Xiao-zhen-Liu opened a new issue, #5881: URL: https://github.com/apache/texera/issues/5881
## Goal Bring the operator output port result cache (MVP) into `main` as a series of small PRs. On a re-run, an output port whose upstream computation has not changed reads its earlier result from storage instead of recomputing it. The code is written and working on a prototype branch; this issue tracks splitting it into reviewable PRs. Design and rationale: see Discussion #5880. ## Scope (MVP) A matched port's result is always reused (full reuse). There is no cost-based decision about whether to reuse, and no eviction or storage limits; those are future work and are not part of this merge. ## Key property When no output port has a matched result, the cache lookup returns nothing and the system behaves exactly as it does today. Each PR below can be reviewed on that guarantee: an empty cache reproduces current behavior. The feature can land inactive and turns on only once results are cached. ## PRs (sub-issues) In dependency order. PRs 1 and 2 are independent and can go in parallel. - [ ] PR 1 (#TBD): Storage foundation (cache table, read/write code, cache key) - [ ] PR 2 (#TBD): Cache state and statistics handling - [ ] PR 3 (#TBD): Scheduler (reuse planner, skeleton generation, schedule assembly) - [ ] PR 4 (#TBD): Turn the feature on (lookup, write, endpoints, cleanup) - [ ] PR 5 (#TBD): Frontend (cache panel and canvas display) Dependencies: 3 needs 1 and 2; 4 needs 3; 5 needs 4. ## Notes - Generated code (jOOQ for the new table, protobuf for the new state) is not committed in this repo; PRs include the source (`.sql`, `.proto`) and the build regenerates the rest. - The prototype branch has merge history and a few unrelated changes mixed in. Each PR will be re-created cleanly on a fresh branch off current `main`, not taken from the prototype branch as-is. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
