Closed johnthagen closed 2 years ago
I reported this to cpython a while back -- there's nothing I can really do about this -- it seems to be a hardware fault
Thanks for the history. It looks like your bug report was here:
Interestingly, it was marked as fixed in 🤔 :
Hey @asottile, so some good news. We think we've isolated the upstream issue to a missing bugfix backport to 3.10:
Do you have the capability to do a test DeadSnakes build targeting a PR on CPython? It would be helpful to know if this truly fixes the DeadSnakes build issue. Thanks!
it is certainly possible, though I usually only backport specific patches for paid customers :S
We're hoping this gets upstreamed into 3.10.9, so you wouldn't need to backport yourself permanently. Was thinking of doing a single CI job against the PR as a test to see if this fixes the issue as it could help the CPython devs.
yeah typically I will land this when the patch version lands (since that takes ~2 seconds of my time to release vs. a few hours to manually backport a patch -- which is why that typically only happens for sponsors)
The backport was merged into the 3.10
branch:
cool, then it will land with the next patch version. unless someone wants to sponsor the work to get it sooner
For those following this, sadly our backport did not fix the build error on 3.10.9 18.04 armhf
... 😢
Related failing armhf
builds:
Anthony, the only thing left I could think the community could help with is if you were able to provide the exact machine specs that the builds run on, we could try to reproduce this build issue. Thanks for the help.
I have no idea, they're not my machines
@johnthagen This is the publicly available information about the builder spec: https://help.launchpad.net/Packaging/BuilderSpecs
If you need more details, you could write an email to feedback@launchpad.net
Afaik we use some special images, ran via lxd, with restricted internet access.
description
As shown here:
The
armhf
build of Python 3.10 is failing to build. This appears to be affecting Tensorflow builds downstream:Could this build be fixed, or is this a known issue/limitation with Bionic builds moving forward?