[R-pkg-team] Bug#913305: r-bioc-s4vectors breaks multiple autopkgtests

[Paul Gevers]

Hi Andreas,

[I'm curious, don't assume a know-it-better tone in my reply.]

On 09-11-18 14:17, Andreas Tille wrote:
> Currently there is a new BioConductor version released and all r-bioc-*
> packages might be in flux and tests temporarily failing.

That is what I somehow expected. I am wondering though, is it only the
tests that fail, or are the packages involved really broken during this
time as well?

> I'd recommend
> to not waste your time on these in the next 1-2 weeks (depending how fast
> Dylan will manage the transition.)

If the answer to my question above is "those packages are broken", then
my next question is, if it is somehow possible to let the package
dependency framework know that this is a transition? If apt would bail
out during installation of the packages saying that the dependencies are
uninstallable, that would be acceptable in my view (that's what happens
with c-library transitions). Before autopkgtesting, how were these
transitions managed? How was it prevented that half of your stack
migrated to testing too early?

So, in all, I wonder if the current situation can be improved at an
appropriate level, such that in cases like this there is no need for a
heads up like "don't look at package stack xyz for the next x days".

Paul

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 488 bytes
Desc: OpenPGP digital signature
URL: <http://alioth-lists.debian.net/pipermail/r-pkg-team/attachments/20181109/0f96ff89/attachment-0001.sig>