SebastianGruza commented on issue #3090:
URL: https://github.com/apache/hugegraph/issues/3090#issuecomment-5702122711
Status after #3184 merged into master as `1a15e762` (2026-09-16), so that
this issue keeps a clear line between what is fixed and what is still open.
**Fixed by #3184 (mitigation, not the codec):**
`HstoreTable.queryByRange(Session, IdRangeQuery)` no longer pushes a
sysprop-only `ConditionQuery` to the store, and both `prepareConditionQuery*`
methods work on a copy instead of mutating the query core still uses. Effect
measured on 5 M edges with 4 sort keys: 13/13 sort-key prefix/range shapes pass
on hstore with id sets identical to rocksdb, against 8/13 store-side decode
failures on master before the merge (details in the PR). Every
`g.V(x).outE(l).has(<sort key>...)` traversal from a known vertex is therefore
usable on master with HStore.
**Still outstanding here:**
1. Item 1 of this issue, the property codec itself: the server writes raw
property values, the store-side `BytesBuffer.readProperty` expects a
cardinality header. Any path that still pushes a `ConditionQuery` and parses
rows on the store trips it. Known such paths: `HstoreTable.queryAll()` (both
exits) and the shard overload `queryByRange(Session, Shard, ConditionQuery)`
(`/traversers/{vertices,edges}/scan`), recorded in my comment above; measured
on the lab they returned rows on our data layout, so whether they fail is data
dependent, not proven safe.
2. The guard mismatch between `prepareConditionQueryList()`
(`containsLabelOrUserpropRelation()`) and `prepareConditionQuery()`
(`userpropConditions()`); no live caller today, to align in the codec fix.
3. Item 2 of this issue, the 22 excluded core tests with the inverted
`Assume` (F4 on our side): untouched by #3184.
4. The versioned sinking format proposed in item 1 has no owner.
If it helps, I can split 1-3 into a checklist in the issue description.
--
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]
