[dev] modules.contrib.nvidiagpu is broken when there's a Nvidia GPU

Bug Report

Summary

Affected module: contrib.nvidiagpu

Description:

Happens in revision a8ab4f95 for me.

File "/mnt/d2/downloads/work/bumblebee-status/modules/contrib/nvidiagpu.py", line 28, in hidden
if "not found" in self._utilization.startswith("not found"):
AttributeError: 'Module' object has no attribute '_utilization'

First glance shows a naming inconsistency most likely due to a typo, because the name lacks an underscore compared to the one from Module.__init__()

Adding an underscore produces a different error:

File "/mnt/d2/downloads/work/bumblebee-status/modules/contrib/nvidiagpu.py", line 28, in hidden
if "not found" in self.__utilization.startswith("not found"):

TypeError: argument of type ‘bool’ is not iterable

A deeper look at this of course shows there’s something wrong with it.

I’ll try to submit a pull request that fixes it for me in a few minutes, but I have no way to test it on a setup when the Nvidia GPU is missing, try testing that use case too if possible before merging, it would be a shame if the change that fixes it for me will break it for others.

[ghost]

Updated the repo after your fix and bumblebee-status runs correctly, feel free to close.

[tobi-wan-kenobi]

Yes, please. I was really torn between the two solutions (as the initial bug clearly demonstrated), but I think searching for ‘not found’ in the whole string should make the solution slightly more stable when the tool’s output changes slightly.

Thanks for the quick PR, though!