SECTION 24 completes at eleven tools with xref_attach, xref_manage,
image_attach, data_extract and drawing_export_dwg, on the four contract members
that genuinely differ per engine -- backends/contracts/refs.py. Everything else
in the sheet track composes the generic members and is written once.
The ActiveX members were read off the AutoCAD 2026 seat's own type library
(C:\Program Files\Common Files\Autodesk Shared\acax25enu.tlb, "AutoCAD 2025
Type Library" 1.0, the library the running 25.x seat registers) before a line of
the COM path was written, and the measurement is in the code beside the calls:
IAcadModelSpace.AttachExternalReference(PathName, Name, InsertionPoint,
Xscale, Yscale, Zscale, Rotation, bOverlay, Password)
IAcadModelSpace.AddRaster(imageFileName, InsertionPoint, ScaleFactor,
RotationAngle)
IAcadBlock: IsXRef (get), Path (get AND put -- so the repath is an assignment,
not a method), Reload(), Unload(), Bind(bPrefixName), Detach(),
XRefDatabase (get)
IAcadDocument.SaveAs(FullFileName, SaveAsType, vSecurityParams)
AcSaveAsType: ac2000_dwg=12 ac2004_dwg=24 ac2007_dwg=36 ac2010_dwg=48
ac2013_dwg=60 ac2018_dwg=64 (=acNative)
The running seat itself was not answering COM while this landed (its window was
Not Responding and GetActiveObject never returned), so the type library -- which
is the same seat's declared surface and reads without a call into the process --
is what the members are measured from; no document of the operator's was opened
or touched. This repository has been bitten four times by a fake that invented a
member, so the COM fakes in the tests model those names and nothing else.
ActiveX angles are radians, so the degree-to-radian conversion happens at that
boundary and nowhere else.
Headlessly, ezdxf.xref.attach carries the attachment (measured on ezdxf 1.4.4:
block flags 20 for an attachment and 24 for an overlay, so "is this an xref" is
flags & (BLK_XREF | BLK_XREF_OVERLAY) -- masking BLK_XREF alone misses every
overlay), add_image_def + add_image the raster, with Pillow reading the pixel
size so the placed image keeps the file's aspect.
Three capability keys, declared in both maps:
dwg_write -- measured absent here (odafc.is_installed() is False), so the
ezdxf entry is a runtime re-evaluation like viewport_render's
matplotlib check rather than a hardcoded False: installing the
ODA File Converter starts the call working with no edit to the
map. The refusal names the install route and drawing_export_dxf.
xref_live -- reload and bind need a live seat's xref manager and are refused
headlessly; list, detach and path are headless and are not
behind the key.
xlsx_write -- openpyxl is optional and absent here; CSV is always available,
so the refusal never blocks the deliverable and says so.
Two numbers the plan predicted wrongly are corrected to what the registry
actually does. `block` outranks `drawing` in _GROUP_TAG_PRIORITY, so the two
xref tools file under `blocks` (9 -> 11) beside block_insert -- which is where
an xref belongs, being a block definition that lives in another file -- and
`drawing` moves 11 -> 12, not 11 -> 14. And `data_extract` must not say "parts
list" in its own text: that is the golden discovery query for
entity_create_table, and a file writer outranking the tool that draws the thing
is a regression the tuning bar catches. It is described by its destination --
a file -- instead.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01FKFStjEFzpGyLq5bMpWxTd
EnvironmentContract carries every member of the track E interface; the two
drawing_properties_* members are @capability('dwgprops') in this worktree only
and become @abstractmethod when group C's implementations merge.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01FKFStjEFzpGyLq5bMpWxTd
Tool count 131 -> 154; 474 -> 1242 collected tests.
The release exists for two measured reasons. A 149-tool catalog costs a client
~40,300 tokens before it has asked for anything, and a drafter searching the way
drafters think -- FILLET, BPOLY, QSELECT, WBLOCK -- scored df=0 against the stock
index, because those words were simply not in the data.
Discovery and batching
- DISCOVERY_MODE=search replaces the catalog with search_tools + call_tool,
ranked over an AutoCAD command and synonym corpus (discovery/aliases.py) on
top of fastmcp's BM25 index. Measured: 40,305 -> 356 tokens advertised.
- cad_batch runs a step list in one round trip; fields= projects result-heavy
responses. The `core` profile is gone -- discovery replaced it.
Honesty debt (Tier 0)
- OCS/ECS: `extrusion` appeared nowhere in this codebase, so mirrored and
rotated geometry reported coordinates that were quietly wrong. backends/ocs.py
translates at the backend boundary; every tool coordinate is WCS on both
engines, and a tilted plane refuses rather than answering with a number that
looks fine.
- drawing_save_as(fmt="dwg") wrote DXF bytes to a .dwg path. drawing_audit never
repaired. analysis_measure_area shoelaced points the caller typed in.
- ezdxf calls gained a per-call timeout, so one hung call can no longer wedge a
server whose document lock is a single asyncio.Lock.
Architecture
- The 1854-line backend ABC is 19 contract modules; @capability(key, reason=)
makes optional methods declarative and a gate fails on undeclared holes.
- 8 layout/viewport tools, CHSPACE, CAD_PROGID escape hatch.
- Wave A: selection filters, boundary tracing, hatch depth, annotation objects,
measurement -- 13 tools of a planned 39. The REGION and 2D-boolean family was
cut on evidence: add_region() yields zero ACIS bytes and the greiner_hormann
substitute loses 28.2% of the area on a square with one semicircular edge.
Verified against a live AutoCAD 2026, which found four COM defects no amount of
reading would have caught -- AcadPViewport has no ViewCenter, and the line that
set it had been swallowed by a bare except since v1.4.
Benchmark v3 (15 tasks) and what it caught
- v2 scored this server 10/10, which carried no information: every task in it
exercised something the server was built around. Five tasks were added because
they can fail, and three did while being written -- a hatch could not be
measured though the refusal listed HATCH as measurable, the boundary tools and
the measure tool disagreed about the same shape (139.2177 vs 139.2699), and a
circle stored as two bulged vertices measured 0.0.
- Competitor reports are NOT re-scored. They are pinned v1.4 runs against v2 and
were never asked these five questions; the chart shows those rows `not run`.
- Correctness A/B: 21/24 -> 24/24, zero regressions.
Performance, stated plainly
The published v1.4 perf report was recorded on CPython 3.14 and the new one on
3.11.15, so the apparent 2-6x speedup is the interpreter, not this release.
Re-measured on one interpreter, v1.5.0 is 28% slower creating entities and 22%
slower on the 10k roundtrip. Fully attributed: EZDXF_CALL_TIMEOUT=0 restores
v1.4.0's exact numbers. That is the deadlock fix being paid for, the knob is
documented, and winning it back is on the 1.6 roadmap.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
