Skip to content

GAP_ANALYSIS.md

Latest commit

 

History

History
581 lines (535 loc) · 29.7 KB

GAP_ANALYSIS.md

File metadata and controls

581 lines (535 loc) · 29.7 KB

Cypher TCK gap + workaround analysis

This document tracks feature gaps between the openCypher TCK and the current GFQL-based conformance harness, along with any temporary workarounds and translation guidelines.

How to use this document

  • When adding or updating a scenario, check if it introduces a new gap.
  • If a scenario is marked skip/xfail, add or reference a gap entry below.
  • Keep the entry short, with a clear root cause and a proposed path forward.
  • Update the entry when the gap is partially or fully addressed.
  • Targettable capability tags (e.g., target-table-ops, target-expr-dsl) are auto-applied to xfail scenarios by feature path for tracking near-term work.

Entry format

  • ID: Short identifier (G1, G2, ...)
  • Status: Open | Partial | Closed
  • Description: What is missing or incorrect
  • Affected scenarios: Scenario keys or TCK paths
  • Workaround: How we currently cope (xfail/skip/translation hack)
  • Next steps: What work would close the gap

Workarounds in use

  • Label matching: Store labels as lists in a labels column and expand to boolean columns label__<name> in the harness. Translate label predicates as n({"label__A": True, "label__B": True}).

Gaps

G1: Cartesian product + projection results

  • Status: Open
  • Description: The harness does not compute cartesian products across multiple MATCH clauses or comma-separated patterns, nor validate row-level projections in RETURN.
  • Affected scenarios: match1-5 (Match1 [5]), match-where3-1, match-where3-2, match-where4-1, match-where4-2, match8-1
  • Workaround: Mark as xfail with expected rows captured in the scenario.
  • Next steps: Add a result comparator for row projections and introduce a cartesian product mechanism (likely outside GFQL core) for multiple MATCH clauses.

G2: Scenario-level row expectations

  • Status: Partial
  • Description: Scenario Expected.rows is recorded but not validated by the runner. The harness can now validate alias-tagged node IDs for single-node returns, but does not compare row-level combinations or projections.
  • Affected scenarios: Any scenario that only specifies rows, or uses RETURN expressions beyond node/edge identity.
  • Workaround: Use node_ids/edge_ids or return_alias when possible; otherwise xfail and note row expectations.
  • Next steps: Extend the runner with row extraction and normalization (alias binding, ordering rules, null handling, projection comparisons).

G3: CREATE parser coverage

  • Status: Partial
  • Description: The minimal CREATE parser handles nodes, labels, simple properties, basic relationships, chained relationship patterns, and simple relationship properties, but does not parse variable-length relationships or advanced Cypher constructs (MATCH/DELETE/UNWIND or computed property expressions) in setup scripts.
  • Affected scenarios: match4-4, match5-25, match5-26, match5-27, match5-28, match5-29, plus any scenario whose setup includes complex patterns.
  • Workaround: Manually craft fixtures or extend parser incrementally.
  • Next steps: Support variable-length relationships and additional pattern forms as needed by the next wave of scenarios.

G4: Parameter binding

  • Status: Open
  • Description: The harness does not support Cypher parameters (e.g. $param) for query execution or comparison.
  • Affected scenarios: match-where1-6, match-where1-9, match-where2-2, return6-17, return-orderby6-1, return-skip-limit1-2, return-skip-limit1-6, return-skip-limit1-8, return-skip-limit2-10, return-skip-limit2-11, return-skip-limit2-14, return-skip-limit2-15, return-skip-limit3-2, with6-5, with-where2-2, with-skip-limit3-2, unwind1-6
  • Workaround: Mark as xfail and capture expected rows in the scenario.
  • Next steps: Add parameter injection support (scenario metadata -> GFQL predicate substitution) and validation for edge-return scenarios.

G5: Disjunctive WHERE predicates (OR)

  • Status: Open
  • Description: The harness does not support OR predicates across node or relationship properties/types.
  • Affected scenarios: match-where1-10, match-where1-11, match-where4-2, match-where5-4, with-where4-2, with-where7-3
  • Workaround: Mark as xfail and capture expected rows in the scenario.
  • Next steps: Add predicate combinators or explicit OR support in the translation layer and runner comparison.

G6: Path variables + length()

  • Status: Open
  • Description: Path variables and path-length predicates (e.g. length(p)) are not supported in GFQL translations or the harness.
  • Affected scenarios: match-where1-12, match-where1-13, return6-8, return6-13, with1-4, with6-4
  • Workaround: Mark as xfail and capture expected node sets.
  • Next steps: Define a path representation in GFQL and implement length checks in the reference enumerator + runner.

G7: Compile-time validation coverage

  • Status: Open
  • Description: The harness does not model Cypher compile-time validation errors (e.g., invalid path property predicates, aggregation in WHERE).
  • Affected scenarios: match-where1-14, match-where1-15, match3-29, match3-30, match4-9, match4-10, match6-21, match6-22, return1-2, return2-18, return4-10, return6-14, return6-15, return6-20, return6-21, return7-2, with4-4, with4-5, with6-8, with6-9, with-orderby1-46-1..with-orderby1-46-10, with-orderby2-25-1..with-orderby2-25-25, with-orderby3-8-1..with-orderby3-8-30
  • Workaround: Mark as xfail with explicit syntax-error reasons.
  • Next steps: Map Cypher error classes to GFQL validation and assert exception types in the runner.

G8: Multi-variable WHERE predicates

  • Status: Open
  • Description: The harness cannot yet translate conjunctive WHERE predicates that bind and filter multiple variables across a complex MATCH pattern (e.g., multiple node aliases, re-used variables, multi-hop joins).
  • Affected scenarios: match-where2-1, match-where2-2, with-where2-1, with-where2-2
  • Workaround: Mark as xfail and capture expected rows/node sets in the scenario metadata.
  • Next steps: Extend the translation layer to support per-alias predicates and multi-pattern binding, then add row-level validation to assert projected property outputs.

G9: Variable comparison joins

  • Status: Open
  • Description: The harness cannot express comparisons between variables (e.g., a = b, a <> b, a.id = b.id, n.animal = x.animal, x.val < y.val), which requires join semantics across bindings and row-level projection validation.
  • Affected scenarios: match-where3-1, match-where3-2, match-where3-3, match-where4-1, match-where6-5, match-where6-6, match-where6-7, match-where6-8, match7-11, match9-8, return6-13, with-where3-1..with-where3-3, with-where4-1
  • Workaround: Mark as xfail and capture expected rows in the scenario metadata.
  • Next steps: Add join-aware translation support (variable comparison predicates) and extend the runner to compare projected rows.

G10: Pattern predicates + variable-length relationships in WHERE

  • Status: Open
  • Description: Pattern predicates used as boolean filters in WHERE (e.g., (a)-[:T]->(b) or (a)-[:T*]->(b)) are not supported, nor are variable-length relationship predicates.
  • Affected scenarios: match-where4-2, with-where4-2
  • Workaround: Mark as xfail and capture expected rows in the scenario metadata.
  • Next steps: Add translation support for pattern predicates and variable-length relationship filters, then validate projected rows.

G11: Comparison predicates + null semantics

  • Status: Open
  • Description: The harness cannot express comparison predicates (e.g., >, <), label predicates in WHERE (e.g., i:TextNode), or null checks (IS NOT NULL) with Cypher's tri-valued logic.
  • Affected scenarios: match-where5-1, match-where5-2, match-where5-3, match-where5-4, match7-25, match9-8, with-where5-1..with-where5-4
  • Workaround: Mark as xfail and capture expected rows in the scenario metadata.
  • Next steps: Add predicate operators and null-handling semantics to the translation layer and runner comparisons.

G12: OPTIONAL MATCH semantics

  • Status: Open
  • Description: The harness does not model OPTIONAL MATCH row preservation, null propagation, or WHERE filtering behavior for optional patterns.
  • Affected scenarios: match-where6-1, match-where6-2, match-where6-3, match-where6-4, match-where6-5, match-where6-6, match-where6-7, match-where6-8, match3-27, match3-28, match7-1..match7-31, match8-2, match9-8, match9-9, with1-5, with1-6, with-where1-3, with-where1-4
  • Workaround: Mark as xfail and capture expected rows in the scenario metadata.
  • Next steps: Add OPTIONAL MATCH semantics to the translation layer and extend the runner to compare projected rows with null handling.

G13: WITH/LIMIT pipeline semantics

  • Status: Open
  • Description: The harness does not support WITH clauses, LIMIT scoping, or mid-query variable pipelines.
  • Affected scenarios: match-where6-5, match3-24, match3-25, match3-26, match3-27, match3-28, match4-8, match6-18, match7-4, match7-5, match7-6, match7-10, match7-21, match7-22, match7-27, match8-1, match8-2, match8-3, match9-6, match9-7, return4-1, return4-11, return6-3, return6-13, return6-16, return6-18, return8-1, return-orderby4-1, return-orderby4-2, return-skip-limit1-3, return-skip-limit2-6, with1-1..with1-6, with2-1..with2-2, with3-1, with4-1..with4-7, with5-1..with5-2, with6-1..with6-9, with7-1..with7-2, with-where1-1..with-where1-4, with-where2-1..with-where2-2, with-where3-1..with-where3-3, with-where4-1..with-where4-2, with-where5-1..with-where5-4, with-where6-1, with-where7-1..with-where7-3, with-skip-limit1-1..with-skip-limit1-2, with-skip-limit2-1..with-skip-limit2-4, with-skip-limit3-1..with-skip-limit3-3, with-orderby1-1..with-orderby1-46, with-orderby2-1..with-orderby2-10, with-orderby2-11..with-orderby2-20, with-orderby2-21..with-orderby2-25, with-orderby3-1..with-orderby3-8, with-orderby4-1..with-orderby4-9, unwind1-3..unwind1-5, unwind1-7, unwind1-11..unwind1-13
  • Workaround: Mark as xfail and capture expected rows in the scenario metadata.
  • Next steps: Add pipeline support (WITH, LIMIT, variable scoping) and extend row-level validation for projections.

G14: Multiple relationship types in MATCH

  • Status: Open
  • Description: The harness does not support relationship type lists in MATCH patterns (e.g., :KNOWS|HATES).
  • Affected scenarios: match2-6, match3-8
  • Workaround: Mark as xfail and capture expected rows in the scenario metadata.
  • Next steps: Add relationship type set support in the translation layer and validate projected relationship outputs.

G15: Label predicates on relationship endpoints

  • Status: Open
  • Description: Label predicates on both sides of relationship patterns are not reliably supported; label boolean columns can trigger schema/type mismatches during multi-hop GFQL filtering.
  • Affected scenarios: match2-2, match3-6, match3-7, match3-25, match3-26, match7-23, match7-28, match9-5
  • Workaround: Mark as xfail and capture expected rows in the scenario metadata.
  • Next steps: Ensure label predicate columns remain boolean across GFQL chain operations and allow endpoint label filtering on relationship matches.

G16: Pattern-level variable binding

  • Status: Open
  • Description: The harness cannot enforce repeated node/relationship variables across pattern parts or MATCH clauses (self-references, cyclic patterns, and shared-variable joins).
  • Affected scenarios: match3-10, match3-12, match3-14, match3-17, match3-18, match3-19, match3-20, match3-21, match3-22
  • Workaround: Mark as xfail and capture expected rows in the scenario metadata.
  • Next steps: Add variable binding semantics across pattern parts (shared variables in a single pattern or across comma-separated patterns/MATCH clauses) and extend row-level validation.

G17: Variable-length patterns in MATCH

  • Status: Open
  • Description: The harness does not support variable-length relationship patterns (*, *min..max) in MATCH, including length bounds, zero-length paths, and variable-length relationship lists.
  • Affected scenarios: match4-1, match4-2, match4-3, match4-4, match4-5, match4-6, match4-7, match4-8, match5-1, match5-2, match5-3, match5-4, match5-5, match5-7, match5-8, match5-9, match5-10, match5-11..match5-29, match6-14, match6-15, match6-16, match6-17, match6-19, match6-20, match7-12, match7-13, match7-14, match7-15, match7-19, match7-20, match9-1..match9-9, return6-8, return6-13, with6-4, with-where4-2
  • Workaround: Mark as xfail and capture expected rows in the scenario metadata.
  • Next steps: Add variable-length matching support to GFQL translation and row-level validation for projected paths/lists.

G18: Named path returns

  • Status: Open
  • Description: The harness cannot return or validate named path values (e.g., p = (...) with RETURN p), including path ordering and formatting.
  • Affected scenarios: match6-1, match6-2, match6-3, match6-4, match6-5, match6-6, match6-7, match6-8, match6-9, match6-10, match6-11, match6-12, match6-13, match6-14, match6-15, match6-16, match6-17, match6-18, match6-19, match6-20, match7-16, match7-17, match7-18, match7-19, match7-20, match9-9, return7-1, with1-4, with6-4
  • Workaround: Mark as xfail and capture expected rows in the scenario metadata.
  • Next steps: Add path materialization support in GFQL and extend the runner to normalize and compare path outputs.

G19: Aggregations + boolean projections

  • Status: Open
  • Description: The harness does not support aggregations (e.g., count(*)) or boolean projections in RETURN (e.g., s IS NULL).
  • Affected scenarios: match7-29, match7-30, match7-31, match8-2, match8-3, match9-5, return2-10, return4-4, return4-6, return4-7, return4-8, return4-9, return4-11, return5-1, return5-3, return5-4, return5-5, return6-1..return6-13, return6-16..return6-19, return8-1, with4-6, with5-2, with6-1..with6-9, with7-2, with-where6-1, with-skip-limit1-2, with-skip-limit2-4, with-orderby1-45-1..with-orderby1-45-10, with-orderby2-22-1..with-orderby2-22-2, with-orderby2-23-1..with-orderby2-23-2, unwind1-4, unwind1-5, unwind1-12
  • Workaround: Mark as xfail and capture expected rows in the scenario metadata.
  • Next steps: Add aggregation support to the translation layer and extend the runner to compute and compare aggregated projections.

G20: IN predicate list membership

  • Status: Open
  • Description: The harness does not translate IN predicate list membership for node/relationship properties.
  • Affected scenarios: match7-17, match9-9
  • Workaround: Mark as xfail and capture expected rows in the scenario metadata.
  • Next steps: Add IN predicate translation to the GFQL predicate layer and update the runner to compare filtered rows.

G21: MERGE clause semantics

  • Status: Open
  • Description: The harness does not support MERGE clause write semantics or the ability to match-or-create nodes/relationships.
  • Affected scenarios: merge1-1..merge1-17, merge2-1..merge2-6, merge3-1..merge3-5, merge4-1..merge4-2, merge5-1..merge5-25, merge6-1..merge6-4, merge6-6..merge6-7, merge7-1..merge7-5, merge8-1, merge9-1..merge9-4, plus match8-2, unwind1-6
  • Workaround: Mark as xfail and capture expected rows in the scenario metadata.
  • Next steps: Add MERGE support to the translation layer (or precompute fixture states) and extend the runner to validate resulting graph state.

G22: Relationship list returns + list functions

  • Status: Open
  • Description: The harness cannot materialize relationship lists for variable-length matches, evaluate list functions (e.g., last(r)), or return list-valued relationship variables in projections.
  • Affected scenarios: match9-1, match9-2, match9-3, match9-4, match9-6, match9-7, match9-9
  • Workaround: Mark as xfail and capture expected rows in the scenario metadata.
  • Next steps: Add list materialization support for variable-length relationships, define list/relationship projection normalization, and implement list function evaluation in the runner.

