A New Hope

[jstonge]

Inspirational work

• allo-montreal is a small project in observable framework where the backend is in python, and the charting is in javascript.
  • metadata.parquet.py is an example where we use Python as backend to produce the metadata of the web app.

Previous work

• Carter's py-allotax: is a proof-of-concept using python subprocess module to intercept the javascript code, and output it to Python environment.
• Template for generating d3 code: This is a proof of concept where we show how to interceptd3.js in python.
• allotaxonometer.js: current repo that is doing the 'backend' work.
• allotaxp is the svelte code for the repo, responsible for producing the data app. It depends on Allotaxonometer.js (this repo).
• allotaxp lines where we write SVGs. This is what we should intercept, somehow. If not in Svelte (could be vanilla javascript), we need to find another way to put the charts together into a single SVG string.
• allotaxp lines that does calculation (combine the two systems) from allotaxonometer.

Where we are

• [ ] Stripping out unnecessary stuff from allotaxonometer.js
  • [ ] what is unnecessary? The current idea is that allotaxonometer.js should only be the 'backend' (code to combine data).
  • [ ] Two issues with current strategy: piping large datasets to the JavaScript code and packaging the JavaScript code with the python package. Both are certainly solveable just haven't gotten there yet.
  • [ ] #19
  • [ ] #20
  • [ ] #10 - add documentation for users (disambiguate among repos and what they're doing), highlighting tools per use case and extensibility. This issue depends on where CLI tool will live. If we make it a python package, documentation will live there for CLI. Different use case documentation will live here.
  • [ ] #21

The dream

• [ ] minimalist python package that obscures node.js dependencies.
• [ ] allotax_plot.py -alpha -csv1 -csv2 -"system 1 name" -"system 2 name"

[amafehr]

Added an issue to discuss organization, where this functionality should live, where code should be pulled from. As a newbie coming in, it seems pretty distributed but some of it makes sense with more documentation.