Gentoo Archives: gentoo-dev

From: Pacho Ramos <pacho@g.o>
To: gentoo-dev@l.g.o
Subject: Re: [gentoo-dev] About adding a *warning* to remind maintainers to check for new PYTHON_COMPAT values
Date: Tue, 11 Jul 2017 11:21:52
Message-Id: 1499772100.24824.18.camel@gentoo.org
In Reply to: Re: [gentoo-dev] About adding a *warning* to remind maintainers to check for new PYTHON_COMPAT values by William Hubbs
1 El lun, 10-07-2017 a las 11:55 -0500, William Hubbs escribió:
2 > On Mon, Jul 10, 2017 at 01:04:10PM +0200, Pacho Ramos wrote:
3 > > Hello
4 > >
5 > > Looking to the list of packages still not supporting python 3.5:
6 > > https://qa-reports.gentoo.org/output/gpyutils/34-to-35.txt
7 > >
8 > > and considering that we should even start testing python 3.6, I think it
9 > > would
10 > > be nice if we could make portage to warn when PYTHON_COMPAT value is not
11 > > updated. It's really frustrating to still see new ebuilds being added with
12 > > obsolete values for PYTHON_COMPAT and relying on a few people looking to
13 > > update
14 > > this. This is also causing huge delays to migrate to newer python versions
15 > > and I
16 > > think it's responsibility of the maintainer to ensure his/her package is
17 > > supported on newer versions or, at least, have a bug and ping upstream for
18 > > the
19 > > cases they need further fixing.
20 > >
21 > > Of course, this wouldn't be a fatal check preventing you from committing a
22 > > package with outdated PYTHON_COMPAT, it would be a warning to remind you to
23 > > update it as soon as possible.
24 > >
25 > > Any issues on trying to go further into implementing this warning? 
26 >
27 > What about the situation where a package is not compatible with newer
28 > versions of python so does not need a PYTHON_COMPAT change?
29 >
30 > I don't think you can assume PYTHON_COMPAT is outdated for a package
31 > just because it doesn't have the latest versions of python
32 > listed. The only time you can know for sure that it is outdated is if it
33 > lists a version of python that no longer exists in the tree.
34 >
35 > William
36 >
37
38 Maybe we could handle it like other warnings: ignore it but ensure we have a bug
39 report to track the issue (I have seen cases of a comment being added showing
40 that a package wasn't ready for python 3.5 but no bug reference listed :/).
41
42 It looks to me like the same we do for parallel build warnings