Repository navigation
fix(gfql): restore cuDF boolean alias shadows - #2272
Merged
Merged
Conversation
added 2 commits
November 1, 2026 02:04
CuDF shared the marker-only early return with generated aliases, so a real Boolean node property hidden behind a same-named node alias stayed as the alias marker. Restore now uses the same alias provenance rule as pandas; cuDF and polars-gpu lanes cover the repro and marker-only control. (#2271)
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
is_cudf_obj(table_df)early return so pandas and cuDF share one alias-provenance rule: only columns explicitly tracked as generated node markers are left as markers.Root cause
The pre-provenance restore logic intentionally skipped all Boolean-base cuDF frames because the code had no node-alias provenance to distinguish a user property from a generated marker. The #2263 fix added
_gfql_rows_node_aliases, but retained the broader cuDF bypass. On master,MATCH (kind:P) RETURN kind.kind AS xtherefore never restored userkind; the projection read the internal Boolean marker. The marker-only control still takes the absent-property route and remains null. Traversal/reentry controls already followed another route.Repro
MATCH (kind:P) RETURN kind.kind AS x[False, True][None, None][False, True]MATCH (kind:P) WITH kind RETURN kind.kind AS x[False, True][None, None][False, True]MATCH (a:P) RETURN a.a AS x[None, None][None, None][None, None]MATCH (kind:P) WITH kind MATCH (kind)-[r:X]->(b) RETURN kind.kind AS x[False][False][False]Validation
55cf673ff77d2e96d270379a86f9a768b959e19ec4fc6db8cbe33ac597f9bfb79703856861935be3(product code unchanged since9f6a5bde;c4fc6db8is test-only: the Polarsrow op 'select'decline branch now coverspolars-gputoo)test_missing_relationship_property_2262.py -k 'not cudf and not polars-gpu'— 4 passed / 4 deselected.test_alias_scoping_semantics.py— 130 passed / 120 expected GPU skips.bin/ruff.sh graphistry: PASS.git diff --check: PASS.07e6f544200c162d42f744dd687f77e6ebff8b90abcfbf66bbbdf09de5b3b1c8.GPU evidence (
graphistry/test-rapids-official:26.02-gfql, cuDF 26.02.01, identity-asserted, skips count as failures)c4fc6db8: 59/59 non-skipped checks green (14 path-filtered non-GFQL lanes skipped). tck-gfql 4289 passed; all 13gfql-routes-offledgers are empty.9f6a5bde: 124 ids, made of the 4 touched + all 120 cuDF/Polars-GPU ids intest_alias_scoping_semantics.py. The 120 match a master55cf673fbaseline: 118 passed, and the same 2 inherited failures (tracked in GFQL Cypher cuDF: 6 test_lowering cuDF cases fail on master (list/dynamic expressions raise host-conversion TypeError; cugraph write flips edge orientation) #2269). Both touched[cudf]cases passed.c4fc6db8: the 4 touched ids, 4 passed / 0 failed / 0 skipped.Closes #2271.
🤖 Generated with Claude Code