[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Bug#814936: transition: mpi-defaults



Control: forwarded -1 https://release.debian.org/transitions/html/mpi-defaults.html
Control: tags -1 confirmed

On 16/02/16 20:58, Mattia Rizzolo wrote:
> Package: release.debian.org
> User: release.debian.org@packages.debian.org
> Usertags: transition
> Tags: confirmed
> 
> Last week I switched the default openmpi implementation on s390x from
> mpich to openmpi.
> 
> Now the default MPI implementation is the same in all release
> architectures, which is nice per se.
> 
> From what I can see in the past there were no rebuilds for this, but (as
> also others pointed out elsewhere) it would only make sense.
> There are also packages already failing to build due to dependencies
> being built against mpich but now building against openmpi.
> 
> According to my grepping the packages affected are a strict subset of
> the ones doing the openmpi transition.
> Though maybe a ben tracker is better for this job.  I try here to forge
> a valid thing, but I'm not sure of what I'm writing:
> 
> is_affected = .build-depends ~ /mpi-default-dev/ & ( .depends ~ /libmpich.*/ | /libopenmpi.*/ )
> is_good = .depends ~ /libopenmpi.*/
> is_bad = .depends ~ /libmpich.*/

This is better:

title = "mpi-defaults";
is_affected = .build-depends ~ /mpi-default-dev/;
is_good = .depends ~ /libopenmpi.*/;
is_bad = .depends ~ /libmpich.*/;
notes = "#814936";
architectures = [ "s390x" ];
ignored = [ ];

I have scheduled level 1.

FWIW this is going to block on openmpi's #814183.

Cheers,
Emilio


Reply to: