On 12/23/21 05:11, Richard Henderson wrote:
> Hi John,
>
> This test has been failing for quite a while. While it is allowed to
> fail, can we either fix this or disable it, to allow the ci to go proper
> green?
This job not only fails on the mainstream pipelines, but also on
all forks recently rebased. This is very annoying, I am in favor
of disabling the job until someone figure out what is wrong.
Phil.
I've been away on holiday, and will be away for another week.
Looks like a few dependencies were updated in the python ecosystem and broke a few things. I tried to fix it in a few hurried minutes, but ... it looks pretty broken:
(33/33) tests/pylint.sh: FAIL: Exited with status: '1', stdout: "************* \nqemu/__init__.py:1:0: F0010: error while code parsing: Unable to load file qemu/__init__.py:\n[Errno 2] No such file or directory: 'qemu/__init__.py' (parse-error)\n", stderr: "************* \nqemu/__init_... (0.54 s)
I am not sure why on god's green earth Pylint has decided it doesn't know how to import namespaced modules anymore. I probably won't be able to fix this until next week, so just disable it for now and I'll fix it when I get back.
--js.
(ps: attempted fix here
https://gitlab.com/jsnow/qemu/-/pipelines/440208449 but there seems to be something very wrong with pylint in the tox environment -- I suspect a dependency of pylint's was updated and is breaking things. I'll have to identify which one it is and update the package reqs to exclude it, but ATM I have no idea which it is.)
(pps, Cleber, what happened to the avocado output here? It's not showing error logs from the failed tests anymore. Did avocado update and break these options?)