[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Accepted dpkg 1.18.19 (source) into unstable



Hi!

On Fri, 2017-01-27 at 16:07:32 +0000, Ian Jackson wrote:
> Hi, Guillem.  I'm afraid I find myself writing a critical email.

Fair enough.

> Guillem Jover writes ("Accepted dpkg 1.18.19 (source) into unstable"):
> > Date: Fri, 27 Jan 2017 05:43:36 +0100
> > Source: dpkg
> > Binary: dpkg libdpkg-dev dpkg-dev libdpkg-perl dselect
> 
> AIUI this has missed the deadline for migration into stretch.

Nope, it went in, in time. As confirmed by Niels on IRC.

> Did you intend this for stretch ?  If not then I don't think
> it was appropriate to upload it to sid.

Yes, that was the intention.

> I have just filed three bugs, at least the first two of which I think
> are troubling for stretch:
> 
>  #852822  signing buildinfo by default breaks compatibility
>  #852821  Dropping Built-For-Profiles is risky
>  #852820  Testsuite-Restrictions field is hard to use

Of those only the first one is an issue, which I had already noticed,
and was planning on fixing in .20 after the migration. But see below.

> If you did intend it for stretch, then I question the wisdom of making
> such large changes so close to the deadline.  If (as I calculate) you
> have missed the formal deadline, you will need a freeze exception.

I'm not really happy about the timing either, but I've been pretty
swamped and didn't have the time up to now. I even had to take
yesterday's off $job (which I'll have to recover during the weekend)
to finish this up. :/ And while this might seem like a bad excuse,
sometimes life does get in the way…

I will miss the deadline due to the SH and MIPS issues, which required
an upload just now, to not break the buildd network. :(

> I think at the very least changes like these:
[…]
> ought not to get a freeze exception and are unwise at this point in
> the release cycle.

Several of those commits are supporting changes as part of the other
commits. The major changes involve reproducible, regression and bug
fixes. The few cleanup commits (that are not supporting changes) are
actually very minor and/or safe IMO.

Was it unwise to do an upload so close to the deadline? Sure. But the
other option was to not do any upload, which would have been worse. And
I even pulled out required changes that were possibly more disruptive!

In any case, I'll file a bug to the release team, and see how to deal
with this now, either an age reduction to 9, or a full review or
whatever, with the understanding that things might possibly need
reverting if they deem it necessary.

Thanks,
Guillem


Reply to: