jrgemignani opened a new pull request, #2284:
URL: https://github.com/apache/age/pull/2284
NOTE: This PR was created with AI tools and a human.
Optimized id() and properties() field access on vertices and edges.
When accessing id(v) or properties(v) on a vertex, the system was generating
inefficient SQL that rebuilt the entire vertex agtype before extracting the
field:
age_id(_agtype_build_vertex(id, _label_name_from_table_oid(labels),
properties))
This forced full vertex reconstruction for every row, even though the data
was already available in table columns.
Added optimize_vertex_field_access() in cypher_expr.c to detect these
patterns and optimize them to direct column access:
- age_id(_agtype_build_vertex(id, ...)) → graphid_to_agtype(id)
- age_properties(_agtype_build_vertex(..., props)) → props
- age_id(_agtype_build_edge(id, ...)) → graphid_to_agtype(id)
- age_start_id(_agtype_build_edge(...)) → graphid_to_agtype(start_id)
- age_end_id(_agtype_build_edge(...)) → graphid_to_agtype(end_id)
- age_properties(_agtype_build_edge(...)) → props
Note: age_label() is intentionally not optimized due to complexity of
cstring-to-agtype string conversion.
Added regression tests in unified_vertex_table.sql to verify the
optimization works correctly for both vertices and edges.
modified: regress/expected/unified_vertex_table.out
modified: regress/sql/unified_vertex_table.sql
modified: src/backend/parser/cypher_expr.c
--
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]
