1216 Commits
Author SHA1 Message Date
PaddleStrokeandClaude Opus 5.5 a6c0e8a959 Parameters: the Filter box finds a name by the starts of its words, so toodia finds tooth_outward_flange_diameter (#11978) (#12695)
* claim: #11978

* wip: Parameters filter and completion match the starts of a name's words (#11978)

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* wip: describe the Parameters filter's word-start matching in the guide (#11978)

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* Parameters autocomplete: list word-start matches after substring matches (#11978)

A name matched only by the starts of its words now comes after every name that contains the typed word, so it never pushes an earlier suggestion out of the list.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

---------

Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
2026-11-07 17:08:48 +00:00
PaddleStrokeandClaude Opus 5.5 3900dc742c Insert Mesh: optional move to origin on import (#12543) (#12698)
* claim: #12543

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* wip: Insert Mesh origin option (#12543)

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* Insert Mesh: name the Move to origin checkbox's handle mesh_units:origin (#12543)

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* Insert Mesh origin: measure in the active component's frame, first placement when it has no path (#12543)

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

---------

Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
2026-11-07 17:08:23 +00:00
PaddleStrokeandClaude Opus 5.5 4f687454fc Canvas: lock a reference image so it cannot be moved or deleted by mistake while sketching (#12310) (#12713)
* claim: #12310

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* wip: canvas lock (#12310)

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* wip: canvas lock: badge binding, save check on the value (#12310)

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

---------

Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
2026-11-07 17:06:42 +00:00
PaddleStrokeandClaude Opus 5.5 4f77836834 Sketch: a test that Symmetry about the origin point accepts two corners of the same rectangle edge (#4850) (#12772)
* claim: #4850

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* Sketch: test Symmetry of two corners of one rectangle edge about the origin (#4850)

Already fixed on main by #4626; this guards it.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

---------

Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
2026-11-07 17:05:01 +00:00
Nathan Dane 8c32875160 Preferences: save and restore panel layouts by workspace (#11306) (#12587)
* claim: #11306

* Preferences: save and restore panel layouts by workspace (#11306)

* Preferences: keep layout keys out of the viewport and fix validation checks

* Preferences: save other settings when panel layout conversion is refused

* Preferences: keep layout key handling scoped and deduplicate web warnings

* fix: retain imported panel column widths

* fix: use shared panel preference controls and tokens

* test: cover custom-height rows in the UI gallery
2026-11-07 07:13:40 +00:00
a2218e8a8d Sketch: Divide Curve places N-1 points on a curve at equal spacing (#8969) (#9664)
* claim: #8969

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* Sketch: Divide Curve places points at equal lengths along curves (#8969)

A new PointAlong constraint holds a point at its share of a curve's
length (measured along the exact curve), so the points follow edits.
sketch.point.divide places them on lines, arcs, circles, ellipses,
splines and conics; trim and break loosen them to point-on-curve.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* Divide Curve: use the solidcore_record segment and vector types after the layer split

Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>

* Divide Curve: exempt sketch.point.divide from kernel cases and show PointAlong glyphs

PointAlong constraints now get a glyph on their point (the midpoint icon),
light their point and curve, and hold an end against Extend like
PointOnCurve does.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

---------

Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Co-authored-by: bherbruck <brennen.herbruck@gmail.com>
2026-11-06 16:44:37 +00:00
austin62jakeandClaude Opus 5.5 cccef6dccd Agents and scripts: Set, Inspect and Frame Render Camera and Apply Render Camera describe their exact parameters and reply (#12288) (#12289)
* claim: #12288

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* Render camera: input and output schemas for Set, Inspect and Frame Render Camera and Apply Render Camera (#12288)

The four commands that set, read, fit and apply the render camera get
typed contracts and replies, in a new `schema/render_camera.rs`, and
leave `schema/unconverted.txt`.

- `RenderCamera` is a saved render setup's camera: its keys, and the
  bounds `RenderSetup::validate` keeps (1e9 mm, 1e6 radians, a positive
  distance, 0 <= fov < pi). The reader decodes it as the handler does,
  through a setup holding only that camera, so each refusal keeps the
  handler's message. The bounds are in the input schema only: a reply
  gives back the camera in the view, which Frame Render Subject can move
  beyond them, so the output schema has the camera's shape alone.
- `camera` and `name` are optional in the types and required in the
  schemas (a `require` transform), so a call without them still gets
  the handler's own message; `#[schemars(required)]` doesn't do this
  next to `serde(default)`.
- The handlers build their replies from the reply types through a small
  `typed_reply`; the JSON is unchanged.

