lutaml-xsd 1.2.110 → 1.2.112

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 16069a6293538146b585e29cf696a42e1294e9f8c211d075bcc8f03f44adcb8f
-  data.tar.gz: 1049c160550ac6905f09ecc512e06536137d9d4b13cade090cc83fc24b9c355c
+  metadata.gz: cdfdd257c5d78df578d997d512d39f17f9c83b10d4a8928708767b41bb6224b9
+  data.tar.gz: c9a0bc6226dd8346324c44af5f93906f851b477895b145ed86a89cc2a70f8538
 SHA512:
-  metadata.gz: 95556f997483a0c1b5b606c0047c7cfcb39f75b3dae29350a61b4613d78cfa7ec7c961fa4f6fe040e84000110aa92b1b87203cdb94cebcbc580ca6055f65c389
-  data.tar.gz: 689fa0a66e8cf18fe44bdaa7081095e631a1b49e6d67d44ba0f49edc6fcf1aaf35f5458ac0d12b901be4435240ecf9cd429b2b2bc7ff42dd94366dee3995db00
+  metadata.gz: f503a443b9e205299c93eadd6c4f878f8fe025a2d0bd0514629fcdd38945f533b948afa656aa3e57bd8b5bfc2ade4213f9c77380d5bcc395c279a23b247c1da6
+  data.tar.gz: 1f533117654e06d3e2acde45f3c053eb99c9c41030b9a844fa598482372db23e24029fe3eaf7514ce7033c8a4c94b9bcd11b62209d00e7ab0b56ccc9458e6e66

data/TODO.conformance.md

@@ -120,7 +120,7 @@ pinned commit) is measured via `rake conformance:w3c`:
 
 | mode | pass-rate |
 |---|---|
-| pipeline | 4617/4696 (98.32%) |
+| pipeline | 4644/4696 measured + 5 budget exclusions (98.96% of measured) |
 
 (2026-11-02, bundled-stack measurement with the escape,
 hint-following, pattern-alternative, identity-crash, engine-set,
@@ -144,7 +144,7 @@ lockfile refresh) → 4411 (93.9%) → 4430 (94.3%) → 4435 (94.4%) →
 → 4557 (97.0%) → 4565 (97.2%) → 4569 (97.3%) → 4562 (drift raw) →
 4563 (97.17%, ref-namespace round) → 4573 (97.38%, group-content
 round) → 4578 (97.49%, attribute-prohibition round) → 4585 (97.64%,
-multi-document round) → 4588 (97.7%, type-binding round) → 4592 (97.79%, model-161 ride) → 4611 (98.19%, engine-406 ride) → 4613 (98.23%, temporal-enumeration round) → 4614 (98.25%, simple-type-children round) → 4617 (98.32%, ic-ancestor-scope round), +656 net over the 1.2.43 baseline. The escapes round (PR #197 — \d→\p{Nd},
+multi-document round) → 4588 (97.7%, type-binding round) → 4592 (97.79%, model-161 ride) → 4611 (98.19%, engine-406 ride) → 4613 (98.23%, temporal-enumeration round) → 4614 (98.25%, simple-type-children round) → 4617 (98.32%, ic-ancestor-scope round) → 4637 (parent-context rounds + chameleon round 1.2.109) → 4637 (nested-type-candidates round, 1.2.110) → 4637 (redefine-cycle round, 1.2.111) → 4644 measured (98.96% of measured, train-adoption round: moxml 0.6.6 + leptris 1.9.418.0), +683 net over the 1.2.43 baseline. The escapes round (PR #197 — \d→\p{Nd},
 \w/\W/\s to their XSD value spaces) collapses Regex 120→69;
 same-step pattern facets now behave as alternatives and
 whitespace-only text reaches the facets intact (PR #200), Regex
@@ -560,26 +560,66 @@ tuple keys, node-set field violations and absent-field rows handled, cross-eleme
   regressions (stash-attributed per row), rspec 1663 green,
   rubocop clean. schR2 stack-overflow note unchanged (crash masked
   by expected-invalid, counted per convention).
-- SCH R2 STACK OVERFLOW (2026-11-11): Schema_w3c/schR2/schR2.i
-  (expected invalid) crashes the assessment with a SystemStackError
-  in lutaml-model 0.8.167's Base#resolved_element_order/
-  delete_deletables — an unguarded recursion over the cyclic
-  redefine graph schR2's documents build (called from our
-  collect_particles). The crash ESCAPES measure()'s rescue
-  (SystemStackError is not a StandardError); in every prior ledger
-  the case silently recorded as a pass via the crash-as-invalid
-  convention on its expected-invalid expectation, so the ledger
-  number never moved. Files/gems/ruby verified identical to the
-  baseline era; the crash reproduces in fresh single-case runs on
-  origin/main on both trains today. Finding for lutaml-model: a
-  cycle guard in resolved_element_order (our org's repo; owner
-  decides the upstream action).
+- SCH R2 STACK OVERFLOW FIXED (2026-11-12, round shipped): the
+  SystemStackError on Schema_w3c/schR2/schR2.i (expected invalid;
+  masked as pass by the crash-as-invalid convention) is OURS, not
+  lutaml-model's — the infinite recursion cycles between the
+  §4.2.2 redefined group body and the redefine TARGET (two DISTINCT
+  objects reachable by the same group name b-g2), so an OBJECT-keyed
+  guard misses it; the cut is NAME-keyed and PATH-scoped in both
+  walkers: leaf_particles marks the group name on its ancestry
+  (same name cannot expand within its own ancestry; sibling reuse
+  still expands fresh) and ContentModelMatcher#compile_group keeps
+  the same path discipline over resolved containers. LEDGER vs the
+  1.2.110 baseline: 4637/4696 measured + the same 5 reZ budget
+  exclusions, ZERO regressions, and batch 4450's crash list is
+  EMPTY (schR2.i now assesses honestly to invalid instead of
+  overflowing). PROVENANCE: batch 03000 was carried over from the
+  1.2.110 ledger — regex-pattern-only cases whose walk never
+  touches the guards (the batch's reI003 quadratic tax made a
+  re-run cost-prohibitive on the contended machine). Also filed
+  lutaml/moxml#540 this round (the 0.6.4 attribute char-ref
+  regression below; repro included).
+- TRAIN ADOPTION COMPLETE (2026-11-12, release pending): the
+  measurement train moves to moxml 0.6.2→0.6.6 + leptris
+  1.9.406.0→1.9.418.0 (moxml#540 CLOSED — the attribute
+  character-reference decode is verified restored: attr
+  `←` decodes to "←" on 0.6.6). SUITE 4644/4691 measured
+  (98.96% of measured) + the same 5 reZ budget exclusions (the
+  nested-repeat walks still exceed the 90s per-case watchdog on
+  418 — a mid-bisect "faster reZ" reading was sweep-churn noise).
+  WINS +11: attgD019/020/021/022/025/027/029/031/033/034 (the
+  strict-attribute class through chameleon/imported attributeGroup
+  refs — the moxml#522 thread, FIXED at last) and elemZ020
+  (leptris-rs#46, the substitution binding). REGRESSIONS −4:
+  wildZ013b/c (FILED leptris-rs#72 — strict anyAttribute admission
+  through namespace-less-import wildcards now reports "not
+  declared (strict)"; the flip is between 1.9.409.0 and 1.9.410.0,
+  bisected) and attZ014a/b (FILED leptris-rs#73 — the
+  multiple-xs:ID-per-element pair drifted to the XSD 1.1 reading;
+  suite-contested 1.0-invalid/1.1-valid, harness takes the 1.0
+  expectation). NOT carried by this train: the leptris-rs#45
+  stE050/055/065/066 legs and attgC035-038 (still failing on 418 —
+  the 407-era wins were lost somewhere in 408-418 and have not
+  returned). TOOLING LESSONS, both the hard way: (a) moxml 0.6.6
+  CANNOT drive leptris ≤ 1.9.409.0 — the adapter reports unusable
+  and silently falls back to nokogiri, so probes answer without an
+  engine; check for the fallback warning before trusting any
+  probe; (b) `bundle lock --update` RE-RESOLVES to latest and
+  silently undoes pin edits — pins go lockfile-version-edit +
+  `bundle install` + an installed-version verify, never
+  `bundle lock`; (c) the per-case sweep's verdict column stores
+  the RAW validity boolean — expected-INVALID rows reading false
+  are PASSES; the first assembly inverted them and manufactured 9
+  phantom reZ regressions before the correction.
 - TRAIN ADOPTION REGRESSION — CHARACTER REFERENCES (2026-11-11):
   the 2026-11-10 adoption verdict ("conformance-IDENTICAL") was
   WRONG for one class: moxml 0.6.4 stopped decoding numeric
   character references in attribute values. Minimal repro:
   `Moxml::Context.new.parse('<x a="&#x2190;"/>').root.attributes.
   first.value` → "←" on moxml 0.6.2, literal "&#x2190;" on 0.6.4.
+  FILED lutaml/moxml#540 (2026-11-12; repro: attribute values keep
+  the reference literal while element text decodes on 0.6.4).
   Downstream this breaks the whole UNICODE-BLOCK pattern-facet
   class — Regex_w3c Arrows/MathematicalOperators/
   ArabicPresentationForms-A/CJKCompatibilityForms/

data/lib/lutaml/xsd/validation/content_model_matcher.rb

@@ -225,7 +225,20 @@ module Lutaml
           container = Array(@group_particles.call(particle)).first
           return empty_fragment unless container
 
-          with_occurrence(particle) { compile_container_body(container) }
+          # §4.2.2 redefine bodies self-reference by name; the body and
+          # the redefine target are distinct objects, so the compile
+          # path set keys on the resolved container's identity — a
+          # container may not compile within its own ancestry, while
+          # sibling reuse still compiles fresh.
+          @group_path ||= []
+          return empty_fragment if @group_path.include?(container)
+
+          @group_path.push(container)
+          fragment = with_occurrence(particle) do
+            compile_container_body(container)
+          end
+          @group_path.pop
+          fragment
         end
 
         # A group ref expands to its declaration's container, whose own

data/lib/lutaml/xsd/validation/rules/content_model_validation_rule.rb

@@ -368,14 +368,31 @@ module Lutaml
             end
           end
 
-          def leaf_particles(particles, repository)
+          def leaf_particles(particles, repository, ancestors = nil)
+            ancestors ||= {}
             particles.flat_map do |particle|
               case particle
               when Lutaml::Xml::Schema::Xsd::Group
-                leaf_particles(expand_group_container(particle, repository),
-                               repository)
+                # §4.2.2 redefine bodies self-reference by NAME; the
+                # body and the redefine target are DISTINCT objects, so
+                # the path set keys on the group name — the same name
+                # cannot expand within its own ancestry, while sibling
+                # reuse still expands fresh.
+                group_name = if particle.name.to_s.empty?
+                               particle.ref.to_s
+                             else
+                               particle.name.to_s
+                             end
+                next [] if group_name.empty? || ancestors.key?(group_name)
+
+                ancestors[group_name] = true
+                body = expand_group_container(particle, repository)
+                leaves = body.nil? ? [] : leaf_particles(body, repository, ancestors)
+                ancestors.delete(group_name)
+                leaves
               when Lutaml::Xsd::Sequence, Lutaml::Xsd::Choice, Lutaml::Xsd::All
-                leaf_particles(collect_particles(particle), repository)
+                leaf_particles(collect_particles(particle), repository,
+                               ancestors)
               else
                 [particle]
               end

data/lib/lutaml/xsd/version.rb

@@ -2,6 +2,6 @@
 
 module Lutaml
   module Xsd
-    VERSION = "1.2.110"
+    VERSION = "1.2.112"
   end
 end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: lutaml-xsd
 version: !ruby/object:Gem::Version
-  version: 1.2.110
+  version: 1.2.112
 platform: ruby
 authors:
 - Ribose Inc.
 autorequire:
 bindir: exe
 cert_chain: []
-date: 2026-11-11 00:00:00.000000000 Z
+date: 2026-11-12 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: liquid