Hi Rebecca
On 2019/09/18 23:51, Rebecca N. Palmer wrote:
Was your local build on a different architecture, or must this be
something random / hardware-or-environment-dependent / a very recent
change in a build-dependency?
I first ran into a problem with the arch:all build of 0.23.3+dfsg-5 [1].
This seemed to be upstream issue #24088 [2], so I tested a src:all build
including the patch on barriere.debian.org.
This should have been caught by the autopkgtest, but I noticed that
python3-statsmodels was missing from the test dependencies.
After the build was successful on barriere, I uploaded with both changes.
At this point, I should have filed a bug against statsmodels to prevent
the migration of 0.9.0-5 to testing, but was expecting the pandas builds
to go smoothly.
I see 0.23.3+dfsg-6 was given back on amd64, and has just built
successfully. I'll keep an eye on the arch:all build, and investigate
further if needed.
Regards
Graham
[1]
https://buildd.debian.org/status/fetch.php?pkg=pandas&arch=all&ver=0.23.3%2Bdfsg-5&stamp=1568792282&raw=0
[2] https://github.com/pandas-dev/pandas/issues/24088
[3]
https://salsa.debian.org/science-team/pandas/commit/b06adee846441925c3b87dc2a8d3c3a845f2e752