G23: RETURN expression projections

  • Status: Open
  • Description: The harness does not evaluate projection expressions (RETURN/WITH/ORDER BY), including property access, arithmetic, list/map construction, label predicates, and literal expressions.
  • Affected scenarios: return2-1..return2-9, return2-11..return2-13, return3-1, return3-2, return3-3, return4-1..return4-9, return4-11, return5-1..return5-5, return6-1..return6-13, return6-16..return6-19, return7-1, return8-1, with-orderby1-23..with-orderby1-45, with-orderby2-1..with-orderby2-10, with-orderby2-11..with-orderby2-20, with-orderby2-21..with-orderby2-24, with-orderby3-1..with-orderby3-7, with-orderby4-1..with-orderby4-9, unwind1-5, unwind1-11, unwind1-13
  • Workaround: Mark as xfail and capture expected rows in the scenario metadata.
  • Next steps: Add RETURN expression evaluation and row-level projection validation in the runner (property access, arithmetic operators, list/map projections, label predicates, and literal expressions).

G24: DELETE clause semantics + deleted entity access

  • Status: Open
  • Description: The harness does not support DELETE clause semantics, side effect validation, or runtime errors for accessing deleted nodes and relationships.
  • Affected scenarios: delete1-1..delete1-8, delete2-1..delete2-5, delete3-1..delete3-2, delete4-1..delete4-3, delete5-1..delete5-9, delete6-1..delete6-14, plus return2-14, return2-15, return2-16, return2-17
  • Workaround: Mark as xfail and capture expected rows/side effects in the scenario metadata.
  • Next steps: Add DELETE support to the translation layer, track side effects, and validate deleted-entity access errors.

G25: ORDER BY semantics

  • Status: Open
  • Description: The harness does not support ORDER BY clauses, ordering guarantees across WITH/RETURN pipelines, or ORDER BY validation (scoping and aggregation restrictions).
  • Affected scenarios: return4-11, return-orderby1-1..return-orderby1-12, return-orderby2-1..return-orderby2-14, return-orderby3-1, return-orderby4-1..return-orderby4-2, return-orderby5-1, return-orderby6-1..return-orderby6-5, return-skip-limit1-1..return-skip-limit1-2, return-skip-limit2-2, return-skip-limit2-4, return-skip-limit2-5, return-skip-limit2-7, return-skip-limit2-8, return-skip-limit2-11, return-skip-limit2-15, return-skip-limit3-1..return-skip-limit3-3, with3-1, with4-6, with-skip-limit1-1..with-skip-limit1-2, with-skip-limit2-1, with-skip-limit2-4, with-skip-limit3-1..with-skip-limit3-3, with-orderby1-1..with-orderby1-46, with-orderby2-1..with-orderby2-10, with-orderby2-11..with-orderby2-20, with-orderby2-21..with-orderby2-25, with-orderby3-1..with-orderby3-8, with-orderby4-1..with-orderby4-9, unwind1-6
  • Workaround: Mark as xfail and capture expected rows in the scenario metadata.
  • Next steps: Add ORDER BY support to the translation layer and extend the runner to validate ordered outputs.

G26: DISTINCT semantics

  • Status: Open
  • Description: The harness does not support DISTINCT projections or DISTINCT handling inside aggregations.
  • Affected scenarios: return4-6, return5-1, return5-2, return5-3, return5-4, return5-5, return6-16, with5-1, with5-2, with-where1-2, with-where4-2, with-skip-limit1-1, with-orderby1-44-1..with-orderby1-44-2, with-orderby2-24-1..with-orderby2-24-2
  • Workaround: Mark as xfail and capture expected rows in the scenario metadata.
  • Next steps: Add DISTINCT handling to projection and aggregation paths and extend the runner to validate distinct row sets.

G27: SKIP/LIMIT semantics

  • Status: Open
  • Description: The harness does not support SKIP/LIMIT clauses or argument validation (including parameterized values).
  • Affected scenarios: return-skip-limit1-1..return-skip-limit1-11, return-skip-limit2-1..return-skip-limit2-17, return-skip-limit3-1..return-skip-limit3-3, with7-1, with-skip-limit1-1..with-skip-limit1-2, with-skip-limit2-1..with-skip-limit2-4, with-skip-limit3-1..with-skip-limit3-3, with-orderby1-1..with-orderby1-46, with-orderby2-1..with-orderby2-10, with-orderby2-11..with-orderby2-20, with-orderby2-21..with-orderby2-24, with-orderby3-1..with-orderby3-8, with-orderby4-1..with-orderby4-9
  • Workaround: Mark as xfail and capture expected rows in the scenario metadata.
  • Next steps: Add SKIP/LIMIT handling to the translation layer and enforce compile/runtime validation of arguments in the runner.

