feat: Adding publication/supplemental info to label

[yimuchen]

• First running version of adding publication information to the plot labels. (Works decently well for loc=2)
• Additional adjustments will need to done for all possible loc/data combinations

[andrzejnovak]

Thanks for the quick turn-around. Could you run this with this test

mplhep/tests/test_styles.py

Line 216 in 048b2b4

def test_label_loc():

so we can check all locs at once? https://github.com/scikit-hep/mplhep/blob/master/tests/baseline/test_label_loc.png

I'd say for locs 1-3 the optimal location is obvious, less so for 0,4

[yimuchen]

Right, still a bunch of work needs to be done for many configurations... Will need quite a bit of tweaking.

[yimuchen]

Updated for decent 1,2,3 settings, Not entirely sure what to do for 0 and 4 for the very extreme cases.

[andrzejnovak]

Thanks! I think for loc=4 the SuppText should go on a new line like in loc=2. For loc=0 I am not sure. Perhaps in-frame or the right edge where the z-axis/cbar label would be.

[yimuchen]

Updated for loc=0 and 4. Made sure that changes to 4 did not change the result without the arXiv ID.

[yimuchen]

I just checked with a friend in ALTAS, I don't think they have a prefered location for arXiv labels (and if they do, I guess they can contribute to change this). So let me know if I should do any additional adjustments to these current defaults.

[andrzejnovak]

Sorry for the delay. Implementation-wise looks good to me. For loc=0 I was rather thinking this could live along the y-axis, outside of the figure like this:

I'd also say the spacing in loc 1,2,3 is a bit tight (in 2,3 this can be seen in the test with some symbols overlaping).

For loc my preference would probably be for the SuppText to be below the xsec information.

What do you think?

[yimuchen]

@andrzejnovak
What do you think about this new tweaked result?

[andrzejnovak]

Thanks a lot! I have a last comment, for loc=0 the orientation of the supptext should be flipped (such that if a person turns their head left, they can read both the supptext and y-axis label. Otherwise LGTM

[yimuchen]

Done!

[andrzejnovak]

@yimuchen it looks like the reference image is missing for the test

[yimuchen]

Should I generate it from my side? Or should we have some CI process to generate these images to ensure consistency?

[andrzejnovak]

The file should be generated locally and included in the tests/baseline. The CI will then check against it

https://github.com/scikit-hep/mplhep/blob/master/CONTRIBUTING.md#generating-reference-visuals

[andrzejnovak]

Thanks a lot @yimuchen. Let me mint a release.