Re: Could we disable "AutoQA" until it can produce usable results?
Witness (a):
https://admin.fedoraproject.org/updates/libguestfs-1.29.31-1.fc22
The log file in this case is 3452 lines long.
Witness (b):
https://admin.fedoraproject.org/updates/ocaml-camlp4-4.02.0-0.9.git87c6a6...
The log file in this case is 3452 lines long (not coincidentally - it
is exactly the same file as above).
And in both cases the log files are a morass of randomness.
Sorry about this. Adam already noted how to search in those logs. Recently, we have
implemented support for task artifacts, and split upgradepath results per build and per
update, so you'll be able to see only the output relevant to you:
http://taskotron-dev.fedoraproject.org/artifacts/20150324/6a4b8934-d24c-1...
We still don't have any way to present it to you easily (you need to know how to
construct these arcane URLs), but we're working on it. And we still haven't
implemented result splitting for depcheck yet.
In your exact case, this "inferior architecture" error message is a known
irregularly occurring issue and it's a bug somewhere in depcheck or dependency solving
stack, but we haven't been able to pinpoint it and fix it yet. Recently we have
received some guidance from libsolv maintainer and we're trying to reproduce the
problem with some debugging info. Unfortunately, it's quite difficult to reproduce.
Disabling the tests until they work flawlessly and we have a nice way of representing the
results is on the table, if many maintainers think it's better than having at least
some logs in some form. However, having this kind of feedback helps us to prioritize the
tasks we work on.