contrueCT commented on PR #2994: URL: https://github.com/apache/hugegraph/pull/2994#issuecomment-5977191254
@bitflicker64 Thanks for the concrete examples. I accept the observation that these wrappers currently trigger the conservative fallback, with possible full-scan and missing-index behavior changes. I will keep this round's requested wrapper expansion outside #2994. The scope boundary was already recorded in [the September 27 reply](https://github.com/apache/hugegraph/pull/2994#discussion_r4114120956) and [the follow-up split](https://github.com/apache/hugegraph/pull/2994#discussion_r4114578733): this PR preserves the source index/error policy across ordinary adjacency with supported suffixes; complex contexts retain conservative local filtering. Generic source-query completeness is handled independently by #3243, and selective pushdown within the fallback remains tracked by #3201. For clarity, `local`, `optional`, `coalesce`, `union`, `choose`, `map(traversal)`, `flatMap(traversal)`, and `math` are currently outside the admitted suffix set. In the reported query shapes, adding them can keep source property predicates local and change an indexed lookup or `NoIndexException` into candidate scanning. These are known compatibility and performance tradeoffs of the retained fallback. Extending admission needs a separate, bounded implementation and validation. A classifier-only probe using TinkerPop 3.5.1 also showed that adding just the proposed classes still rejects `union` and `choose` through their internal `EndStep` / `HasNextStep`; this was not a full HugeGraph runtime test. Please treat this expansion as follow-up work under #3201. For this PR, I will address reproducible defects that violate the declared supported-path contract, including result loss, incorrect predicate handling, or broken paging. Expanding support for additional explicitly conservative shapes should not keep reopening the merge scope. Your suggested alternative of explicitly documenting the exclusions is the approach I am taking here. -- 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]
