On 26. 01. 21 15:52, Alexander Scheel wrote:
Well, I understand your sentiment against mass spec changes, but there are
cases, where it currently cannot be avoided (e.-g. when a targeted mass rebuild
is needed for a soname bump). W.g. when we update Python from 3.9 to 3.10 we
will need to rebuild (close to) 4000 packages. How are we supposed to check what
the latest package in rawhide is and how do I act accordingly if it is not
dist-git HEAD? (Also, mass rebuild.)
Just responding to this one point (and in light of the parent's post
as well)....
...is there a way that the Python bump can be done alongside the usual
(and pre-scheduled) mass-rebuild of Fedora packages?
I realize Python release bumps are becoming a regular occurrence but
I'd rather we piggy-back off an existing deadline (mass-rebuild) than
impose other artificial restrictions.
I am afraid I don't understand. The actual rebuild needs to happen in a side tag
which is not similar to the mass rebuild side tag. Also if we mix it up with the
regular mass rebuild, many packages would fail until the Python bump is
finished. The actual Python rebuild will happen 1-2 months before the mass
rebuild to allow us to merge the side tag in time for the mass rebuild.
That's about it for the official rebuild that happens once fro each Python
release and is scheduled in
https://fedoraproject.org/wiki/Changes/Python3.10#Important_dates_and_plan
There are more somewhat possible rebuilds that might follow, see for example
https://lists.fedoraproject.org/archives/list/python-devel@xxxxxxxxxxxxxxxxxxxxxxx/message/VZCGTCZQLPCXIFHQFKOVVDJC2PL7PW5N/
I bring this up because, for a week, dogtag-pki was FTBFS in Rawhide
(and, your scripts caught it) but I couldn't understand_why_: my
local builds were succeeding just fine, it was only remote
scratch/non-scratch builds that were failing. It turns out to be a
change in my local pki-core (which was installed), but which wasn't
yet updated in the dependency in Fedora.
This is, admittedly, a case for simplifying PKI's deliverables (and
perhaps, make it a single source package), but still: even
well-intentioned packagers occasionally get stumped and unknowingly
push bad updates.
This policy idea is mostly to avoid the cases that are intentional ("let's push
this and see what happens even when I have no idea yet"), not to avoid mistakes.
Mistakes will keep happening.
If we instead schedule time around mass-rebuild, I think this would
reduce stress for all parties.
In order to reduce stress during the side tag rebuild, we keep rebulding rawhide
with Python 3.10 contiguously. This is to have time to fix the issues, instead
of pushing Python 3.10 and filing bugs later, suddenly having many urgent
problems to solve.
--
Miro Hrončok
--
Phone: +420777974800
IRC: mhroncok
_______________________________________________
devel mailing list -- devel@xxxxxxxxxxxxxxxxxxxxxxx
To unsubscribe send an email to devel-leave@xxxxxxxxxxxxxxxxxxxxxxx
Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/
List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines
List Archives: https://lists.fedoraproject.org/archives/list/devel@xxxxxxxxxxxxxxxxxxxxxxx
