+1 maintenance report

Olivier Gayot olivier.gayot at canonical.com
Mon May 16 08:02:15 UTC 2022


I was on +1 maintenance last week. I focused on -proposed migrations 
going up from the bottom of the list in excuses.

Thanks to @slyon for sponsoring, re-triggering autopkgtests on my behalf 
and providing advise throughout the week!

Many packages with regressions were missing logs so I requested a bunch 
a re-triggers. racket-mode, namely, failed to re-trigger with "does not 
have any test results".

### armci_mpi

Autopkgtest failed on ppc64el. The autopkgtest process got interrupted 
by SIGPIPE. Test green after re-triggering.

Other regression on s390x was reported by Graham and acknowledged by the 
Debian maintainer:

### python-fluids

Autopkgtest would fail on all architectures. Analysis showed that the 
upstream test-suite was running over all .py files from the tree (with 
some exceptions).
In Debian/Ubuntu, tests would run over the test files under the .pc/ 
directory ; where quilt would store the original test files after 
applying patches.

1.0.9-1ubuntu1 was uploaded and .debdiff forwarded to Debian.


Unfortunately, there is another regression in python-fluids (likely 
endianness related) on s390x that I did not investigate.

### libcloud

With the new upstream version, autopkgtest would fail on all 
architectures when http_proxy/https_proxy variables were defined. 
Another test would fail if Internet was not reachable.


Another test was also failing on s390x due to endianness issue.

https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=1010891 (applied)

All changes (except for disabling the test that requires Internet 
access) have been sent and merged upstream.


Also all patches were forwarded to Debian.

3.4.1-2ubuntu2 was uploaded and made it to the -release pocket.

### pysdl2

With the new upstream release, Debian enabled the upstream test suite 
(based on pytest) in autopkgtest. In earlier versions, the upstream test 
suite was not run when running autopkgtest. Therefore, the regressions 
detected are not necessarily the result of the new upstream release.

Many regressions (all seemingly related to endianness) occur on s390x. 
Some of them seem non-trivial to fix. I started the fix some in 
ppa:ogayot/kinetic-proposed but was far from being done by EOW last week.


There is potentially a mix of issues in both:
  * the tests,
  * pysdl2,
  * libsdl2.

I opened
https://github.com/py-sdl/py-sdl2/issues/227 against pysdl2 upstream to 
get some help but did not get any feedback yet.


More information about the ubuntu-devel mailing list