Adding pytest-mpl to ultraplot workflow

[cvanelteren]

No description provided.

[beckermr]

The second call to pytest is still failing. IDK why.

[beckermr]

I think I got it...

[cvanelteren]

Yeah not sure if that second tag is necessary, but 🤞

[beckermr]

And I give up. Sorry for the noise...

[cvanelteren]

hmm strange let me give it a crack.

[beckermr]

Maybe the rc stuff needs to be in a file and not just a string on the command line?

[cvanelteren]

I am wondering if the checkout erases the files. Locally that command runs form me.

[beckermr]

yeah this command prints

% python -c "import ultraplot as plt; print(plt.config.Configurator())"
abc: False
autoformat: True
borders: False
cmap: Fire
coast: False
cycle: colorblind
edgefix: True
grid: True
gridminor: False
inlineformat: retina
innerborders: False
lakes: False
land: False
margin: 0.05
negcolor: blue7
ocean: False
poscolor: red7
reso: lo
rivers: False
style: None
unitformat: L
...

which doesn't match what the docs expect.

[cvanelteren]

Seems to be working now!

[cvanelteren]

Look at all that green ;-)

[cvanelteren]

woop woop! 🥳

[beckermr]

One more item. We need to test that this fails if we change something. For example, let's push a simple style change to this PR, ensure the tests fail, and then back that change out.

[cvanelteren]

Smart! With backing out you mean just a tmp commit like here now right?

[cvanelteren]

looks good :)

[beckermr]

The workflow doesn't appear to upload results images in the upload step. I assume we'd want this?

See here: https://github.com/Ultraplot/ultraplot/actions/runs/13055908903/job/36426869444

[beckermr]

We have to swallow the failure in the job step and then reraise it in GHA. I can look up how to do this as I have done it in the past.

[beckermr]

We need to continue on error for the test step (https://docs.github.com/en/actions/writing-workflows/workflow-syntax-for-github-actions#jobsjob_idstepscontinue-on-error) and then exit 1 in a final step using the value of steps.<step_id>.outcome and it that value is success (https://docs.github.com/en/actions/writing-workflows/choosing-what-your-workflow-does/accessing-contextual-information-about-workflow-runs#steps-context)

[cvanelteren]

cool this works!

[cvanelteren]

It is bit odd that we have to look at the "succesful" part of the test, but happy that we can test these baselines now 😄

[beckermr]

Huzzah indeed! LGTM and thanks for the hard work here!