Feature Details
Hi @AgniveshChaubey , I’ve been going through several recent graph-related issues and PRs, and I’m noticing a recurring pattern around:
- layout inconsistencies
- handle positioning issues
- stale graph state after updates
It seems like many of these are symptoms of the same underlying problem rather than isolated bugs.
Instead of addressing them individually, I wanted to propose defining a small “graph determinism contract” to make the graph pipeline more predictable and consistent.
A rough outline could include:
- stable node/edge identity rules
- clear triggers for layout + collision recomputation
- consistent handle anchoring (especially around
$ref and $defs chains)
The idea would be to first align on these invariants, and then implement fixes against this shared contract rather than patching issues case-by-case.
If this direction sounds reasonable, I can:
- draft a short RFC / design note
- identify affected areas in the current pipeline
- follow up with small, focused PRs
Would really appreciate feedback on whether this aligns with the intended direction for the graph system.
Good First Issue
Contribution
Feature Details
Hi @AgniveshChaubey , I’ve been going through several recent graph-related issues and PRs, and I’m noticing a recurring pattern around:
It seems like many of these are symptoms of the same underlying problem rather than isolated bugs.
Instead of addressing them individually, I wanted to propose defining a small “graph determinism contract” to make the graph pipeline more predictable and consistent.
A rough outline could include:
$refand$defschains)The idea would be to first align on these invariants, and then implement fixes against this shared contract rather than patching issues case-by-case.
If this direction sounds reasonable, I can:
Would really appreciate feedback on whether this aligns with the intended direction for the graph system.
Good First Issue
Contribution