Validate IO

[liamhuber]

Layer type checking on top of ports.

At a high level:

• Don't allow ports with incommensurate types to connect (a simple prerequisite to work on Ontological connections #21)
• Show input readiness graphically (Colour-code output by readiness #67)
• Don't allow input to be manually modified if it's connected on the graph or if there's no good widget for its data type

Implementation TODOs:

• Modify DTypes
  • Import and inherit from ryvencore.dtypes.DType
  • Add valid_instance_classes
  • Add an __eq__ or match or something operator that compares class or instances
  • Implement a bunch of dtypes, overriding the matching criteria as needed (e.g. for Choice)
• Spoof Session, Script, and Flow as described above, so that you can add the match criterion to check_connection_validity whenever the input has a dtype
• Handle allowing None for dtypes
• Override NodeOutput to make it work with dtypes just like NodeInput
• Add create_output_dt to my Node
• Override setup_ports in my Node
• (De)serialize the new dtype fields
• DEBUG: get connections working again
• Add a front-end check to PortWidget to re-colour it when the value matches the dtype
• Add an input_ready method to Node that checks the match of all input ports
  • Use it in the Lammps node to avoid a run unless all input is present
    • Also add delete_existing_job=True for now, since we don't otherwise handle loading jobs nicely.
• Tests would be wise (not there for everything, but at least most of the new stuff)
• Update demo notebooks (old nodes might have some loading after this)

[github-actions]

👈 Launch a binder notebook on branch pyiron/ironflow/validate_io

[liamhuber]

Reminder to self: A quick end-of-day notebook test showed this introducing a nasty performance hit once it's actually rendered, even though the tests run as fast as ever. Try again after rebooting and if the problem persists, fix it.

Edit: performance is fine after rebooting.

[review-notebook-app]

Check out this pull request on 

See visual diffs & provide feedback on Jupyter Notebooks.

---

Powered by ReviewNB