On 02/05/2017 03:10 AM, Hans de Goede wrote:
Hi All,
So a lot has been said about $subject and FESco has asked me
to send a mail to the devel list describing the what and why
of this change.
There's two issues here that are important not to conflate: whether the
glvnd update should go into f25, and the poor handling of the process of
pushing it into f25 thus far. While I do have strong opinions on the
latter point, I'll try to only address the former.
I don't think anyone will disagree with you that the libGL handling
between mesa and third-party drivers is not great. Making the situation
better is an admirable goal, and I'm glad there's work being done to
improve things. However, changes as large as this run up against the
stable update guidelines[1]:
"A major version number reflects a more-or-less stable set of features
and functionality. As a result, we should avoid major updates of
packages within a stable release. Updates should aim to fix bugs, and
not introduce features, particularly when those features would
materially affect the user or developer experience."
This change definitely introduces features that materially affect the
user experience, which means that it falls under the policy and requires
an exception from FESCo *before* it can be considered for inclusion in
f25. It seems we're in a place now where the new functionality is
backed out and the unnecessary breakages have been addressed, so we can
now take a step back and seek permission instead of rushing it out and
asking for forgiveness.
The argument that someone blogged that this change would be maybe
available in f25 so we should therefore push it into f25 is frankly not
at all convincing in light of the above. A blog post doesn't override
distribution policy, and doesn't exempt anyone from going through the
processes to coordinate such a change with the rest of the project.
It's ultimately up to FESCo as to whether or not this change qualifies
for an exemption. Personally, I think it doesn't. Pushing it into a
release mid-stream seems overly aggressive to me when f26 is only 4
months away. Keeping the status quo in f25 for another 4 months won't
be the end of the world - we've gotten this far with the sub-optimal
libGL handling. If you really want f25 users to be able to test this
change, you can maintain it in a copr and allow people to opt-in (and
back out if there are issues.) Even if you do think the change is
technically ready, it has only had a couple of weeks to settle and there
may be other issues that testers haven't found yet.
In addition, I think a change of this magnitude should have really gone
through the change process, not just announced on a blog that many of us
don't read. The point of the change process is to provide a uniform way
to let everyone developing the distribution know a big change is coming,
provide a way for everyone to test against the change so they're not
caught by surprise when e.g. sway and steam games stop working, and to
provide a path to back out and regroup if the change has unforeseen
consequences. In other words, to avoid exactly what happened here. The
f26 system-wide proposal deadline was only a few days ago - there may
still be time to write up and submit this as an f26 change if FESCo
agrees to entertain it.
tl;dr: I say wait for f26 and go through the formal change process, but
it's ultimately up to FESCo.
Rich
[1] https://fedoraproject.org/wiki/Updates_Policy#Stable_Releases
[2] https://fedoraproject.org/wiki/Changes/Policy
_______________________________________________
devel mailing list -- devel@lists.fedoraproject.org
To unsubscribe send an email to devel-le...@lists.fedoraproject.org
