://github.com/s3tools/s3cmd/blob/master/setup.py
[6]: https://github.com/poljar/weechat-matrix/blob/master/requirements.txt
https://github.com/lutris/lutris/blob/master/requirements.txt#L30