figure.py: make plotly an optional dependency

[suvayu]

Please confirm that this PR has done the following:

• Tests Added
• Documentation Added
• Name of contributors Added to AUTHORS.rst
• Description in RELEASE_NOTES.md Added

Description of PR

With this PR, I am proposing to make plotly an optional dependency.

One of the feedbacks I got for friendly data (which uses pyam) was that there's a version mismatch with plotly with their environment, preventing them from using the Python API directly from their analysis code.

I noticed that plotly is used only in one place in figures.py, so I thought maybe this is an acceptable proposal. Since it's easier to talk in code, I wrote this PR. If I have misunderstood the importance of that module, please feel free to close this.

Thoughts?

[danielhuppmann]

Thanks @suvayu, don't see an issue with making plotly an optional dependency. But I'm not sure if the current implementation works, I suggest to rather follow the approach used in the datareader module, see

pyam/pyam/datareader.py

Line 50 in 6435898

if not HAS_DATAREADER: # pragma: no cover

Also, you'll have to add the new dependency type to all GitHub actions workflow install steps, see

pyam/.github/workflows/pytest.yml

Line 39 in 6435898

run: pip install -e .[tests,deploy,optional-io-formats,tutorials]

(and the same in the other workflow recipes).

[suvayu]

Hi @danielhuppmann , thanks for your comments, I have pushed commits addressing them, please let me know if I missed something.

[danielhuppmann]

Thanks! Letting the tests do their thing...

[danielhuppmann]

The workflow recipe for building the docs also needs the new optional plotting dependency so that it can build the gallery.

pyam/.github/workflows/build-docs.yml

Line 39 in 6435898

run: pip install -e .[tutorials,docs]

And the ReadTheDocs config will then also need that...

https://github.com/IAMconsortium/pyam/blob/main/readthedocs.yml

[codecov]

Codecov Report

Merging #549 (18acff0) into main (6435898) will decrease coverage by 0.1%.
The diff coverage is 100.0%.

@@           Coverage Diff           @@
##            main    #549     +/-   ##
=======================================
- Coverage   93.6%   93.4%   -0.2%     
=======================================
  Files         48      48             
  Lines       5299    5301      +2     
=======================================
- Hits        4961    4955      -6     
- Misses       338     346      +8     

Impacted Files	Coverage Δ
setup.py	0.0% <ø> (ø)
pyam/figures.py	38.0% <100.0%> (+6.5%)	⬆️
pyam/datareader.py	50.0% <0.0%> (-50.0%)	⬇️
tests/test_datareader.py	69.5% <0.0%> (-13.1%)	⬇️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 6435898...18acff0. Read the comment docs.

[suvayu]

Done :)

[danielhuppmann]

Thanks, looks good! I marked the first two to-do items in the PR description as not applicable - do you want to add your name to the list of contributors?

[suvayu]

Thanks, I feel this is too trivial a change for a mention in the Authors list :-p

[danielhuppmann]

Thank you @suvayu for resolving dependency conflicts between pyam and some of the modelling frameworks out there!