Re: References to registries -> debian-med policy?
Hi Steffen,
On Mon, Sep 11, 2017 at 04:16:20PM +0200, Steffen Möller wrote:
>
> > https://bugs.debian.org/cgi-bin/pkgreport.cgi?maint=debian-med-packaging@lists.alioth.debian.org
> >
> > It is not that all bugs are hard to fix - some have patches attached
> > which just need to be applied in VCS and a build test would help to make
> > sure a sponsor just needs to upload. There is also the maintainers
> > dashboard
> >
> > https://udd.debian.org/dmd/?email1=debian-med-packaging@lists.alioth.debian.org&email2=&email3=&packages=&ignpackages=&format=html
> >
> > which provides information about other problems like packages of other
> > maintainers / teams that are buggy and prevent a testing migration. It
> > makes perfectly sense to lend a helping hand there as well.
>
> I just updated bwa to 0.7.16 ;) and likely will update some more as I add
> the references to the registries. The watch files are very nice, we should
> praise them a bit more.
Hups, what do you mean by praise? You will fix one if you will see a
broken one??? I really, really hope that everybody will do so since
otherwise its a bug.
> The references to catalogues will also help
> (actually,
> it has helped already when upstream had moved to git and our watch
> file looked at the old location, still) to be informed from their side about
> new releases.
I admit that some upstreams "silently" move to some other location.
>From time to time I'm checking the validity of watch files and currently
I'm just bumping version=4 inside watch files after checking.
> The work of yours on clutalx and the Qt5 migration is impressive. If I
> get this right, then the provisioning with containers that provide the older
> Qt4 libraries would not work - am I right?
I did not more nor less than fixing bug #874851 (didn't I mention that
there are lots of bugs to fix? :-P ) There will be no Qt4 in buster and
I'm afraid clustalx upstream will not care to much. Thus I tried to
safe one of our highest popcon packages. I have no idea what this might
mean for containers. Just test whether the Qt5 build works as soon as
possible and lets fix bugs that might be reported.
> >> For the trivial bits, which includes updates and backports, the answer
> >> shall be online editable package instructions and automated testing. The
> >> testing of the larger picture in my mind shall happen with the CWL.
> > Testing will uncover not yet detected problems. The bug tracking
> > system contains *detected* problems we just need to fix.
> >
> This is correct, except that I am mostly after the extra confidence
> in particular for backports that everything is fine as in "no difference
> with what sid/testing provides".
>
> What must not happen is that the workflow-based tests are too
> sensitive and detect more differences than they are relevant - good
> old syntax vs semantic issues and ... well there is this semiotic
> triangle thingy (pun intended)
> (https://en.wikipedia.org/wiki/Triangle_of_reference)
> that we can also discuss the biological relevance of any difference
> between versions observed. I have no exact idea about how we
> get there and where we end up, but I am confident that it will be
> interesting.
So if you are converned about testing workflows in different
distributions you should write test suite for testing / unstable first
and then setup a stable machine running the test suite with the
backports.
Kind regards
Andreas.
--
http://fam-tille.de
Reply to: