Control: reopen 1002789 Control: found 1002789 1.12.0+ds1-5 Control: retitle 1002789 python-pycdlib: FTBFS during tests if /tmp is not a tmpfs Control: severity 1002789 important
On 03/03/24 at 23:29 +0100, Thomas Goirand wrote: > On 3/1/24 15:36, Lucas Nussbaum wrote: > > Control: reopen 1002789 > > Control: found 1002789 1.12.0+ds1-5 > > > > Hi all, > > > > Unfortunately, I can still reproduce this failure with version > > 1.12.0+ds1-5, as shown in the attached build log. > > Sorry about that. > > > > Let us (Santiago or myself) know if it would be useful to provide a VM > > to reproduce this issue. > > > > Lucas > > Hi Lucas, > > How do you explain that the package builds fine on my laptop, and on the > Debian buildd servers, as 1.12.0+ds1-6 (that only contains a cosmetic > clean-up of .pytest_cache on clean, compared to -5) built find after I > uploaded it? See: > https://buildd.debian.org/status/package.php?p=python-pycdlib&suite=sid > > I'm therefore closing this bug again. > > It's frustrating, I would have like to understand. Maybe something related > to /tmp mounted as tmpfs vs a "real" partition? Right, it looks like it only builds fine if /tmp is a tmpfs. Which looks like a valid bug, isn't it? (I can reproduce the failure on my laptop with /tmp on the same filesystem as the build dir, and then the failure disapears if I mount -o bind /dev/shm /tmp). Note that closing a bug just because you don't understand it does not look right. I'm setting the severity to important since we know how to make it build. Lucas