[0.1.x] pyodide 0.24.0, widget shim patches

[bollwyvl]

References

• fixes Option "Using a custom Pyodide distribution" (specified in jupyter-lite.json) not working #59
• alternative to Update to Pyodide 0.24.0 #61

Changes

• update pyodide version references
• rename repodata.json to pyodide-lock.json
• update for micropip 0.4.0
• update widget shims vs ipywidgets 8.1.1

Notes

• this is currently targeting 0.1.x
  • could target main, and then re-do a bunch of stuff
  • either way, a bunch of work on Support jupyterlite-core 0.2.0a0 #58

[github-actions]

👈 Try it on ReadTheDocs

[bollwyvl]

Looking fine on RTD.

[jtpio]

• could target main, and then re-do a bunch of stuff

We can ask Meeseeks to backport to main. Although #58 also seems to be updating widgetsnbextension so potential conflics.

[michaelweinold]

@jtpio, since you asked back in #61:

peewee (and sqlite3 in general) work fine now 🎉

from peewee import *

# Define a database object
db = SqliteDatabase('/tmp/my_database.db')

# Define a model class
class Person(Model):
    name = CharField()
    age = IntegerField()

    class Meta:
        database = db

# Create the table in the database
db.connect()
db.create_tables([Person])

This of course with the limitations already discussed in:

• Error when trying to use sqlite3 #35

[jtpio]

Thanks!

[jtpio]

@meeseeksdev please backport to main

[bollwyvl]

sqlite3 in general work fine now

my heart jumped a little until the disclaimer...

of course with the limitations already discussed

😿

[michaelweinold]

@bollwyvl, if you're interested in moving the sql filesystem issues forward, please get in touch with me at (email). We might have some financial resources available to hire someone on UpWork to work on this specific issue.

[bollwyvl]

For my purposes, public comms are preferred, but certainly we could set up a call on the Jupyter zoom or something if there are folk that want to talk about it... i'm in US/Central, but can be pretty flexible on times. Usually right before (or after, but now getting late for GMT+) the block of Notebook/Lab calls on Wednesdays.

The challenge is the potential developer pool for ServiceWorker/DriveFS support is... very small. It's been a fair piece since i tracked down a few of the remaining issues (probably before the split to a dedicated repo). My day job test developer instinct would be to fully instrument down to the (virtual) syscall level on both real CPython instance, one the MEMFS that does work (e.g. what you get in Firefox private browsing), and one against the ServiceWorker-backed one... but I don't have an effective test bench for doing that.