search

NCAS-CMS / cfunits

A Python interface to UNIDATA’s UDUNITS-2 library with CF extensions:
http://ncas-cms.github.io/cfunits
MIT License
11 stars 8 forks source link

Test Actions update (generate & upload coverage reports) #14

Closed sadielbartholomew closed 3 years ago

sadielbartholomew commented 3 years ago

Test and, if/when working, merge, an update to the GH Actions run_test_suite workflow to upload a(XML format) coverage report for one representative job to the Codecov coverage service. See individual commit messages for the sub-steps required to achieve this.

(Direct equivalent to NCAS-CMS/cf-python#121 for cf-python, so see also the description on that PR for the desired outcome.)

sadielbartholomew commented 3 years ago

Oh, I should remove the Python 3.5 jobs (the equivalent of which have been removed in cf-python and cfdm already), as v. 3.5 has now reached end-of-life and I will shortly deprecate 3.5 (& document that) across the libraries. (Will do that outside this PR since it is not strictly related.)

sadielbartholomew commented 3 years ago

Again I am not seeing Codecov comment on the PR, as happened with cfdm but not with cf-python (the former of which did not have a custom config set when it went in first by PR, whereas the latter did). I suspect somehow in the Codecov YAML config I have disabled it without meaning to. Let me see if I can get a comment before I merge by editing the config...

sadielbartholomew commented 3 years ago

I haven't explicitly set to turn the PR comment off so having checked the Codecov documentation it seems the issue is that I have set the branch key to master thinking this was setting the PR base branch, but it is in fact setting the PR head branch (generally, though that setting could pass through to the comment setting). I'll remove that & see what happens.

sadielbartholomew commented 3 years ago

Retriggering by opening & closing...

sadielbartholomew commented 3 years ago

it seems the issue is that I have set the branch key to master thinking this was setting the PR base branch, but it is in fact setting the PR head branch (generally, though that setting could pass through to the comment setting). I'll remove that & see what happens.

No, considered that theory debunked as there is still no Codecov bot comment appearing. I'll merge this as-is since the comment aspect is just a nice-to-have.