Travis-CI linting set-up means that commits to files with pre-existing issues fails

[martin-cs]

If you touch files in master that have linting issues then travis-ci will fail even if you haven't introduced these faults. Possible solutions / work-arounds:

1. Move linting to the end of the tests, not the start (my preferred solution).
2. Make everyone submit linting clean up patches for all files before they edit them (not recommended).
3. Ignore the lack of testing in this case (the current solution).
4. Make linting failures warnings not errors for CI.

Thoughts?

[tautschnig]

Actually linting should be done in a separate (i.e., parallel) job; at least one current pull request already touches .travis.yml, so maybe we should that get merged first and then rewrite the "script" entries. We can actually get rid of the "install" section as well.

Then, of course, 2 should happen, possibly in a concerted bug-squashing session. (There are way too many open pull requests at the moment, with varying assignees. Those should get done before looking into the re-formatting at large scale.)

[martin-cs]

On Sun, 2016-12-18 at 09:35 -0800, Michael Tautschnig wrote: Actually linting should be done in a separate (i.e., parallel) job; at least one current pull request already touches .travis.yml, so maybe we should that get merged first and then rewrite the "script" entries. We can actually get rid of the "install" section as well. Then, of course, 2 should happen, possibly in a concerted bug-squashing session. (There are way too many open pull requests at the moment, with varying assignees. Those should get done before looking into the re-formatting at large scale.)

All sounds great, I'm just concerned that the current configuration means that things aren't getting tested due to simple "failures". Cheers, - Martin

[tautschnig]

#364 (which needs #362) gets us back in the position where linting failures can be clearly distinguished from other test failures.

[tautschnig]

A somewhat large amount of cleanup is now in #364, but also #333 helps to reduce the number of linting warnings.

[thomasspriggs]

This issue looks to be no-longer-relevant, as we have -

• Moved to github actions rather than Travis.
• Moved the linting checks into a parallel job.

Please feel free to either re-open this issue, or open a new one if there are still linting related problems.