This small PR adds missing test coverage for invalid -prune parameter values / combinations:
Not sure if the tests fit into feature_config_args.py or should rather be moved into feature_pruning.py; the latter though seems to be run less often due to being very memory-hungry.
@ryanofsky any opinion on where these tests should go? I'm not sure feature_config_args.py is the right place.
Concept ACK. Thanks for adding tests.
As these are pruning-specific, I think it would be clearest to group them with the pruning tests, not the general argument/configuration tests.
Code review ACK e2ff385e138562fb3e1cc63bdd58715a2d8bad98
<!--e57a25ab6845829454e8d69fc972939a-->
The following sections might be updated with supplementary metadata relevant to reviewers and maintainers.
<!--2502f1a698b3751726fa55edcda76cd3-->
| Coverage | Change (pull 22684, ffa8e6c828622c5011a2e43c028196bb57d2aadc) | Reference (master, 42b00a37580a11fb) |
|---|---|---|
| Lines | -0.0113 % | 83.9607 % |
| Functions | +0.0000 % | 76.0762 % |
| Branches | +0.0015 % | 52.5537 % |
<sup>Updated at: 2021-08-16T16:10:13.125878.</sup>
Looks like feature_pruning is excluded from the coverage reports as one of the extended tests.
I merged this in commit b935abb9eb2eeaceb5621e110e1889c2ce1b7734, but GitHub didn't close this pull request
I reported this as https://support.github.com/ticket/1261937
Reply from staff:
I went through our logs and was able to verify what you described here. Sometimes we have a slight database glitch that causes some pull requests to misbehave like this.
There's a workaround that some of our users have come up with - if you add a label and then remove the label, it will often update the status to closed.
In the meantime, I've reached out to our engineering team to help investigate further. While we don’t yet have a specific ETA, we’ll be sure to update you as soon as there’s any news to share.