G28: UNWIND clause support

  • Status: Open
  • Description: The harness does not execute UNWIND clauses or list expansion semantics.
  • Affected scenarios: return-orderby1-1..return-orderby1-12, return-orderby4-1, return-skip-limit1-3, return-skip-limit2-1, return-skip-limit2-6, return-skip-limit3-3, with-orderby1-1..with-orderby1-22, with-orderby1-43..with-orderby1-45, with-orderby3-7-1..with-orderby3-7-10, unwind1-1..unwind1-13, union1-3, union2-3
  • Workaround: Mark as xfail and capture expected rows in the scenario metadata.
  • Next steps: Add UNWIND translation/execution support and row-level comparison for list projections.

G29: Temporal values and functions

  • Status: Open
  • Description: The harness does not parse or evaluate temporal functions and values (date, time, localtime, datetime, localdatetime) or their ordering semantics.
  • Affected scenarios: with-orderby1-11..with-orderby1-20, with-orderby1-33..with-orderby1-42, with-orderby1-45-6..with-orderby1-45-10, with-orderby2-11..with-orderby2-20
  • Workaround: Mark as xfail and capture expected rows in the scenario metadata.
  • Next steps: Add temporal literal/function support in the parser and expression evaluation/ordering layers.

G30: CALL clause + procedure support

  • Status: Open
  • Description: The harness does not support CALL clause execution, procedure registry/lookup, argument validation, or YIELD projections.
  • Affected scenarios: call1-1..call1-13, call2-1..call2-6, call3-1..call3-6, call4-1..call4-2, call5-1..call5-2, call5-3-1..call5-3-2, call5-4-1..call5-4-11, call5-5..call5-8, call6-1..call6-3
  • Workaround: Mark as xfail and capture expected rows in the scenario metadata.
  • Next steps: Add procedure registry stubs, CALL clause translation, and YIELD projection validation, then extend the runner to compare row outputs.

G31: UNION / UNION ALL semantics

  • Status: Open
  • Description: The harness does not support UNION/UNION ALL composition, distinct handling, or column alignment validation across union branches.
  • Affected scenarios: union1-1..union1-5, union2-1..union2-5, union3-1..union3-2
  • Workaround: Mark as xfail and capture expected rows in the scenario metadata.
  • Next steps: Add UNION clause translation, row normalization, and column alignment validation for UNION / UNION ALL branches.

G32: CREATE clause semantics

  • Status: Open
  • Description: The harness does not support CREATE clause write semantics (node/relationship creation, property/label side effects).
  • Affected scenarios: create1-1..create1-20, create2-1..create2-24, create3-1..create3-13, create4-1, create5-1..create5-5, create6-1..create6-10
  • Workaround: Mark as xfail and capture expected rows in the scenario metadata.
  • Next steps: Add CREATE support to the translation layer and validate side effects on nodes/edges/properties.

G33: SET clause semantics

  • Status: Open
  • Description: The harness does not support SET clause updates for properties or labels, nor side effect validation.
  • Affected scenarios: set1-1..set1-11, set2-1..set2-3, set3-1..set3-8, set4-1..set4-5, set5-1..set5-5, set6-1..set6-21
  • Workaround: Mark as xfail and capture expected rows in the scenario metadata.
  • Next steps: Add SET support to the translation layer and validate property/label updates with side effects.

G34: REMOVE clause semantics

  • Status: Open
  • Description: The harness does not support REMOVE clause updates for properties or labels, nor side effect validation.
  • Affected scenarios: remove1-1..remove1-7, remove2-1..remove2-5, remove3-1..remove3-21
  • Workaround: Mark as xfail and capture expected rows in the scenario metadata.
  • Next steps: Add REMOVE support to the translation layer and validate property/label removals with side effects.

