Add pydantic/ipyautoui version of dashboard

[mwcraig]

This PR adds a draft of the pydantic/ipyautoui version of the dashboard.

[mwcraig]

@nicole-brewer @JuanCab -- this is ready for a look.

I haven't written the answers yet, but will work on that later today or tomorrow.

I know I want to add a bit more to the ipyautoui widget (i.e. more in notebook 3b) and to save that to a python module.

The notebook I added, 3c, is a possible extension we could plan to do, or include but do only if time permits, or drop.

Feedback is very much welcome, especially in places where you find it confusing.

[mwcraig]

@jgunstone @ollyhensby -- I wanted to let you know that we're planning to use/promote ipyautoui as part of a widget tutorial at SciPy 2024. Any feedback is welcome (note these are not fully polished yet)...and if you find anything here that might be useful to the ipyautoui documentation feel free to use it.

But mostly thanks again for writing ipyautoui -- it is so much easier than doing widgets by hand!

[jgunstone]

hi @mwcraig - this is great news, thanks for letting us know. We're so happy that you're enjoying using 'ipyautoui', and that it will feature in your tutorial at SciPy. I'll take a look over the notebooks when I get a chance and will defo harvest for the docs :) thanks again

[mwcraig]

Do you want to create a version fo the entire dashboard using the pydantic/autoui code? If you do, it could illustrate how much more compact that code is.

:rofl: I had sort of forgotten until I reviewed #8 that there was more to the dashboard than these controls.

[mwcraig]

Thanks for the review -- I'll plan on making the revisions Friday morning.

[mwcraig]

@nicole-brewer @JuanCab -- I've done a fairly extensive rewrite of this. The first notebook walks through pydantic and mentions ipyautoui. The second discusses ipyautoui some more. The third has class-ified versions of the remaining widgets, and introduces adding traitlets to a class. The fourth assembles the pieces into the dashboard and introduces link as another way to accomplish what observe does.

I'm looking forward to your feedback on this and am happy to drop/rearrange/expand as needed. I'm a little worried I went overboard here...

[JuanCab]

I performed a code review the afternoon of June 24. I did check all the %answer worked as expected...

• In 03b_ipyautoui.ipynb

  • Answers have not been exported to their answer key files yet.
  • where you first generate ui2 to show it fails the validation check... you might want to say it is because it assumed a default value of 0 (since none was provided) for the window_size.
  • %answer key/03b/05.py is missing an answer. :)

• In 03c_other_dashboard_widgets.ipynb might be better called 03c_other_dashboard_elements.ipynb. In any case:

  • I would match the axes names, titles, and legend as shown in PR#8, its low-cost and clarifies the plot quite a bit.
  • Care to explain the @tr.observe('smoothing_info') decorator on select_and_plot method of DataAndPlot?
  • You do NOT use ipydatagrid when displaying the data table with DataAndPlot. Might be nice to do that for consistency with the "02" notebooks.

• In 03d_dashboard2.ipynb:

  • when I executed the cell with main_widget in it, NOTHING was displayed! This is intentional based on the text, but a little disconcerting... Not even a white bar to indicate the cell executed...
  • When I tried to nb_export the notebook, I got a strange error FileNotFoundError: [Errno 2] No such file or directory: '03d_deashboard2.ipynb' because there is a typo, it should be '03d_dashboard2.ipynb' not '03d_deashboard2.ipynb'

[JuanCab]

03e_more_ipyautoui_pydantic.ipynb contains the misspelt SmoothingSetttings instead of SmoothingSettings (where I stutter less with the t).