On 20.03.2015 08:47, Stefan Sperling wrote:
> On Fri, Mar 20, 2015 at 08:34:00AM +0100, Branko Čibej wrote:
>> I just noticed that we forgot to bump the displayed copyright year.
>> Fixed in r1667941 and nominated for backport to 1.9.x, 1.8.x and 1.7.x.
>> I also vetoed the 1.7.20 and 1.8.13 releases because of the wrong year
>> ... we really shouldn't release with wrong legalese, and we already
>> allowed 1.9.0-beta1 to slip through with that buglet.
>>
>> Sorry about not noticing this earlier, I realize we already have enough
>> votes tor 1.7.20 and 1.8.13; but I really think we should pull these
>> tarballs.
>>
>> -- Brane
> If we decide to pull these releases based on this problem, then I'm
> against making everyone re-run tests for this. Just allow people to
> diff the tarballs and submit a new signature based on that.
Note that we do not "make" anyone tun tests in order to vote for a
release; it's up to you to decide how thorough you want to be when
validating a release. And yes, just diffing the tarballs against the
previous version and confirming a one-character change is surely enough
in this case.
> Could we have a buildbot test for this kind of problem?
We can add tests to our test suite, that would be easiest, IMO. I'll
have a go at that.
> Should our rat-report bot (which I can't seem to locate in the maze
> of buildbot right now) perhaps check for this?
The RAT report builder is currently turned off because it was reporting
spurious failures; either because of a bug in buildbot itself, or a
problem with the slave VM. And the RAT slave didn't even build the sources.
-- Brane
Received on 2015-03-20 09:06:57 CET