Vendor W3C MathML 4 grammar (marked modifications); rework metanorma-mathml wrapper

grammars/basicdoc.rnc

@@ -1068,9 +1068,10 @@ latexmath = element latexmath { text }
 ## Encoding of AsciiMath
 asciimath = element asciimath { text }
 
-## Encoding of MathML: the official W3C MathML 3.0 grammar (namespace
+## Encoding of MathML: the official W3C MathML 4 grammar (namespace
 ## http://www.w3.org/1998/Math/MathML), via the metanorma wrapper in
-## grammars/mathml/. See grammars/mathml/README.adoc and basicdoc-models#35.
+## grammars/mathml/. See grammars/mathml/README.adoc, basicdoc-models#35,
+## and the MathML 3 -> 4 migration: basicdoc-models#39 / metanorma#585.
 mathml = external "mathml/metanorma-mathml.rnc"
 
 StemAttributes =

grammars/mathml/README.adoc

@@ -1,17 +1,48 @@
-= Vendored MathML 3.0 RelaxNG grammar
+= Vendored MathML 4 RelaxNG grammar
 
-`mathml3*.rnc` are the **unmodified** W3C MathML 3.0 RelaxNG Compact schema
-files, retrieved from https://www.w3.org/Math/RelaxNG/mathml3/ and used under the
+`mathml4*.rnc` are the W3C MathML 4 RelaxNG Compact schema files, retrieved
+from https://www.w3.org/Math/RelaxNG/mathml4/ on **2026-07-16** and used under
+the
 https://www.w3.org/Consortium/Legal/2002/copyright-software-20021231[W3C Software
-and Document License].
+and Document License]. They are generated by W3C from
+https://github.com/w3c/mathml-schema.
+
+IMPORTANT: unlike the MathML 3 vendoring (which was unmodified), these files
+carry **marked modifications**, each tagged in-place with a
+`METANORMA MODIFICATION (basicdoc-models#39)` comment preserving the upstream
+text. They fall into two classes: (a) the draft-generated grammar violates
+RELAX NG's duplicate-attribute restriction throughout the full-profile
+composition (jing rejects it outright: `data-other`, `dir`/`mathsize`/
+`mathvariant`, `id`, `alttext`, `maxsize`, `href`, `attributionsrc`,
+`displaystyle`, `encoding`); (b) the `anyElement`/`annotation-xml` combine
+re-opens and a nested-include override had to be adjusted so the metanorma
+wrapper's foreign-markup bans remain expressible for both jing and libxml2.
+Diff against the upstream snapshot to enumerate; re-audit all of them when
+re-syncing at CR/REC (upstream may have fixed the duplicates by then).
+
+Include graph: `mathml4.rnc` includes `mathml4-presentation.rnc` (which
+includes `mathml4-core.rnc`) and `mathml4-content.rnc` (which includes
+`mathml4-strict-content.rnc`).
+
+IMPORTANT: MathML 4 is, as of the retrieval date, a W3C **Working Draft
+nearing Candidate Recommendation** (the schema is generated "for MathML4
+drafts"). This snapshot is pinned as retrieved; **re-sync when MathML 4
+reaches CR, and again at REC** (metanorma/metanorma#585 caveat 2). The target
+is MathML 4 **full** — not MathML Core, which drops Content MathML.
 
 `metanorma-mathml.rnc` is the metanorma wrapper: it `include`s the W3C grammar
 (`start = math`) and augments it with metanorma's owned extension attributes
-(currently `data-metanorma-numberformat` on `<mn>`). `basicdoc.rnc` references it
-via `mathml = external "mathml/metanorma-mathml.rnc"`, so the wrapper's scope
-contains the W3C grammar and avoids any define-name collision with basicdoc.
+(currently `data-metanorma-numberformat` on `<mn>`), and bans the foreign
+markup escape hatches (`annotation-xml`, `anyElement`). `basicdoc.rnc`
+references it via `mathml = external "mathml/metanorma-mathml.rnc"`, so the
+wrapper's scope contains the W3C grammar and avoids any define-name collision
+with basicdoc. Note the documented v4 loosening the wrapper cannot subtract
+(embedded XHTML/SVG in token content via `textorHTML`) — see the wrapper's
+closing comment and metanorma/basicdoc-models#39.
 
-MathML version is **3** (v4 is not yet officially published). The authoritative
-list of `data-metanorma-numberformat` parameters lives in plurimath
-(`Formatter::Standard::DEFAULT_OPTIONS`); add new `data-metanorma-*` attributes to
-`metanorma-mathml.rnc` as they are coined. See metanorma/basicdoc-models#35.
+MathML version is **4** (migrated from 3: metanorma/basicdoc-models#39,
+umbrella metanorma/metanorma#585). The authoritative list of
+`data-metanorma-numberformat` parameters lives in plurimath
+(`Formatter::Standard::DEFAULT_OPTIONS`); add new `data-metanorma-*`
+attributes to `metanorma-mathml.rnc` as they are coined. See
+metanorma/basicdoc-models#35.