Add DodoSource

[nrbgt]

References

• continues Add more "well-known" Sources #44
• continues Create example app #7

Code changes

• add some more demo deps
  • ipydatagrid
  • ipytree
    • already quite busy, wouldn't add much
  • ipylab
  • updated ipywidgets and jupyterlab_widgets
    • update serialization for strict JSON
• add DodoSource
  • find nodes and lnks
    • tasks
    • files
    • subtasks
    • task dep
    • folders-of-files
• example notebook
  • make panel
  • load source
  • make grids
  • link grids and selection
  • run tasks
    • logs without killing UI
    • without blocking kernel
  • highlight
    • dependency paths following file_dep and task_dep links
    • upstreams
    • downstreams
  • collapse nodes and edges
• issues noticed while working on app
  • handle in-place updates (for certain cases)
  • handle linkHandlers and nodeHandlers better for facets
  • fix multiple shapes with "canary" attributes (e.g. Rectangle.width and Text.text)
  • add line_dash to more shapes (via HasStrokeAndFill)
  • fix center point of custom Rectangle and Ellipse shapes
• docs
  • changelog
  • clean up some hard-coded positions for facets in docs
  • update jupyterlite-core and pyodide-kernel

User-facing changes

• users of the binder demo will be able to try out the grid against an evolving system

• users of the lit demo will see:

  note	screen
  custom shapes
  running tasks based on task selection
  ipylab icons
  shortest path with particles
  upstream/downstream highlight

Backwards-incompatible changes

• n/a

[github-actions]

Try it on ⬅️ ReadTheDocs or Binder ➡️

[sanbales]

Looking good, just some small things I ran into

[sanbales]

Looks good to me, just a couple of small comments, but happy to merge if you are cool with it.

[sanbales]

Just wanted to double check this was still going in this PR...

[nrbgt]

yep, it's in here, and the demo shows some improvement in the stability of the render, though the vanilla forces still encourage stuff to move

[sanbales]

Can't tell the difference between these two. Only bringing it up in case it was an unintentional commit.

[nrbgt]

it has just been compressed.

[sanbales]

Ran a doit lab locally and played with the DodoApp.ipynb for a while.

Seems to be working well. Only things I've noticed (and these may be PEBCAK or not intended):

1. selection from datagrid to graph does not seem to work (i.e., selecting a row on the grids, does not highlight nodes or links on the graph)
2. sometimes the between particles do not seem to appear for things someone might expect (e.g., from env to setup)

Tried it on lite as well but didn't see the Dodo notebooks in there.

[sanbales]

Would the same happen with height?

[nrbgt]

it would visually hide it, but you'd still pay the canvas price

[nrbgt]

thanks for the review! 😍

didn't see the Dodo notebooks in there.

Yep, they don't work in lite because subprocess.Popen. Also doit doesn't import, due to some unvendored stdlib packages... to that end, I started some playground stuff on doitoml, but if course doit relies on subprocess.Popen as well, so only pure-python actions work.

the between particles

will look into it. Trying to get fancy with non-blocking stuff. Got some ideas on how that could be improved.

selection from datagrid

will look into it

[nrbgt]

selection from datagrid

will look into it

oof, this is general failure, even setting it programmatically doesn't work.

node_selection.selected = [0]
# doesn't highlight `all`

the issue is that we don't update the graphData.nodes._selected when a selection update comes in, and then node-selection thing is getting clobbered by the fancy text junk.

[nrbgt]

fixed selection on 43202f1

[nrbgt]

the between particles
will look into it.

Ah, this might due to it actually being a bit more accurate now, as it will only show particles if there is a path of the form:

[file] --> file_dep --> task --> targets --> [other file]

(Or the reverse)

Initially, it kinda lied, as it would follow sibling paths like:

[file] --> file_dep --> task --> file_dep --> [other file]

So basically: a lot of the nodes aren't particularly sensitive to other nodes. If there's a concrete example, happy to look into it!

[nrbgt]

I've updated the label to show the path count, a la

[sanbales]

the between particles
will look into it.

Ah, this might due to it actually being a bit more accurate now, as it will only show particles if there is a path of the form:

[file] --> file_dep --> task --> targets --> [other file]

(Or the reverse)

Initially, it kinda lied, as it would follow sibling paths like:

[file] --> file_dep --> task --> file_dep --> [other file]

So basically: a lot of the nodes aren't particularly sensitive to other nodes. If there's a concrete example, happy to look into it!

Yeah, that makes sense. No, I don't think it's worth it to go beyond what you've done. It's working nicely!

[sanbales]

I vote we pull the trigger. The only thing I could find was that when you try to open some files, it says it can't find them. Think it is for files that start with .? Obviously, testing this on 🪟ers, so 🐧 may be fine...