Re: Use and abuse of the unreproducible tag
On 14.09.2016 00:11, Santiago Vila wrote:
> On Tue, Sep 13, 2016 at 11:31:44PM +0200, Matthias Klose wrote:
>> On 13.09.2016 18:25, Adam D. Barratt wrote:
>>>
>>> Regardless of whether there's consensus that you agree with, it's an RC bug to
>>> not build within the same release, and has been for several releases now.
>>
>> so the solution is easy. just implement a check that the package still builds
>> reproducible in testing *before* it migrates to testing?
>
> You are very funny :-) That would not work, as very often it's updated
> packages entering testing that breaks already existing ones.
>
> We deliberately break things in testing and unstable all the time,
> (new GCC releases, new debhelper breaking backwards compatibility, etc),
> because we assume that we will be able to fix whatever is broken.
>
> But we still want packages in testing to be buildable in testing, so
> the only solution is really periodic archive rebuilds, and continuous
> integration systems like the one in reproducible builds which rebuild
> packages over and over again, both in testing and unstable.
sorry, I don't share you view that this might be funny. You are known to make
issues RC issues like the missing build-indep/build-arch targets. I think the
RC severity of such reports is at least questionable. This is very much
different than adopting to be compatible with a new upstream software.
Your behavior is essentially pushing back infrastructure issues back to package
maintainers, who should not be involved by such decisions. So your next topic
should be to look at
https://buildd.debian.org/status/package.php?p=openjdk-9&suite=experimental
and file issues on the package instead of the infrastructure (hey, that's the
easier way).
not amused, Matthias
Reply to: