WAP 1.2.1 WBXML SCR Ledger

Version: v0.4 Status: effective SCR extracted; Class C applied; direct decoder evidence partially closed

Purpose

Define the exact WBXML 1.3 feature-level obligations for the selected WAP 1.2.1 Class C data client and distinguish a WBXML-capable browser from a host that merely invokes an unverified decoder executable.

The machine-readable authority is:

• spec-processing/source-manifests/wap-1.2.1-wbxml-scr.json
• spec-processing/source-manifests/wap-1.2.1-selected-normative-clauses.json

Validate it with:

node scripts/check-wap-wbxml-conformance-ledger.mjs
node scripts/check-wap-selected-normative-clauses.mjs
node scripts/check-wap-conformance-ledger.mjs

Effective authority

The normative target is:

1. WAP-192-WBXML-20010725-a, WBXML 1.3
2. WAP-192_105-WBXML-20011015-a, applied afterward

The approved SIN says the July 2001 SCR was not in the appropriate format and that part of WAP-192.101 had not been rolled into the updated specification. Its corrected section 9 table supplies the effective actor-specific WBXML-S-* and WBXML-C-* rows. Tracked-change inspection also resolves the effective wording for processor instructions and removed XML metadata.

The profile selection is:

• WAP-215 target: CCR-CLASSC-C-001
• selected feature group: WBXML:MCF
• WAP-221 meaning: all mandatory client features in the specification SCR

Effective totals

Scope	Count
Active WBXML SCR rows	15
Mandatory rows, all actors	11
Optional rows, all actors	4
Client/decoder rows	3
Server/document/encoder rows	12
Class C-required client rows	3
Rows outside the selected client profile	12

There are no optional client rows in the corrected table.

Selected Class C client rows

SCR	Effective feature	Code status	Primary work lane
WBXML-C-001	Binary XML Structure	partial	WML-203, R0-08, historical T0-07 linkage
WBXML-C-010	Encoding default attribute values	partial	WML-203, R0-08
WBXML-C-011	Binary/literal token equivalence for tags and attributes	partial	WML-203, R0-08, historical T0-07 linkage

The selected-row audit is:

• implemented: 0
• partial: 3
• missing: 0
• direct normative WBXML tests: 3
• boundary-only tests: 0

These are feature-level evidence counts, not a WBXML compliance percentage. The three rows expand into 47 client-applicable deduplicated clauses covering section 5 and its subsections plus sections 6.3 and 6.4. The WML-203 direct-evidence tranche records 42 fixed-outcome fixtures citing all 47 selected clauses and promotes all 47 to implemented. All three feature-level parent rows stay partial because their broader document-family limitations remain.

Current implementation evidence

transport-rust owns the boundary, consistent with the repository architecture:

• map_success_payload_response recognizes application/vnd.wap.wmlc;
• decode_wmlc invokes the built-in, pinned lowband-wml13-wbxml/0.3.0 decoder;
• WML token-table selection is keyed by the carrying MIME media type, and carrying-protocol charset metadata takes precedence over the WBXML header;
• decoder output and element nesting are bounded;
• header order, WBXML 1.3 version and WML 1.3 public identifiers, multi-byte integers (including legal leading zero-valued groups), supported charset termination, string tables, tag/attribute parser states, page-zero WML tokens, global strings/entities/extensions across unassigned pages, literal flags, PI and opaque structure, and malformed input have deterministic outcomes;
• the typed token registry accepts the full one-octet page index independently for tag and attribute spaces, resolves assigned page zero, rejects unassigned pages 1 through 254, and rejects unregistered implementation-specific page 255 deterministically;
• every default and fixed attribute in the selected WML 1.3 DTD is reconstructed before textual handoff;
• every assigned WML page-zero tag, attribute-start, and attribute-value token has a generated binary/literal equivalence pair;
• original WBXML bytes and the WMLC media type are preserved on success;
• failures use the structured WBXML_DECODE_FAILED path.

Direct source-derived evidence is in transport-rust/tests/fixtures/transport/wbxml_wml13/conformance.json and transport-rust/src/tests/wbxml_conformance.rs. The former permissive sample corpus now has fixed failure expectations under the pinned decoder and remains robustness-only evidence.

Cross-layer evidence is in the schema-v2 transport-rust/tests/network/interop/wdp_cdpd_ipv4_seed.json case selected_cdpd_wbxml_unitdata_round_trip and wml_203_reconstructed_wdp_sdu_matches_text_engine_behavior. The test reconstructs the opaque WDP delivery payload, assigns the WMLC media type only at the fetch boundary, and compares native engine state/rendering with the paired canonical text deck. pnpm test:story WML-203 exercises that paired text deck through the WASM engine boundary; it does not move WBXML decoding into the simulator.

There are no unpromoted selected client clauses. The source sentence tracked as WBXML-CL-CHARSET-UNREPRESENTABLE-NAME explicitly terminates tokenisation and therefore belongs to the unselected server/encoder profile under WBXML-S-001; it is preserved there as not-assessed, not counted as decoder evidence. WBXML-CL-TOKEN-CODE-PAGES is direct-evidence-backed by the bounded registry and fixed outcomes; this does not claim token tables that WML 1.3 does not assign.

The parent rows also retain broader limitations: externally supplied implied attribute values and non-WML document families. Generic application/vnd.wap.wbxml routing remains a separate WAE gap.

Server and encoder rows

The 12 WBXML-S-* rows cover binary-document structure, tokenisation, processing instructions, comments/declarations, parsed and unparsed entities, well-formedness, validation, default attributes, and literal attribute-name encoding.

They are retained for source completeness and for a future independently shippable encoder/gateway module. They are not selected by the current Class C client claim and have not been implementation-audited in this pass.

Remaining WBXML work

1. Extend engine structural parity beyond the focused canonical deck fixture.
2. Add generic application/vnd.wap.wbxml routing and non-WML token-table registries only if those document-family profiles enter selected scope.
3. Extend default-attribute reconstruction beyond WML 1.3 only if another version or document-type profile is selected.
4. Audit the 12 server/encoder rows, including unrepresentable-name tokenisation, only when that independently shippable profile is claimed.

Modern safety, streaming, performance, and diagnostics may improve the implementation. They cannot replace these strict decode outcomes or turn an unsupported token path into a silent success.

Source handling

The WAP-192 PDFs, private text extractions, and temporary page images remain outside Git pending redistribution approval. The repository stores only source identities, hashes, normalized requirement mappings, and audit evidence.