Add new recipe (16) by summer student: projection subplots

[sadielbartholomew]

This is the first of three recipes created this summer by our first summer student of two, Natalia. It has been prepared in the form consistent with the existing recipes.

I have only added the Python script - @bewithankit, I see four different files per recipe, is this one sufficient or do I need to generate the other three (plot_N_recipe_codeobj.pickle, plot_N_recipe.ipynb and plot_N_recipe.rst) from this for this PR, or can this be done at release time (or is the creation automatic from release r other procedures)? I will be adding the other five student-made recipes in the next few (working) days - all are in the same format as the canonical Python script.

Given the nature of recipes, the script itself should be self-contained and self-explanatory. But for context, I will add that, given the summer placement was only two weeks including time for making a poster, there wasn't time to do a full feedback round, so I have included some commits to Natalia's final script to apply linting including line wrapping; simplification of some logic and comment wording; and notably to replace one projection, 'lcc', with another, 'merc', because there is a bug with the former (see NCAS-CMS/cf-plot#75) which breaks the script, so to avoid that bug.

Expected output

Note there is a cf-plot bug (yet another one, eek, NCAS-CMS/cf-plot#77) preventing us from disabling the axes for all plots, which would make everything consistent given some of the projections don't have default in-built axes labelling.

[bewithankit]

Hi @sadielbartholomew!

Thank you, and thanks to Natalia as well for submitting this recipe. I've reviewed the Python script, and it's well-structured and clearly commented—great job! 🙌

Regarding your question, the additional files (plot_N_recipe_codeobj.pickle, plot_N_recipe.ipynb, and plot_N_recipe.rst) are generated automatically as part of the build process. What I used to do is add all the necessary files after generating the HTML and then provide a link to the temporary deployment of the webpage for reviewers to preview (for an example, see PR #688).

Note

Additionally, for each new recipe, you'll need to edit the docs/source/recipes/recipe_list.txt file to include the newly added recipe with its corresponding filters, as outlined in DOCUMENTATION.md.

With these considerations in mind, I believe this PR could be merged once other files are added/updated. Great work on preparing this recipe from Natalia's summer project!

[sadielbartholomew]

Thanks Ankit for your prompt review. 🙂

Additionally, for each new recipe, you'll need to edit the docs/source/recipes/recipe_list.txt file to include the newly added recipe with its corresponding filters, as outlined in DOCUMENTATION.md.

Aha, that makes sense - I was wondering where the filter keywords would be added. I think across the student recipes we can add some new ones too to reflect other topics/themes, but I will check in with you that you think they are suitable.
Right-o, I shall get this done and re-commit, and do this for the other 4 PRs too.

[sadielbartholomew]

Hi again @bewithankit, I've added this to the recipe_list.txt now, including adding a new keyword 'subplot' and marking that for two other recipes which use subplots - let me know if you think this works. Natalia's second recipe (#819, to be updated shortly) also uses subplots as a major part of the recipe, so I thought it would be useful to add it in especially in light of that.

Regarding your question, the additional files (plot_N_recipe_codeobj.pickle, plot_N_recipe.ipynb, and plot_N_recipe.rst) are generated automatically as part of the build process. What I used to do is add all the necessary files after generating the HTML and then provide a link to the temporary deployment of the webpage for reviewers to preview (for an example, see PR #688).

Getting the environment required for Sphinx (echo 'ERROR: Must use sphinx version 2.4.5. Got 7.4.7' etc.) is a bit of a pain, so let me suggest that I add all five recipes via PRs like this one, with the Python script and addition to the recipe_list.txt, then I do a final PR where I add all of the built material from those five to the repo finally, show they show up on the site. That way, I don't have to get the Sphinx env set up and enabled each time, and all the source and built materials stay separate in PR terms. What do you think?

[sadielbartholomew]

Given how busy Ankit is with his PhD we have agreed I can merge this without second review. At release time I will enact the plan I outlined above:

let me suggest that I add all five recipes via PRs like this one, with the Python script and addition to the recipe_list.txt, then I do a final PR where I add all of the built material from those five to the repo finally, show they show up on the site

So, merging, and there will eventually be a follow-on PR to get all the new recipes built and included.