G35: Expression evaluation (expressions suite)

  • Status: Open
  • Description: The harness does not evaluate expression-only queries from the TCK expressions suite (boolean/list/map/string/null/temporal/etc), so standalone RETURN expression semantics are not validated.
  • Affected scenarios: tck/features/expressions/** (keys expr-*)
  • Workaround: Mark as xfail with expected rows captured in scenario metadata.
  • Next steps: Implement expression evaluation + row-level comparison in the runner (including list/map/temporal semantics) and hook into GFQL execution.

G36: UseCases suite coverage

  • Status: Open
  • Description: The harness does not execute the TCK use case scenarios (counting subgraph matches, triadic selection), which depend on full pattern matching, optional matches, aggregation, and row-level projections.
  • Affected scenarios: tck/features/useCases/** (keys usecase-*)
  • Workaround: Mark as xfail with expected rows captured in scenario metadata.
  • Next steps: Implement row-level evaluation + optional match semantics and translate use case patterns to GFQL once bindings and projection checks are in place.

Notes

  • Keep this doc aligned with tests/cypher_tck/scenarios.py and plan updates in plans/cypher-tck-conformance/plan.md.

Coverage impact snapshot (tag-based)

Counts below are rough, tag-based xfail totals from tests/cypher_tck/scenarios.py. Tags overlap, so totals are not additive.

  • Expression evaluation (G35): 2,599 xfail scenarios tagged expr (temporal 1,069; list 219; string 32; typeConversion 47; boolean 150; comparison 84; aggregation 142).
  • Pipeline semantics (G13/G25/G26/G27): with 372, orderby 339, limit 241, skip 19, distinct 21 (all xfail).
  • Row projection + bindings (G1/G2/G23): match 138 xfail, return 134 xfail (row-level comparison and join/binding semantics missing).
  • Update clauses (G21/G32/G33/G34/G24): create 78, merge 77, set 53, remove 33, delete 45 (all xfail).
  • Procedures / union / unwind (G30/G31/G28): call 49, union 12, unwind 76 (all xfail).
  • UseCases suite (G36): usecase 30 (all xfail).

Implementation lift + risk notes

Heuristic guidance for staging work: prioritize high-lift, low-risk items that vectorize cleanly in pandas/cuDF. "Lift" reflects rough xfail counts above.

  • Row projection + bindings (G1/G2/G8/G9/G23): Lift high (match 138 + return 134 xfail) with spillover into many clause suites. Risk medium-high due to join semantics, alias scoping, null handling, and row order rules. Vectorization: join/merge + projection, but requires careful row normalization.
  • Pipeline semantics (G13/G25/G26/G27): Lift high (with 372, orderby 339, limit 241, skip 19, distinct 21). Risk medium: variable scoping and aggregation boundaries are tricky, but core operations map to sort/groupby/drop_duplicates/iloc.
  • UNWIND (G28): Lift medium (unwind 76). Risk medium: explode semantics are vector-friendly, but null/empty-list behavior and pipeline placement need care.
  • Expression evaluation (G35): Lift very high (expr 2,599) but risk high. Recommend staging: start with pure, deterministic scalar ops (literals, boolean/comparison, basic arithmetic) before lists/strings, then temporal and error semantics last.
  • OPTIONAL MATCH (G12): Lift high (touches many MATCH/WHERE scenarios) with high risk: left-join + null propagation across multi-hop bindings; sensitive to Cypher's tri-valued logic.
  • UNION (G31): Lift low-medium (union 12). Risk low-medium: concat + distinct with column alignment; vectorization straightforward once row projection normalization exists.
  • Update clauses (G21/G32/G33/G34/G24): Lift medium (node/edge creation and mutation: create 78, merge 77, set 53, remove 33, delete 45). Risk high: side effects, transactional semantics, and post-mutation validation.
  • CALL procedures (G30): Lift low-medium (call 49). Risk high: external procedure registry, signature/side-effect semantics.
  • UseCases suite (G36): Lift low-medium (usecase 30). Risk high: depends on OPTIONAL MATCH, aggregation, and row projections.