PatternFacet spelling; align ToldBNode/LexicalBNode traits with their definitions - #74
Open
ericprud wants to merge 2 commits into
Open
PatternFacet spelling; align ToldBNode/LexicalBNode traits with their definitions#74ericprud wants to merge 2 commits into
ericprud wants to merge 2 commits into
Conversation
Fixes the long-standing misspelling in the validation manifest (49 tests). Downstream harnesses that name this trait need the same one-word change, e.g. jena's ShexT.java tPaternFacet. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
ToldBNode: the focus node is a blank node.
LexicalBNode: the schema tests the lexical form of some blank node in
the data.
Verified with an instrumented validator (recording every string facet
evaluated against a blank node label) plus label-perturbation runs
(does the status change when blank node labels do?):
+ ToldBNode on 1focusMaxLength-dot_{pass-bnode-short,pass-bnode-equal,
fail-bnode-long}: their foci are _: nodes; their Length/MinLength
siblings already carry the trait.
+ LexicalBNode on 1focusMinLength-dot_pass-bnode-long: status flips
from conformant when labels shrink.
+ LexicalBNode on 1iriRefLength1_fail-bnode-equal,
1literalPattern_fail-bnode-match, 1iriPattern_fail-bnode-match:
each measures a facet against a blank node label (status is saved by
the nodeKind mismatch, but the lexical probe is the test's point, and
their siblings carry the trait).
- LexicalBNode off 1focusBNODE_dot_fail-iriFocusLabel-equal and
1focusBNODELength_dot_fail-iriFocusLabel-equal: no blank node appears
anywhere in these tests; the label under measure is an IRI's.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
ericprud
pushed a commit
to shexjs/shex.js
that referenced
this pull request
Aug 10, 2026
Runs the whole shexTest validation manifest twice - once through neighborhood-rdfjs over an in-memory store (the oracle), once through neighborhood-sparql over a SPARQL endpoint - and requires agreement modulo blank node labels. TEST_sparql=true gates it (npm run test-sparql); SPARQL_ENDPOINT=<url> targets an external store. The bundled endpoint (comunica in a worker thread, reached through an Atomics-parked synchronous HTTP bridge) is as hostile as the spec allows: blank node labels are drawn from a small rotating pool per response, and any query mentioning a label is rejected outright. The DECEPTICON - hand-written decoy topologies plus a mirror of every shexTest graph into its own namespace - is loaded beside every test's data; it is unreachable from any focus, so it must change nothing, but an unanchored description collects it and fails its denotation check. Skips are trait-driven (ToldBNode, LexicalBNode). The suite is green against current shexTest main (1180 passing, 30 skipped) and stays green once shexSpec/shexTest#74 tightens the tags (the three mistagged told-bnode-focus tests move from label-skips to focus-skips, and three lexical probes whose status the nodeKind gate saves become skips). External stores additionally skip, with the reason named, any test whose literals they refuse (QLever 400s ill-typed booleans) or rewrite (xsd:byte comes back xsd:int) - Endpoint#rejected and Endpoint#literalsMissing. Full sweep against native QLever with ad-freiburg/qlever#3190 applied: 713 passing, 0 failing, 454 skips with stated reasons. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
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
1 participant
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.
Two manifest-metadata fixes, found while running the shexTest validation suite differentially (in-memory vs SPARQL endpoint) in shex.js.
1.
sht:PaternFacet→sht:PatternFacetFixes the misspelling on all 49 tests that carry it. Downstream harnesses that name this trait need the same one-word change — known references:
jena-shex/.../runner/ShexT.javadeclarestPaternFacet(plus its vendored copy of this suite)bindings/emacs/tests/shextest_ranked_manifest.json); harness code treats traits as opaque strings2. Nine trait alignments
Semantics audited: ToldBNode = the focus node is a blank node; LexicalBNode = the schema tests the lexical form of some blank node in the data. Checked all 1189 tests two ways: an instrumented validator recording every string facet evaluated against a blank node label, and label-perturbation runs asking whether the reported status changes when blank node labels do.
Add
ToldBNode(foci are_:nodes; Length/MinLength siblings already tagged):1focusMaxLength-dot_pass-bnode-short,…_pass-bnode-equal,…_fail-bnode-longAdd
LexicalBNode:1focusMinLength-dot_pass-bnode-long— status flips from conformant when labels shrink1iriRefLength1_fail-bnode-equal,1literalPattern_fail-bnode-match,1iriPattern_fail-bnode-match— each measures a facet against a blank node label (the nodeKind mismatch keeps the status stable, but the lexical probe is the test's point, and their siblings carry the trait)Remove
LexicalBNode(no blank node appears anywhere in the test; the measured label is an IRI's):1focusBNODE_dot_fail-iriFocusLabel-equal,1focusBNODELength_dot_fail-iriFocusLabel-equalAfter these fixes the traits are exact: shex.js's SPARQL suite now skips purely by
ToldBNode/LexicalBNode(no content sniffing) and runs green — 1137 passing, 30 skipped-by-trait, over an endpoint that scrambles bnode labels per response. PyShEx'smanifest_tester.pyalready skips by these traits and picks up the three newly-tagged ToldBNode tests for free.validation/manifest.jsonldregenerated withbin/genJSON.js.🤖 Generated with Claude Code