Bug#954654: transition: hdf5
On 22/04/2020 11:16, Sebastiaan Couwenberg wrote:
> On 4/21/20 11:32 AM, Emilio Pozuelo Monfort wrote:
>> On 09/04/2020 11:09, Emilio Pozuelo Monfort wrote:
>>> On 09/04/2020 06:17, Sebastiaan Couwenberg wrote:
>>>> On 4/6/20 12:54 PM, Emilio Pozuelo Monfort wrote:
>>>>> hdf5 is currently blocked from migrating to testing on mpich due to #954244.
>>>>
>>>> mpich migrated to testing. hdf5 will need some help to migrate, not all
>>>> bad rdeps will be autoremoved eventually.
>>>
>>> I couldn't really understand what was going on from the normal hint output:
>>>
>>> got: 37+0: a-1:a-0:a-15:a-0:i-20:m-0:m-0:p-0:s-1
>>> * armel: cdo, code-saturne, code-saturne-include, harp, libadios-bin, libharp-dev, libharp10, libncarg-bin, libncarg-dev, libncarg0, ncl-ncarg, python3-adios, python3-cdo, python3-harp, python3-sfepy
>>>
>>>
>>> Looking closely at those packages didn't give me any hint as to why they would
>>> be uninstallable (not even on armel) so I added a full hint-type hint and did
>>> a test run, and I got this:
>>>
>>> I: [2020-04-09T08:49:35+0000] - Removed 9 of 40 cruft item(s) after the changes
>>> I: [2020-04-09T08:49:35+0000] - easy: 55+0: a-10:a-3:a-3:a-3:i-23:m-3:m-3:p-3:s-4
>>> I: [2020-04-09T08:49:35+0000] - * amd64: cantor-backend-scilab, cdo, python3-cdo, ruby-hdfeos5, scilab, scilab-cli, scilab-full-bin, scilab-minimal-bin, scilab-test
>>> I: [2020-04-09T08:49:35+0000] - * arm64: cdo, python3-cdo, ruby-hdfeos5
>>> I: [2020-04-09T08:49:35+0000] - * armel: cdo, python3-cdo, ruby-hdfeos5
>>> I: [2020-04-09T08:49:35+0000] - * armhf: cdo, python3-cdo, ruby-hdfeos5
>>> I: [2020-04-09T08:49:35+0000] - * i386: cdo, python3-cdo, ruby-hdfeos5
>>> I: [2020-04-09T08:49:35+0000] - * mips64el: cdo, python3-cdo, ruby-hdfeos5
>>> I: [2020-04-09T08:49:35+0000] - * mipsel: cdo, python3-cdo, ruby-hdfeos5
>>> I: [2020-04-09T08:49:35+0000] - * ppc64el: cdo, python3-cdo, ruby-hdfeos5
>>> I: [2020-04-09T08:49:35+0000] - * s390x: cdo, python3-cdo, ruby-hdfeos5
>>> I: [2020-04-09T08:49:35+0000] - FAILED
>>>
>>> Basically we could remove cdo, ruby-hdfeos5 (or wait for it to be a candidate
>>> but I would remove it now to finish this) and scilab. Unfortunately after checking
>>> with dak, I found that tasksel depends on meta-kde, which depends on cantor, which
>>> depends on scilab, which means that scilab is going to need a fix for #955694 in
>>> order for this transition to finish.
>>
>> scilab got fixed, and openmpi became a blocker but also migrated. Some other
>> packages received uploads but after hinting them hdf5 was finally able to migrate.
>
> It looks like hdf5 didn't actually migrate:
>
> $ rmadison -s testing hdf5
> hdf5 | 1.10.4+repack-11 | testing | source
>
> 1.10.6+repack-1 is only in unstable.
>
> britney_output/2020/20200421030001.txt.gz shows a bunch of packaged
> involved in the transition as accepted, but also that the hint failed.
>
> britney_output/2020/20200421040001.txt.gz shows that hdf5 is not a valid
> candidate (or it already migrated).
ftr, that last part confused me and made me believe that the package had
migrated (since I had been looking at it and had added various hints that should
have made the transition migrate, but obviously something went differently and
it didn't). Since then the transition got entangled with netcdf and pdal, so we
decided to rebuild those two packages in tpu to disentangle them. After a few
more hints to get other rdeps to migrate (some had seen new uploads or were
blocked due to a failing autopkgtest) hdf5 finally migrated along with all its
rdeps.
Cheers,
Emilio
Reply to: