Remove personal names, local paths and fork-only process notes from code comments,
docs and the plugin manifest (repository now points at the upstream project), and
drop the handoff section from the plan.
Co-Authored-By: Claude Sonnet 5 <[email protected]>
What it is, how to check it on Linux or Windows without a Mac, the reading order,
the small edits to existing code, what each area proves, the decisions that are the
maintainer's, and what is deliberately not done.
Co-Authored-By: Claude Sonnet 5 <[email protected]>
Runs the plugin API tests, the plugin crate and Python model tests, and the host's
real-IPC plugin tests against a freshly staged plugin and built runner, and fails
if either environment variable target is missing so the tests cannot skip silently.
Co-Authored-By: Claude Sonnet 5 <[email protected]>
On Windows the temp directory has backslashes, which a Python string literal reads
as escape sequences (C:\Users is a unicode escape), so the RasterImage and OLE
lifecycle tests could not even create their entity. Windows accepts forward slashes.
Found by the first Windows run of the fork build workflow; Linux passed all 60.
Co-Authored-By: Claude Sonnet 5 <[email protected]>
Portable builds (executable plus plugins/opencad.python where the host looks for
it) and the real-IPC plugin tests on ubuntu and windows. Started by pushing a
fork-build-* tag or by hand; publishes nothing. release.yml stages the plugin
only in the macOS bundle.
Co-Authored-By: Claude Sonnet 5 <[email protected]>
A saved user preference (on by default) toggled by the SCRIPTCOMMANDS command
gates HostApi::run_command; Cancel stays allowed so a waiting command is never
stranded. The runner refuses the SCRIPTCOMMANDS command itself, so a script
cannot change it. Documents why no helper-thread or stack-growing dependency was
added: native builds link with at least 8 MiB, the nesting depth is fixed, and the
deepest step found needs 2 to 4 MiB in a debug build.
Co-Authored-By: Claude Sonnet 5 <[email protected]>
Drop two unneeded mut bindings in tests and mark the generated dict/enum
converters allow(dead_code), since a kind can leave one unused.
Co-Authored-By: Claude Sonnet 5 <[email protected]>
Rewrite the host-model design note for the current architecture (entities, tables,
commands, nested validation, known limits), refresh the plan's current-state header,
the ledger and gaps-doc status lines (cadcodec PR #48), the README status and the
superseded ocs.command notes, and drop dangling references to a sibling project that
is not in this repository.
Co-Authored-By: Claude Sonnet 5 <[email protected]>
polyline_close/open/width/reverse/join (a line or arc is turned into a polyline
first), array_path and array_3d, written against observed prompts and checked
for exact geometry in the real-IPC audit. Polylines default their pick point to
the first vertex. The audit runs on a large-stack thread because PEDIT's
line-to-polyline step needs more than a test thread's 2 MiB in a debug build.
Co-Authored-By: Claude Sonnet 5 <[email protected]>
The generated dict-to-struct conversions for nested records defaulted every
absent field and ignored unknown keys, so an LwPolyline built from
{'x': .., 'y': ..} vertices became zero-coordinate vertices. They now refuse an
unknown key (naming the keys the record takes) and a manifest list,
required_struct_fields, enforces the fields that define a nested record's
geometry (vertex locations, hatch edge ends and centres, section symbol points).
Regression test over real IPC; every existing lifecycle audit still passes.
Co-Authored-By: Claude Sonnet 5 <[email protected]>
Each is written against the prompt sequence observed in the exploration harness
and checked for exact geometry in the real-IPC audit. HATCH is left to
create_entity because its command needs an interactive boundary pick.
Co-Authored-By: Claude Sonnet 5 <[email protected]>
HostApi::run_command drives the real OCS command through the same primitives
the automation channel uses, synchronously on the host thread: run a line,
start a command and answer point/text/token/entity/selection/enter prompts, or
cancel (closing anything the command opened). Refused while another command is
active, for a denylist (QUIT, NEW, OPEN, SAVE*, PY_* ...) and off the active tab.
Python: doc.command, doc.start_command and doc.modify (offset, trim, extend,
fillet, move, copy, rotate, scale, mirror, erase). Real-IPC audit checks exact
geometry for every wrapper plus guards and undo; IPC and Python unit tests; docs.
Co-Authored-By: Claude Sonnet 5 <[email protected]>
TableOperation gains simple-linetype create/modify/rename/delete (standard names
taken, rename follows layers and entities, guarded delete) and paper-space
layout create/rename/delete/set-current/set-page (default page setup and sheet
viewport, the application's own layout switch, Model protected). CTAB now reports
the current layout. Real-IPC audit with DWG/DXF persistence and undo, IPC and
Python unit tests, docs.
Co-Authored-By: Claude Sonnet 5 <[email protected]>
BlockEntityAdd adds a validated entity to a block definition (owner set by the
host, same validation as a model-space entity, self-nesting inserts refused),
reusing the scene's block-editor routing. Deleting a block member now also
drops its handle from the block record. Real-IPC audit with an attribute
definition inside a block, DWG/DXF persistence and undo; IPC and Python unit
tests; docs.
Co-Authored-By: Claude Sonnet 5 <[email protected]>
TableOperation gains BlockCreate (copy entities into a definition shifted by the
base point, optionally erasing originals), BlockModify, BlockRename (inserts
follow) and BlockDelete (refused while referenced; removes members, markers and
orphaned draw-order tables). Real-IPC audit with DWG/DXF persistence and undo,
IPC and Python unit tests, docs; the DXF loss of the block description is pinned
by a canary.
Co-Authored-By: Claude Sonnet 5 <[email protected]>
Extends TableOperation with text style create/modify, dimension style
create/modify (DimStyle fields as a validated JSON object), rename, delete and
set-current. Reuses OCS's style in-use scan and reference-following rename
(four style_ops methods widened to pub(super)); a renamed text style also
updates dimension styles that name it. Real-IPC audit with DWG/DXF persistence
and undo, IPC and Python unit tests, docs. Four cadcodec findings are pinned by
canaries.
Co-Authored-By: Claude Sonnet 5 <[email protected]>
Adds the additive v7 TableOperation request and HostApi::table_operation for
layer create/modify/rename/delete/set-current. Refusals are decided before the
undo step is recorded. Python wrapper, real-IPC audit with DWG/DXF persistence,
IPC round-trip test, Python unit test and docs.
Co-Authored-By: Claude Sonnet 5 <[email protected]>
A scripted RasterImage now reuses one ImageDefinition per file and registers
it in ACAD_IMAGE_DICT, which owns it, as for an AutoCAD-attached image. The
reactor is deliberately not written because its owner convention cannot be
verified. An ignored spike records that OCS's own IMAGE command creates bare
images that lose their file path on save.
Co-Authored-By: Claude Sonnet 5 <[email protected]>
The shared driver now activates the layout that owns a viewport and requires
a drawn frame. The host assigns a unique id of at least 2 and derives
custom_scale as height / view_height on every scripted create or edit, as
MVIEW does, so custom_scale is read-only. A control showed the default id 0
is not invisible, so the id matters for uniqueness.
Co-Authored-By: Claude Sonnet 5 <[email protected]>
EmbedPicture builds the frame with OCS's own EmbeddedImage and
build_embedded_ole, so the opaque storage is never scripted. PNG, JPEG and
BMP are stored as read and other formats are re-encoded as PNG. The lifecycle
test now creates through Python; a new test covers formats, aspect, origin,
layer and refusals.
Co-Authored-By: Claude Sonnet 5 <[email protected]>
points is canonical; the host derives the point counts and refreshes the
end/tick/label projections with sync_display_fields when it changes, and
validates that style and view-representation references are existing class
objects. The lifecycle test passes every gate in DWG; DXF reopens the symbol
as an unknown entity because cadcodec dispatches SECTIONLINE only inside
blocks (pinned by a canary and already in the cadcodec write-up).
Co-Authored-By: Claude Sonnet 5 <[email protected]>
SolidOperation::Boolean runs synchronously, surfaces the kernel's refusal
with a plain reason and leaves the operands untouched, consumes them on
success unless keep_operands, and records one undo step. The lifecycle test
uses planar boxes with exact volumes and covers chaining, refusals, both
formats and undo; an ignored test shows a curved refusal reaching the script.
Co-Authored-By: Claude Sonnet 5 <[email protected]>
Times brep::combine in release against the pinned cadkernel: about 1-2 s for
curved operands, 1-2 ms for planar, versus 23 s in debug. Records the
kernel's frequent refusals, the host-thread and 30 s script-budget
implications, and recommends a synchronous Boolean operation that surfaces
refusals.
Co-Authored-By: Claude Sonnet 5 <[email protected]>
Adds SurfaceFromProfile and Extrude kernel operations (doc.solids.surface and
doc.solids.extrude), accepts surfaces in Transform, and maps Surface with
editable isolines. Fixes a real bug found by the test: the surface kind field
overwrote the entity-type kind key in Python dicts, so it is exposed as
surface_kind, with a catalog test guarding against name collisions.
Co-Authored-By: Claude Sonnet 5 <[email protected]>
The Python unit test's fake host omits the properties list the real host
always supplies. The coercion tolerates its absence, and a new test covers
tuples for point properties and lists of them. The previous commit slipped
past the Python tests because the gate did not include them.
Co-Authored-By: Claude Sonnet 5 <[email protected]>
RegionFromProfile reuses the REGION command's kernel calls; doc.solids.region
exposes it with layer and delete_source options. The lifecycle test builds
regions from real circle and polyline entities and covers refusals, a move,
both formats and undo. create_entity now accepts coordinate tuples for point
properties and every generated float setter accepts Python ints.
Co-Authored-By: Claude Sonnet 5 <[email protected]>
An additive HostApi::solid_operation builds cadkernel primitives and applies
rigid transforms, verifying each result lifts back losslessly and refusing a
payload it cannot lift. Exposed as doc.solids in Python; Solid3D, Body and
Region are mapped read-only with the opaque payload preserved. The lifecycle
test passes through the real runner with analytic volumes and exact extents as
oracles, DWG and DXF reopen with re-edit, and undo restores exactly.
Co-Authored-By: Claude Sonnet 5 <[email protected]>
OCS already creates, transforms and combines bodies through cadkernel, and
solid_to_sat / kernel_acis_body verify each result lifts back losslessly.
Two ignored spike tests measure volumes and DWG/DXF round trips. The
blocker is an operation channel, not geometry; the note recommends an
additive solid_operation host method and phases the work.
Co-Authored-By: Claude Sonnet 5 <[email protected]>
Frame corners and the aspect lock are scriptable; the opaque storage and
envelope are excluded and refused on edit. The audit embeds a real PNG
through OCS's own builder and proves it survives DWG and DXF byte-for-byte.
Python creation stays a follow-up.
Co-Authored-By: Claude Sonnet 5 <[email protected]>
Documents six cadcodec reader/writer gaps with source references, canary
tests and suggested fixes (nothing filed). Verifying them showed the Spline
DXF weight loss was an OCS bug: the writer emits weights only for a rational
spline, so the host now sets flags.rational from the weights. Spline passes
every gate.
Co-Authored-By: Claude Sonnet 5 <[email protected]>
Insert can be created by naming an existing ordinary block; the host refuses
unknown blocks, model/paper space and containment cycles, and keeps legacy
inserts of missing blocks editable. The legacy Polyline is update-only with a
hint toward Polyline2D and Polyline3D because both writers convert it on
save. The generator gains update_only_message.
Co-Authored-By: Claude Sonnet 5 <[email protected]>
Compares every writable ATTDEF field through DWG and DXF. DWG round-trips the
complete state; acadrust's DXF reader handles only eight group codes, so
alignment, width factor, oblique angle, attribute flags, field length,
generation flags, lock and text style reopen as defaults. Pinned by canary
expectations and recorded in the ledger.
Co-Authored-By: Claude Sonnet 5 <[email protected]>
Runs every first-generation kind through the real-runner lifecycle driver.
The audit found Ellipse unvalidated, zero-height Text/MText and non-finite
Arc angles and Point/Line thickness accepted, and no creation-time checks
for most kinds. validate_basic_entity now covers them and refuses only
problems an edit introduces. Twelve kinds pass every gate; Polyline,
Polyline2D, Polyline3D and Spline carry pinned persistence findings, and
Insert is update-only.
Co-Authored-By: Claude Sonnet 5 <[email protected]>
Picks up cadcodec 5b682ed and cadkernel 6f046af and the plotstyle
layer-usage speedup. The adapter's Cargo.lock now resolves acadrust to
5b682ed. All host, API and adapter tests pass and no acadrust canary
assertion tripped.
Co-Authored-By: Claude Sonnet 5 <[email protected]>
Region, Body, Solid3D and Surface need a kernel-backed create/transform API
that regenerates ACIS data; Ole2Frame needs a valid embedded-storage
fixture. No opaque payload is rewritten. This completes the ordered queue.
Co-Authored-By: Claude Sonnet 5 <[email protected]>
Light placement, intensity, colour, attenuation and shadow fields are
writable with host validation. staged_python_light_lifecycle_over_real_ipc
passes every gate except DXF cast_shadows, pinned by a canary.
SectionSymbol is recorded as blocked: it needs host-created style and view
representation objects and its raw point counts have no documented
derivation.
Co-Authored-By: Claude Sonnet 5 <[email protected]>
Bounds, scale and rotation are writable with viewport and scale reference
rules. staged_python_view_border_lifecycle_over_real_ipc passes every gate
in DWG; a DXF DRAWINGVIEW reopens as an opaque entity, pinned by a canary.
Co-Authored-By: Claude Sonnet 5 <[email protected]>
Curated view fields are writable; the owner must be a paper-space block and
frozen layers must exist. staged_python_viewport_lifecycle_over_real_ipc
passes every gate except a paper-space canvas oracle, so the kind is
recorded as integration-tested.
Co-Authored-By: Claude Sonnet 5 <[email protected]>
Scripts place an underlay against an existing UnderlayDefinition; the host
checks that the definition exists and matches the underlay type.
staged_python_underlay_lifecycle_over_real_ipc passes every gate except DXF
persistence: acadrust's reader returns the degree-valued rotation
unconverted and repeated saves compound it. Pinned by canary expectations.
Co-Authored-By: Claude Sonnet 5 <[email protected]>
Maps the placement, display and clip-boundary fields with rectangular and
polygonal clip validation. staged_python_wipeout_lifecycle_over_real_ipc
passes the full gate including DWG and DXF reopen with re-edit.
Co-Authored-By: Claude Sonnet 5 <[email protected]>
An earlier bulk edit for the RasterImage flags message also rewrote MLine's
test expectation; MLine.flags is a manifest override with its own message.
Co-Authored-By: Claude Sonnet 5 <[email protected]>
Scripts name an image file; the host reads its pixel size, creates and links
the ImageDefinition and derives size and the default clip boundary.
staged_python_raster_image_lifecycle_over_real_ipc passes every gate in OCS
with a real PNG. ACAD_IMAGE_DICT and reactor linkage are not created, so the
kind is recorded as integration-tested.
Co-Authored-By: Claude Sonnet 5 <[email protected]>
The host rebuilds the derived spline from the generating parameters on
scripted create/edit. staged_python_helix_lifecycle_over_real_ipc passes
every gate except DXF handedness, which acadrust's reader never applies;
the gap is pinned by a canary via per-format expectations in the shared
mesh lifecycle driver.
Co-Authored-By: Claude Sonnet 5 <[email protected]>
Maps all three mesh kinds with host validation of grid counts, face indices
and edge topology, and skips sub-record common data in the generator.
A shared run_mesh_lifecycle driver passes the full gate for each kind,
including DWG and DXF reopen with re-edit.
Co-Authored-By: Claude Sonnet 5 <[email protected]>
Maps rows, columns, merged ranges, style and placement through the generic
generator (now with tuple and usize support), adds grid/merge/reference
validation, and keeps DXF cell merge dimensions in step with merged_ranges.
staged_python_table_lifecycle_over_real_ipc passes the full gate including
DWG and DXF reopen.
Co-Authored-By: Claude Sonnet 5 <[email protected]>
Extends the adapter generator (Color, LineWeight, mixed unit/payload enums,
fixed float arrays, bitflags-2 newtypes, keep_types override) and maps
MultiLeader with a merge-style context patch, host validation and
reference checks. staged_python_multileader_lifecycle_over_real_ipc passes
the full gate including DWG and DXF reopen.
Co-Authored-By: Claude Sonnet 5 <[email protected]>
Color converts as {'kind': ByLayer|None|ByBlock|Index|Rgb, 'value': ...}.
Leader.override_color is now writable; neither DWG nor DXF stores it, which
is recorded as a fourth persistence blocker with a canary assertion.
Co-Authored-By: Claude Sonnet 5 <[email protected]>
Adds a hand-written flat Dimension model to the adapter (manifest
manual_kinds hook) and a synthetic-property section to the host coverage
policy. The host derives actual_measurement and the base definition point on
scripted create/edit. staged_python_dimension_lifecycle_over_real_ipc passes
the full gate for every subtype including DWG and DXF reopen. Python ints are
now accepted for coordinates and angles.
Co-Authored-By: Claude Sonnet 5 <[email protected]>
Maps vertices, flags, justification, normal, scale and style with host
validation and MLineStyle binding. The host recomputes vertex direction,
miter and element parameters on scripted create/edit, so scripts only supply
positions. staged_python_mline_lifecycle_over_real_ipc passes the full gate
including DWG and DXF reopen.
Co-Authored-By: Claude Sonnet 5 <[email protected]>
Adds Leader validation and annotation/dimension-style reference checks to
ocs_plugin_api, maps 15 writable and 5 read-only fields in the adapter, and
adds staged_python_leader_lifecycle_over_real_ipc covering C/R/E/D/U/I/V/P.
The W gate stays open: acadrust's DXF writer omits the annotation handle and
DXF reads drop annotation_offset; DWG R2010+ does not store text size or the
hookline flag. Each gap is pinned by a canary assertion.
Co-Authored-By: Claude Sonnet 5 <[email protected]>
Found while building an out-of-tree scripting plugin from scratch against
this doc and the actual ocs_plugin_api/host source. Two categories:
Fixed outright (docs/examples didn't match the real code):
- The worked CadModule example returned Vec<RibbonGroup> from
ribbon_groups(); the real trait requires &[RibbonGroup] - the example
wouldn't compile. Fixed with the OnceLock pattern the host's own
benchmark code already uses.
- Reconciled a contradiction between "process isolation removes the need
for a shared Rust ABI" and the exact-rustc_version requirement described
a few paragraphs earlier in the same doc.
- The roadmap's "Next" list still had external automation as unstarted;
--mcp/--serve are real and operational (src/cli.rs, src/mcp.rs).
- Added the missing plugin.toml to docs/plugin-template-v2 (its own
DESIGN.md already claimed one could be installed).
- Documented that plugin-template's CI floats rust-toolchain@stable, which
doesn't guarantee matching the host's exact rustc - added acadrust_source
to both templates' plugin.toml plus the release.yml step that fills it
from Cargo.lock's resolved acadrust source, alongside the existing
rustc_version fill.
Flagging for your call, not fixed here (found the behavior, not sure it's
the intended one, prefer to ask than guess at a fix):
- acadrust_compatible()'s ABI gate only runs at all if plugin.toml declares
acadrust_source - an undeclared field is silently treated as compatible,
unlike rustc_version (undeclared = incompatible). The same shortcut
appears in two places: ExternalPlugin::acadrust_compatible
(src/plugin/external.rs) and the inline check inside fetch_release_info
(src/plugin/marketplace.rs). If this is deliberate (e.g. back-compat with
manifests from before the field existed), no action needed beyond a
comment; if not, both call sites need the same fix.
- The command line still uppercases typed/pasted input even with the `>`
literal-spaces prefix (Message::CommandInput, src/app/update/mod.rs) -
`>` only stops a space from auto-submitting the line, it doesn't skip
`s.to_uppercase()`. A plugin command needing a case-sensitive typed
argument has no escape today short of automation.
- fetch_release_info (src/plugin/marketplace.rs) returns every release
under a linked repo with no filter on the manifest's own id field (which
it already parses to build ReleaseInfo) - a repo hosting more than one
plugin under tag-namespaced releases surfaces all of them mixed
together in one plugin's version list.
Co-Authored-By: Claude Sonnet 5 <[email protected]>
The ad-hoc codesign path signed the whole bundle with one --deep pass,
which never carried --entitlements to the QuickLook appex, so PlugInKit
silently rejected it as unsandboxed. Sign nested code individually
instead, matching the Developer ID path.
Separately, ThumbnailProvider drew the extracted preview at its native
pixel size anchored at the CGContext origin (bottom-left) instead of
scaling it to fill request.maximumSize, producing a small thumbnail
stuck in one corner instead of a properly sized, centered icon.
Note: Gatekeeper still rejects an ad-hoc or unnotarized Developer-ID
signature outright, and PlugInKit periodically re-validates and drops
any registered plugin that fails that check — so these fixes are
necessary but not sufficient for thumbnails to render from a build
that isn't both signed and notarized.
Co-Authored-By: Claude Sonnet 5 <[email protected]>
The DMG dropped just the bare .app with no indication of where it's
supposed to go, unlike the "drag the icon onto Applications" layout
every other Mac app uses. Stage the .app next to a symlink to
/Applications before building the image, then lay out the Finder
window's icon view (via a temporary read-write DMG, since a compressed
UDZO image can't be mounted for editing afterward) so both items are
visible side by side.
The Finder/AppleScript layout step is best-effort and fails gracefully:
this script mirrors the CI build-macos job, which likely has no
interactive Finder session for AppleScript to reach at all, so the
essential fix (the symlink itself, which doesn't depend on Finder
scripting) still lands even when the cosmetic pass can't run.
Fixes#769.
Co-Authored-By: Claude Sonnet 5 <[email protected]>
PropValue::ParamRow/ParamAddRow, the PropParamInput/Commit/Delete/AddNew
messages, their command_driver handlers (on_prop_param_commit/_delete/
_add_new, resolve_named_parameter_edit), and their render functions all
already existed — but nothing ever constructed the "Parameters" section
that would route data to them, so the whole inline add/rename/redefine/
delete workflow was unreachable from the UI. Adds that missing section
to the no-selection page in refresh_properties, alongside a new
PropValue::ParamsVisibilityToggle header row (mirrors the existing
show_constraint_values setting) for toggling whether constraint pills
show their driven value/parameter name.
Users can now manage named parameters inline in the Properties panel
(live resolved values, add/rename/redefine/delete) as an alternative to
the separate PARAMETERS modal, which continues to work unchanged.
Co-Authored-By: Claude Sonnet 5 <[email protected]>
Six one-shot alignment commands for the current selection —
ALIGNLEFT/ALIGNHCENTER/ALIGNRIGHT/ALIGNTOP/ALIGNVCENTER/ALIGNBOTTOM —
distinct from the existing point-picking ALIGN/ALIGN3D command: each
nudges every selected object (or complete group, via new
Scene::selected_object_units) so its bounding-box edge/center lines up
with the extreme edge/center already present in the selection, closer
to a word processor's "align objects" than AutoCAD's 3-point ALIGN.
Wired as plain one-click Tool entries in the Modify extension panel,
next to the existing ALIGN entry, matching that panel's flat-list
convention (no point prompts, so no dropdown submenu needed).
Co-Authored-By: Claude Sonnet 5 <[email protected]>
Constraint delete: a Properties-panel row delete button, viewport
click-to-select on a constraint glyph pill with Delete-key support, and a
right-click "Delete Constraint" menu item — all routed through one
undoable Scene::selected_constraint-aware path.
Fixes two named-parameter bugs found while testing that feature:
- DIMCONSTRAINT: a parameter named like a mode-switch keyword ("r", "d",
"a", "x", "y") could never be referenced, since the keyword check ran
before the parameter-name check.
- PARAMETERS modal Apply rebuilt Scene::named_parameters but never
refreshed the Properties panel, leaving renamed/deleted rows stale
until an unrelated refresh (e.g. a selection change) happened to occur.
Co-Authored-By: Claude Sonnet 5 <[email protected]>
value toggle, and redesigned constraint-ribbon icons
Properties panel:
- New "Parameters" section (shown with no selection) listing every named
parameter with inline name/formula editing, add, and delete, backed by
new PropValue::ParamRow/ParamAddRow and command_driver handlers
(resolve_named_parameter_edit, on_prop_param_commit/_delete/_add_new).
A fresh parameter's formula now defaults to "1", not "0" — a 0-length
seed value is a genuine solver singularity for a Distance constraint
driven by that parameter, with no defined direction to grow back out of.
- New "Constraints" section (shown for a single selected entity) listing
every constraint touching it as a clickable PropValue::EntityLink that
selects the linked entities in the viewport, danger-tinted when the
constraint conflicts.
Constraint pills:
- New Options > General toggle, "Show values and parameter names on
constraint markers" (on by default, persisted). Off, every pill shows
just its bare glyph instead of glyph+driven-value/parameter-name — the
glyph alone is enough to see a constraint is present, without the value
text covering nearby geometry on a dense sketch.
Constraint-ribbon icons:
- Replaced all 18 unicode/letter glyph icons (including the plain "F",
"L", "M", "S" letters) with proper SVGs under assets/icons/constrain/,
matching the Modify section's two-color style (#B4B6B9 white / #6DB7ED
blue, consistent stroke width). Tangent reuses the Draw section's
white-arc-plus-blue-dots motif.
Co-Authored-By: Claude Sonnet 5 <[email protected]>
An ocs_gcs Ellipse stores focus1 as an absolute point rather than an
offset from center, so a constraint that pulls only on center (e.g.
Concentric) left focus1's absolute coordinates untouched while center
moved — changing the derived focus1-center vector that major_axis and
minor_axis_ratio are computed from on write-back, so the ellipse visibly
rotated and reshaped even though nothing asked it to.
solve_scope now adds two Difference constraints per registered ellipse
pinning focus1.x-center.x and focus1.y-center.y to their pre-solve
values, so focus1 translates rigidly with center unless something
actually constrains it. Fixed's Ellipse arm no longer re-pins focus1
directly since that's now redundant with these rules.
Co-Authored-By: Claude Sonnet 5 <[email protected]>
The DWG writer truncates an XRecord Chunk to u8::MAX bytes, so a sketch's
constraint set or named-parameter table silently lost every entry on the
next DWG open once serialization passed ~255 bytes (found live: a
23-constraint demo round-tripped fine through DXF but came back empty
through DWG). Split the blob across multiple same-code 310 entries on
save and reassemble them on load, entirely on the OCS side.
Co-Authored-By: Claude Sonnet 5 <[email protected]>
None of the 17 constraint commands (HCONSTRAINT, VCONSTRAINT,
PCONSTRAINT, QCONSTRAINT, ECONSTRAINT, TCONSTRAINT, NCONSTRAINT,
NRCONSTRAINT, LCONSTRAINT, FXCONSTRAINT, SYCONSTRAINT, CCONSTRAINT,
EDCONSTRAINT, CPCONSTRAINT, MPCONSTRAINT, OCCONSTRAINT, DCONSTRAINT,
ACONSTRAINT) carried an inventory::submit! registration, so they were
invisible to command-line autocomplete, the MCP commands listing, and
the ALIASEDIT shortcut editor (which rejects a custom shortcut target
that isn't in this same registry). Each constraint module now
registers itself, with a regression test pinning the full set.
Co-Authored-By: Claude Sonnet 5 <[email protected]>
Arcs are now full constraint participants (center/radius via the existing
Circle-shaped solver path, plus real endpoints kept consistent by arc-rules
CurveValue constraints); adds Diameter/DistanceX/DistanceY/ArcLength/Normal
constraint kinds and initial Ellipse entity support (center-based
constraints only). DWG-native class mappings and gaps (ArcLength, Ellipse)
are grounded against real ObjectARX headers rather than guessed, so files
saved with DWG-native constraints stay honest about what AutoCAD can
actually represent.
Also fixes a case-sensitivity bug where a lowercase named-parameter
reference typed at a dimensional constraint's value prompt was rejected
(the command line uppercases SingleToken input before the parameter-name
lookup), and adds a "Used by" column to the Named Parameters panel.
Co-Authored-By: Claude Sonnet 5 <[email protected]>
Root cause (found by instrumenting build_fallback and reading
ttf-parser's own source, not guessed): cosmic-text's fallback search
picks macOS's system font (San Francisco / SFNS.ttf) for any codepoint
no stroke font or explicitly-requested family covers. It's a variable
TrueType font, and ttf-parser's outline_glyph() resolves a variable
glyph's contours through its gvar table's deltas relative to whatever
variation coordinates are currently set on the Face — an unset Face
has none, so outline_glyph silently returns None (not an error) for
such a glyph. The caller then still returned Some(Glyph { strokes:
vec![], .. }): a "resolved" glyph with nothing to render, no error
anywhere.
Two-part fix:
1. pin_default_variations() sets every variation axis to its own
default value before outlining, applied at all TTF outline-
extraction sites (both build_fallback variants, both build_shaped
variants, and the plain by-family glyph() lookup). This alone
wasn't sufficient for San Francisco specifically — even pinned, its
outline still didn't resolve for the glyphs this bug was found on,
apparently because ttf-parser can't retrieve base glyf data gvar
needs for at least some of that font's accented Latin glyphs.
2. So build_fallback() now treats "resolved a glyph ID but couldn't
extract its outline" as a real failure (outline_from_face returns
None, not an empty glyph) and falls through to a direct fontdb scan
for any other installed font that also covers the character and
can actually be outlined, rather than silently giving up.
Full workspace test suite: 813 passed, 0 failed (previously 812
passed, 1 failed — this was the one pre-existing failure carried
through the entire session).
Co-Authored-By: Claude Sonnet 5 <[email protected]>
Brings in 54 commits from origin/main: a selection-performance overhaul,
surface thickening, GPU-native wide/tapered stroke rendering, continued
ocs_doc_api evolution, and assorted fixes/polish (block double-click
editing, DWG hatch/database repair, clipboard, collapsible Properties
sections).
Six real conflicts, all where this branch had independently re-applied
a fix upstream later revised or already carried natively:
- src/app/automation.rs, src/app/command_driver.rs: combined both
sides — kept this branch's sketch-constraint/named-parameter loading
on "new"/"open", adopted main's Scene::clear()/material_base_dir and
its newer rebuild_derived_caches() over the older bump_geometry().
Scene::clear() (src/scene/entity.rs) now also resets
sketch_constraints/named_parameters, so every caller gets this for
free instead of each call site repeating it.
- src/io/mod.rs, src/scene/render_graph.rs: adopted main's design
outright (commit 973af87d, authored by Hakan directly) — a large
MINSERT is no longer purged as "corrupt" at load, only its rendering
is bounded, preserving the source data. This branch's own earlier
design (reject at load) predates that revision.
- src/scene/pipeline/mesh_gpu.rs: adopted main's valid_rgba_size()
guard (a real robustness improvement over this branch's version).
- src/snap.rs: replaced wholesale with main's version — this branch's
copy was an explicit verbatim snapshot from an earlier upstream PR
(commit 4c264b8c) with zero constraint-system-specific content, and
main has since evolved it considerably further.
Also removed two now-dead duplicates the merge surfaced without a
conflict marker (both branches had independently fixed the same
things): the OPTIONS/OP command dispatch/autocomplete entries, and
automation.rs's open_finalizes_and_purges_like_the_ui_open_path test
(kept main's version, which covers strictly more: solid_models/
material_base_dir reset on "new").
Full workspace suite passes (812 lib tests + this branch's own 24
constraint/persistence integration tests) except the one pre-existing,
unrelated font-test failure already present on main. Zero clippy
findings in the hand-written merge resolution itself; the ~1,171
warnings clippy reports are pre-existing in the 54 absorbed commits,
not introduced here.
Co-Authored-By: Claude Sonnet 5 <[email protected]>
New ConstraintKind variants — Concentric, CenterPoint, Colinear, Midpoint,
Fixed, PointOnCurve, EqualDistance, Symmetric — each built entirely from
existing ocs_gcs (FreeCAD planegcs) primitives, no new solver math needed.
build_constraint now returns Vec<Rc<dyn Constraint>> instead of
Option<Rc<...>>, generalizing the old Coincident-only "needs a second
system-level constraint" special case to every kind that needs one.
Wired into the ribbon (7 new buttons), a new point+pre-selected-entity
command for CenterPoint/Midpoint/PointOnCurve, a new 4-point command for
EqualDistance, and the existing DWG-native class-name mapping (all 8 reuse
the same generic "plain Geometrical node" path already built for
Coincident/Horizontal/etc.). Deliberately not modeled: Normal and
EqualCurvature, which for this app's Line/Circle-only entity model would
just be Perpendicular and Equal wearing different DWG class names.
Also makes DWG-native constraint persistence opt-in (Options > General,
off by default) rather than unconditional: a scope that doesn't yet have
the native graph is only given one when the setting is on, but a scope
that already has one — from an earlier save, or a file that came from
real AutoCAD — keeps getting it resynced regardless of the setting, so
toggling it off never freezes a stale graph in place.
Co-Authored-By: Claude Sonnet 5 <[email protected]>
Implements docs/dwg_constraint_compatibility_design.md's staged plan in
full: materializes the sketch-constraint set as acadrust's real
AssocNetwork/Assoc2dConstraintGroup object graph on save, additive to
the existing OCS_SKETCH_CONSTRAINTS XRecord, so files also carry
constraints AutoCAD/BricsCAD recognize as their own native objects
(per Hakan's request on the upstream project). Covers every
ConstraintKind, including AssocVariable wiring for named-parameter
driving values, and clears+rebuilds on each save.
DWG round-trips the full dependency chain; DXF only keeps the
constraint-group shell since acadrust's own DXF writer deliberately
omits AssocDependency-family objects. Both behaviors are covered by
tests and documented in the design doc's progress section, along with
two other implementation-time findings (skipping the NOD-anchored
top-level network, and the DWG writer's root-node/connections wire
format).
Co-Authored-By: Claude Sonnet 5 <[email protected]>
Hakan (upstream maintainer) asked for the sketch-constraint system to
persist using the constraint object model already exposed in acadrust/
cadcodec (Assoc2dConstraintGroup, AssocConstraintNode), so a file saved
by OpenCADStudio carries constraints AutoCAD/BricsCAD recognize as
their own native objects, not just an opaque custom XRecord blob.
Covers:
- The full ConstraintKind -> AssocConstraintNodeData mapping table,
grounded in build_constraint's actual refs shapes (the real contract,
not the doc comments).
- The geometry-node mapping for SketchRef (line/circle/arc/point
references) onto Line/BoundedLine/Circle/Arc/ImplicitPoint nodes.
- The AssocNetwork -> AssocAction -> Assoc2dConstraintGroup graph shape
actually needed, not just a flat data swap.
- A proposed additive persistence strategy: write both our own
OCS_SKETCH_CONSTRAINTS XRecord and the native graph, since our model
carries things (DrivingValue::Named) the native graph can't yet
losslessly express.
- Known gaps: Equal has no length/radius sub-kind split natively,
several AutoCAD constraint types we don't model at all, Tangent
involving an Arc (a pre-existing gap in the constraint system,
unrelated to DWG compatibility).
Three of five open questions are resolved against Autodesk's own
primary sources rather than left as inference: the ObjectARX 2012 SDK's
own C++ headers (mirrored on GitHub, Autodesk's own copyright notice
grants free use/distribution -- it's the SDK Autodesk publishes for
third-party interop) and Autodesk's live public ObjectARX reference
site. Confirmed: the associative network's dictionary key
("ACAD_ASSOCNETWORK"), the AcDb::ImplicitPointType enum settling how
SketchRef's point markers map to constraint geometry nodes, and that
AssocVariable's evaluator field can be left empty (Autodesk's own doc:
"the default evaluator for current acad version will be used"). Two
questions remain and need a real write-then-open test rather than more
reading: whether AutoCAD tolerates a partially-populated graph, and
DWG version gating on fields not yet checked.
Suggested staged approach: prove out Horizontal/Vertical on a single
Line first (sidesteps the value-dependency wiring), verify it opens
cleanly in real AutoCAD/BricsCAD or a DWG inspector, before writing
more of the graph on an unverified foundation. Design only -- no
implementation yet.
Co-Authored-By: Claude Sonnet 5 <[email protected]>
Named parameters (docs/named_parameters_design.md, stages 3-4):
- ParameterTable (src/scene/named_parameters.rs): a document-wide table of
named formulas, each resolved through a shared expression evaluator with
cross-parameter references, cycle detection, and division-by-zero/parse
error reporting. SketchConstraint::driving_param becomes DrivingValue
(Literal | Named) instead of a bare f64, resolved at solve time by
sketch_solve::build_constraint via Scene::named_parameters.
- Persistence (src/scene/named_parameters_persist.rs): same lazy XRecord
model sketch_persist.rs already established, synced around open/save.
Bumps the sketch-constraint XRecord's own format version (1 -> 2) since
DrivingValue's shape doesn't line up with a version-1 blob's bytes.
- PARAMETERS command and its editor (src/ui/window/named_parameters.rs):
add/remove/edit rows, live validation, duplicate-name detection: Apply
defines the table and re-solves every constraint scope with a
named-parameter-driven constraint (command_driver.rs). Distance/Angle
constraint prompts (modules/draw/constrain/value.rs) now accept either a
literal or an existing parameter's name, mirroring how the constraint's
own value prompt already worked.
- Undo scope-down carried over from this project's own established pattern
(constraint-system stage 9): the geometry an Apply moves is one undo
step, but the parameter table edit itself has no undo entry yet — a
documented follow-up, not a silent gap.
UI scale setting (src/app/settings.rs, app/mod.rs, ui/window/options.rs):
a whole-interface scale percentage (50-200%), wired directly into iced's
own per-window scale_factor hook rather than per-panel font-size
plumbing, so it applies immediately to every panel, dialog, and the
ribbon at once. Persisted alongside the existing lineweight-display-scale
setting.
OPTIONS/OP command dispatch (app/commands/view.rs, app/commands/mod.rs):
OPTIONS was registered for autocomplete but had no dispatch arm — the
Start-tab welcome screen's own button was the only way to reach it.
Found live-testing the UI-scale setting, unrelated to that feature
itself. "OP" is AutoCAD's own short alias.
help/: a new self-contained static user-help manual (manual.md as the
authored source, tools/build_help.py splits it into the committed static
HTML pages plus search index) — not yet wired into the application.
64 new/updated tests across the named-parameters and constraint-prompt
changes.
Co-Authored-By: Claude Sonnet 5 <[email protected]>
