[Bug] CI error: flake8 Expected - Waiting for status to be reported

[neteler]

Describe the bug
Maybe well known but I see numerous PRs failing with

flake8 Expected — Waiting for status to be reported

This happens esp. with pure C language PRs where a flake8 test doesn't make too much sense.

C related changes:

• d.legend: handle 0 for log (-l) #1507
• Add mod(x,y) function to r.mapcalc and r3.mapcalc doc #1617
• v.out.ogr: Fix a typo and remove -z flag mention from warning messages #1694
• r.mfilter: implement parallelization with OpenMP #1708
• r.neighbors: implement parallelization with OpenMP #1724
• ...

But also Python changes

• grass.pygrass: Improve exception handling in Info.open (#1555) #1559
• modules: Replace bare except by Exception in r.unpack and i.in.spotvgt #1578
• style: scripts/v.report/v.report.py:fixed F841 local variable 'numcols' is assigned to but never used #1609
• i.tasscap: utf-8 encoding for non-ASCII chars #1643
• r.proj: Add test cases #1663
• ...

It would be nice to see the flake8 test treated in a better way.

[ninsbl]

Flake8 is actually a required check, impeding merging PRs where the flake8 check is not reported as successful (even if it is). The simplest way in the github web UI I found to trigger re-run of the stale check is to close and re-open the PR:
https://stackoverflow.com/questions/52408592/how-to-relaunch-github-check-without-pushing-new-commits/52412240

[agiudiceandrea]

@neteler @ninsbl what about #1694?

[wenzeslaus]

This happens...with pure C language PRs...It would be nice to see the flake8 test treated in a better way.

One reason the test is required is a limitation of GitHub: You cannot avoid accidental pushes from local repos to the main branch in any other way than requiring a one of the test jobs to pass, but what the job actually does is not important in the context of protecting the branch. With the branch protection rule requiring a passed check, a PR is required for modifying the branch because that's the only way how to get a job running on relevant code before the actual push.

The check depends on the particular name of the job, so for old PRs, there is simply no job of that name (and that's why close/open or code update is needed to trigger new jobs).

[neteler]

Issue seems to be gone, closing.