Re: Bug#877419: Bug#877700: RM: pandas [arm64 armel armhf mips mips64el mipsel s390x] -- ROM; Some build time tests are failing on specific architectures
Hi Yaroslav,
On Fri, Oct 13, 2017 at 01:51:51PM -0400, Yaroslav Halchenko wrote:
>
> On Fri, 13 Oct 2017, Andreas Tille wrote:
>
> > 1. proceed as I suggested here:
> > https://lists.debian.org/debian-science/2017/10/msg00001.html
> >...
> > Is there anybody who is no happy about this?
>
> sorry to be the pain, I am ... And first of all -- thanks for taking
> care about pandas!
No need to sorry.
> I would disagree on the complete disabling of the tests. Selective
> annotation of failed tests at least allows maintainer's judgement on
> either any particular failure of critical importance. Disabling ALL
> tests would let a completely broken package into the Debian for that
> architecture (not to mention that we did have failures on big-endians
> signal generic problems in the code, which just weren't triggered on
> x86s)... then why to have it there in that architecture at all?
I might try to pick some of the failed tests from the logs. I did so
once with kind of iterative uploads for python-cogent package by
checking the logs of the failing architectures. If you think the proper
way would be to login to each single architecture and build there this
would not fit into my time frame I'm willing to spent on this task.
> If ftp masters and porters insist on "better to have a broken one than
> none", I would argue that unlike typical software, where bugs are usually
> "obvious" to the user (things just fail), in scientific/data oriented
> software bugs are more inconspicuous -- you just place data in and get
> some garbage out possibly without realizing it.
>
> In summary my 1c: I am ok with RM for those archs (again), or
> partial disabling of the tests, but not ok with overall disabling of
> the test suite
I'll give the partial disabling a try.
> > 2. Upload after migrating the package to Debian Science as
> > we previously agreed upon.
>
> do you mean migrating of a Vcs?
As discussed before: Changing Vcs and Maintainer field.
Thanks for raising your opinion
Andreas.
--
http://fam-tille.de
Reply to: