[DrunkOnJava]

Summary

Brings the stale rows of docs/data/spec-coverage.tsv (and the page generated from it) up to what main measures: seven rows leave partial for closes, §22 AcDb:Template leaves missing for partial, §9 keeps closes with its evidence and a live tracker, and the "Last measured" line moves to the current main run.

Spec reference

ODA v5.4.1 sections §9, §22, §20.4.3, §20.4.4, §20.4.5, §20.4.46, §20.4.60, §20.4.68 and §20.4.75. Refs #127, #133, #140, #147, #148 (all stay open: no row here becomes complete).

Evidence for every state that moves is CI run 40532346172, the newest green ci.yml run on main (e7c53e1), artifact corpus-measurement; all its jobs passed, Corpus (public DWG) included.

Test plan

• python3 tools/ci/spec_coverage.py --write, then --check: spec-coverage: 139 rows, 135 required sections, 0 problems.
• python3 tools/oracle/spec_notes.py <oracle.log of the run> docs/data/spec-coverage.tsv run on a copy of the TSV changes nothing: every oracle note already matches the run's oracle.log (52 sections annotated, no cell differs).
• python3 tools/maintainer/train/arch.py check: clean.
• Shown on real files, no new unit tests. What the run says, from corpus-results.json: 52 drawings, 15,541 decoded, 533 skipped, 0 errored, and the per-kind errors map is empty on every drawing. None of TEXT, ATTRIB, ATTDEF, MTEXT, VIEW, DIMSTYLE or HATCH appears among the 28 skipped kinds. Every file of corpus-logs/coverage-*.txt reports err 0.

State changes (closes here means what the page defines: the decoder ends on its exact boundary on every corpus record of the type; no row is complete because R13 has no AutoCAD-written drawing, #345):

Section	Was	Now	What the row said, and what proves the move
§20.4.3 TEXT, §20.4.4 ATTRIB, §20.4.5 ATTDEF	partial	closes	The note quoted 29 TEXT records each on ACadSharp's R14, R2000 and R2004 saves and 4 on export_sample.dwg (R2018) as erroring (#226). #258 read the TEXT, ATTRIB and ATTDEF branches as §20.4.3 to §20.4.5 print them (118 erroring records close); the run has errored 0.
§20.4.46 MTEXT	partial	closes	The note quoted 36 records each on the R14 and R2004 saves as erroring. #261 reads the R2000+ line spacing and the R2004+ background block; the 72 records close (the CHANGELOG records the same); the run has errored 0.
§20.4.60 VIEW	partial	closes	The pre-R2007 records on ACadSharp's R14, R2000 and R2004 saves errored. #263 reads the section's pre-R2007 record and #284 the view mode (group 71); the run has errored 0 and the oracle's 6 records have 120 values equal, 0 different (unchanged).
§20.4.68 DIMSTYLE	partial	closes	The note quoted 3 records each on the R2007 and R2010 saves as erroring. #293 reads the BL reactor count of the R2007+ entry prefix and they close; #297 had decoded the 20 R14 records; the run has errored 0.
§20.4.75 HATCH	partial	closes	The note said the 8 R14 records are never routed. #264 routes the R14 class name (all 8 close, equal to the R2000 saves); HATCH is absent from the run's skip histogram and errors. The oracle does not score HATCH yet (48 paired records unscored); the row says so.
§22 AcDb:Template	missing	partial	dxf_header::measurement reads $MEASUREMENT from it and dwg-to-dxf writes it (#320). Measured departure recorded in the source: from R2007 the description length counts UTF-16 units where §22 counts bytes. probe_dxf_header over the 42 drawings with an export: 9,583 variables written, 9,410 agree, 173 hold a stated relation, 0 disagree, 0 extra (dxf-header-probe.log). Not exposed: the description text and the other fields. Not measured: R14 and R2000, whose template is §3.2.4's and is not read.

Not moved, on purpose:

• §9 AcDb:Header stays closes. It cannot be complete while R13 has no file. The note now carries the per-release evidence from header-probe.log of the run (8 of 8 drawings close on each of R14, R2000, R2004, R2007, R2010 and R2013, 4 of 4 on R2018, 0 disagree on every one), which is what closed Header variables, R2000 #128, Header variables, R2004 and R2007 #129 and Header variables, R2010 to R2018 #130; its tracking cell moves from the closed A table-driven header decoder with release ranges #126 to Header variables, R13 and R14 #127, which tracks R13. The DXF writer figures in the note are refreshed to the run's (9,583 written, 9,410 agree, 173 by relation, 0 disagree).
• §20.4.58 LTYPE stays partial: 4 values still differ from the export (LTYPE dash shape flag: the DWG sets 8 where the export writes 2 #242).
• §20.4.54 LAYER (true colour not surfaced, LAYER does not surface its true colour (R2004+) #245), §20.4.62 UCS (no corpus record, UCS reads its base-UCS handle inline and skips OrthographicType on R2000-R2004 (§20.4.62) #285) and the control objects (Control objects' num_entries disagree with the exported TABLE 70 on the ACadSharp saves #241) keep the gaps their notes name.

Page text: the "Last measured" line now reads 2026-10-27, main e7c53e1, with the run linked, and says the oracle notes match the run's oracle.log while a note's other numbers cite the PR or run that measured them. The paragraph that said closes came from a local 19-file corpus re-measured later "by the 0.2.0 measurement work" now says it comes from the Corpus CI job on the 52 public drawings (#225).

Not changed, though also behind: docs/release-1.0-criteria.md (its R3 row quotes 67 closes and 31 partial, and R8 quotes 308 erroring records; its "Last measured" line is 2026-10-21). A status there changes only in the PR that links the new evidence, so that is a separate change.

Source-provenance declaration

By submitting this PR I certify that my contribution does not incorporate executable code from any source whose license is incompatible with Apache-2.0. Specifically:

• No Autodesk DWG SDK source (RealDWG, ObjectARX, ObjectDBX) or NDA-protected Autodesk documentation was consulted or imported.
• No Open Design Alliance Teigha / Drawings SDK source was consulted or imported. The ODA's public Open Design Specification for .dwg files PDF is explicitly allowed (see CLEANROOM.md).
• No GPL-licensed DWG implementation source (LibreDWG, etc.) was consulted or imported. This matters because dwg-rs is Apache-2.0 and must remain compatible.

My contribution is based on:

• The ODA's freely-redistributable Open Design Specification for .dwg files (v5.4.1);
• Public reverse-engineering notes, research papers, or blog posts cited inline;
• Reading algorithm-description comments (not executable code) in permissively-licensed projects (MIT / Apache / BSD) to resolve spec ambiguities — disclose any such reading in this PR body so it can be recorded in CLEANROOM.md.

See CLEANROOM.md for the full allowed/forbidden source list and the honest scope of what "clean-room" means for this project.

Developer Certificate of Origin

By submitting this PR, I certify that the contribution complies with the
Developer Certificate of Origin 1.1.
Specifically, I affirm that the contribution is my own work (or rightfully
licensed to me) and that it can be distributed under the project's Apache-2.0
license.

I will sign off commits with git commit -s when asked, which adds a
Signed-off-by trailer to the commit message asserting the DCO.

[chatgpt-codex-connector]

Codex Review Summary

This comment shows the latest Codex review activity on this pull request.

Review	Status	Commit	Review trigger
📝 Code Review	✅ Completed 2026-10-28T16:38:27.586575Z	974d24d	New commits

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review" or "@codex security review".

Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings.