Incremental Computation Graphs
Web page · https://dziban.net/posts/incremental-computation-graph
First seen by Oriel on .
AI summary
The article explains how Yabu uses a Salsa-based computation graph to cache parsing, type-checking, and compilation, then invalidate only results affected by changed inputs.
- Mutable inputs, such as function source, feed pure computations whose results are memoized along with their dependencies.
- When computations call one another, the system records a runtime dependency graph; a changed type signature can invalidate dependent type-checking results.
- Memoization entries track their value, verified-at and changed-at revisions, and dependency revisions to determine whether cached results need rechecking.
- A red-green walk validates dependencies up to unchanged inputs, then updates verification revisions so subsequent evaluations in the same revision can skip repeated work.
- The Yabu example is simplified: it omits complications in the staging system and does not fully check how other functions type-check against a changed function.
#incremental computation #memoization #compilers #type checking
Generated by gpt-5.6-sol. Check the original for details.
Metric trends
Last 30 days · Up to 120 recorded samples per source and metric, with reposts kept separate · First and latest retained
Points
Score reported by the source.
- Lobsters: 13, latest observation ; +12 over observed period since
Comments
Comment count reported by the source.
- Lobsters: 3, latest observation ; +3 over observed period since
Aliases
| Source | External ID | URL |
|---|---|---|
| Lobsters | 6utrvz | https://dziban.net/posts/incremental-computation-graph/ Read discussion |