mbutrovich opened a new issue, #4705: URL: https://github.com/apache/datafusion-comet/issues/4705
Triage pass over the open `requires-triage` queue, per the project [Bug Triage Guide](https://github.com/apache/datafusion-comet/blob/main/docs/source/contributor-guide/bug_triage.md). - Total issues processed: 15 (12 triaged, 3 skipped, 0 failed) - Priority counts applied: `priority:critical` 1, `priority:medium` 7, `priority:low` 4 - Guide: [docs/source/contributor-guide/bug_triage.md](https://github.com/apache/datafusion-comet/blob/main/docs/source/contributor-guide/bug_triage.md) Labels have already been applied and `requires-triage` removed from each issue listed under "Triaged". A reviewer should spot-check the calls and close this issue when satisfied. To correct a label, edit the affected issue directly. ## Triaged ### priority:critical - [Bug] make_timestamp does not throw under spark.sql.ansi.enabled=true ([#4679](https://github.com/apache/datafusion-comet/issues/4679)) - Area labels: `area:expressions` - Rationale: under ANSI mode Spark throws on out-of-range arguments while Comet silently returns NULL, ignoring `spark.sql.ansi.enabled`; the guide classifies config-ignoring silent divergences as critical (decision-tree step 1). ### priority:medium - [Proposal] Scan I/O acceleration: node-local fragment cache, asynchronous prefetch, and cache-affinity scheduling ([#4695](https://github.com/apache/datafusion-comet/issues/4695)) - Area labels: `area:scan` - Rationale: net-new performance feature with a workaround today (re-fetch on every scan); missing-capability gap per decision-tree step 3. - pref: Use builtin compression for arrow ipc writer ([#4655](https://github.com/apache/datafusion-comet/issues/4655)) - Area labels: `area:shuffle` - Rationale: shuffle write-path performance improvement; current per-batch ZSTD framing works, so it is an optimization with a workaround. - Add optional native Lance scan support ([#4632](https://github.com/apache/datafusion-comet/issues/4632)) - Area labels: `area:scan` - Rationale: opt-in new reader; Lance tables already work through Spark's Lance reader, so this is a feature gap with a fallback. - perf: use aligned slice access in SparkUnsafeArray bulk append ([#4626](https://github.com/apache/datafusion-comet/issues/4626)) - Area labels: `area:shuffle` - Rationale: native shuffle bulk-append performance work; the existing per-element path is correct, so this is an optimization with a workaround. - [EPIC] Support Spark interval types (CalendarInterval / YearMonthInterval / DayTimeInterval) and interval expressions ([#4540](https://github.com/apache/datafusion-comet/issues/4540)) - Area labels: `area:expressions` - Rationale: unsupported types cause fallback to Spark for any interval-carrying query; missing-feature gap with a Spark fallback workaround. - [EPIC] Provide JVM/codegen-dispatch implementations for Incompatible expressions so they never fall back by default ([#4506](https://github.com/apache/datafusion-comet/issues/4506)) - Area labels: `area:expressions` - Rationale: feature work to keep Incompatible expressions native via codegen dispatch instead of falling back; functional gap with a workaround. - Gate non-default StringTypeWithCollation inputs on Spark 4.0 datetime expressions ([#4646](https://github.com/apache/datafusion-comet/issues/4646)) - Area labels: `area:expressions` - Rationale: non-default collations are silently treated as compatible and run natively; reporter tracked this as medium (split from #4502). See escalation note below. ### priority:low - Add `CONTRIBUTING.md` with link to the contributor guide ([#4694](https://github.com/apache/datafusion-comet/issues/4694)) - Area labels: none - Rationale: documentation/tooling task with no runtime impact (decision-tree step 4). - Move static support decisions from serde convert into getSupportLevel ([#4673](https://github.com/apache/datafusion-comet/issues/4673)) - Area labels: `area:expressions` - Rationale: behavior-preserving internal refactor of the serde framework; no user-facing effect (decision-tree step 4). - Discussion: remove incompatible native implementations that have a codegen dispatch fallback? ([#4654](https://github.com/apache/datafusion-comet/issues/4654)) - Area labels: `area:expressions` - Rationale: design discussion, no functional bug or regression; lowest bucket per decision-tree step 4. - Discussion: Should Comet add geospatial (ST_*) function support? ([#4455](https://github.com/apache/datafusion-comet/issues/4455)) - Area labels: none - Rationale: scope/design discussion, no functional bug; lowest bucket per decision-tree step 4. ## Escalations to consider - Gate non-default StringTypeWithCollation inputs on Spark 4.0 datetime expressions ([#4646](https://github.com/apache/datafusion-comet/issues/4646)) - Today these datetime expressions run natively for non-default collations where results "may diverge" from Spark, which matches the guide's decision-tree step 1 ("Can this bug cause silent wrong results?") and could justify `priority:critical`. Left at `priority:medium` because the reporter and #4502 tracked it as medium and the divergence only occurs with non-default collations on Spark 4.0; flagging for reviewer judgment. ## Skipped - needs more info - Bug triage results: 2026-06-11 ([#4625](https://github.com/apache/datafusion-comet/issues/4625)) - Prior triage summary issue, not a bug; left `requires-triage` in place for the reviewer to close. - Bug triage results: 2026-06-01 ([#4548](https://github.com/apache/datafusion-comet/issues/4548)) - Prior triage summary issue, not a bug; left `requires-triage` in place for the reviewer to close. - Bug triage results: 2026-05-26 ([#4441](https://github.com/apache/datafusion-comet/issues/4441)) - Prior triage summary issue, not a bug; left `requires-triage` in place for the reviewer to close. ## Notes - The guide lists `spark 4` as a pre-existing area indicator, but no `spark 4` label currently exists in the repo (only `spark sql tests` does). #4646 is Spark-4.0-specific and would otherwise have received it; left off and noted here rather than creating a new label. -- 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] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