Tests check each camera and aspect case against both the schema and the
handler, keep today's messages in strict, and match every reply against
its schema in warn.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

---------

Co-authored-by: Claude <noreply@anthropic.com>
2026-11-06 13:16:05 +00:00
3bcd93e2ca Sketch: a preference to turn off scaling the sketch to its first dimension (#4109) (#6807)
* claim: #4109

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* Sketch: a preference to turn off scaling the sketch to its first dimension (#4109)

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* Test the first-dimension preference at small, typical and large values

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* Engine: the scale_sketch doc names the Sketch preference

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* Translate the first-dimension preference into the complete catalogs (ja, zh)

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* Add the Arabic, Hebrew and Vietnamese texts for the first-dimension preference

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

---------

Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Co-authored-by: bherbruck <brennen.herbruck@gmail.com>
2026-11-06 09:31:40 +00:00
selune 3a70b43708 render: add CPU and GPU rough Glass, films, dispersion and caustics (#12200)
Add saved optical models, reciprocal rough interface transport, hard-light
connections, and bounded photon caustics for surfaces and selected media.
Keep inactive Glass sampling behavior and document admission and model limits.

Expose native controls and provide original CPU/GPU diagnostic captures.
2026-11-06 09:05:30 +00:00
Ahmad KhanandAhmad Khan 89191a1278 Parameters: the parameter list names the sketch that reads a parameter in a point coordinate (#12293) (#12350)
* claim: #12293

* Test: a parameter read only by a sketch point lists the sketch in used_by (#12293)

* Parameters: used_by names the feature that reads a parameter itself, as the graph does (#12293)

parameters.list built used_by from named parameter expressions only, so a
parameter read by a sketch point coordinate, a sketch pattern or a sheet
metal or plastic rule listed no user. The names a feature reads through its
own expressions now come from one helper that both parameters.graph and the
parameter table use.

---------

Co-authored-by: Ahmad Khan <ahmad.khan@shiritai.co>
2026-11-06 08:03:55 +00:00
Ahmad KhanandAhmad Khan 55d07a258c Insert Fastener: one fastener per hole, head at the open end, when a hole is picked at both ends (#12191) (#12404)
* claim: #12191

* Test: a hole picked at both ends gets one fastener at its open end (#12191)

* Insert Fastener: picks on one hole seat one part, at the hole's open end (#12191)

Picks were merged only when they seated at the same point, so a hole picked
at both ends (a blind hole's rim and floor edge, a through hole's two rims)
got a second fastener, turned over. Picks on one axis line of one body now
seat one part. When one of them is a blind hole's floor edge, the part is
seated at the rim instead: hole_wall now says whether the picked end of the
wall opens onto air, by classifying a point on the axis just past it.

* Insert Fastener: coaxial blind holes from opposite faces stay two holes, a fastener each (#12191)

---------

Co-authored-by: Ahmad Khan <ahmad.khan@shiritai.co>
2026-11-06 07:11:07 +00:00
Mario GeigerandClaude Opus 5.5 1589e18f2a Point clouds: remember the source file and reload it (#8818) (#8821)
* Point clouds: remember the source file and reload it (#8818)

A cloud keeps the absolute path it was inserted from. pointcloud.reload
reads it again, keeping the cloud's name, placement, units and colouring,
and the filters on properties the file still has; the Edit Point Cloud card
has a Reload from file button.

The file is read beside the window through the file loader, with progress
and Cancel, as an insert is. A thinned cloud stays thinned to as many points,
which reads an unchanged file as the same points; any other is read as a new
insert would be, so a file that grew comes in whole.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* Point clouds: a reload asks before it reaches a network location (#8818)

A design names the cloud's file, and a design may come from anyone. On
Windows, only asking about `\\host\share\x` connects to that computer and
may send it the user's sign-in. So `pointcloud.reload` now reads through
`vfs::read_design_path`, as linked designs do, and Reload from file asks
with the same prompt as Update Links.

The button's label uses its stable message id. The strings gain Korean,
which a new label needs. `pointcloud.reload` gets its line in
tests/kernel-cases/EXEMPT, which the "Every command has a case" job needs.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* Point clouds: a reload reads under the limit the cloud was read under (#8818)

The limit was inferred from the counts, as if a cloud with fewer points than its
file had been thinned. A point that cannot be read makes the same difference:
such a PLY read as half its points on an unchanged file, since a limit one below
the file's count is a stride of two, and a file that grew was cut to the old count.
Keep the limit itself: a read is a function of the file and the limit.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* Point clouds: a reload drops a value filter when the file stops naming the values (#8818)

A set of values filters a label, and `pointcloud.edit` refuses one on a property with no names.
A reload kept such a filter while the property stayed, so a file that dropped its names comment
left a cloud every later edit refused. One predicate now says what is refused and what a reload
drops.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* Point clouds: a reload keeps its undo step and ends a play it moved (#8818)

A reload accepted with the Edit Point Cloud card open left the card's undo
boundary before it, so the card's next edit folded the reload into itself and
one Undo took back both. The boundary now moves past the reload.

A play told its filter by place alone. When a filter before it went, by a
reload, an agent's edit or an undo, the next filter took that place and the
play's range, on screen and written when the play ended. A play now names its
filter's property and ends, unwritten, when that place holds another.

`engine.batch` and the control hook compared command ids exactly, while the
engine takes any case: `POINTCLOUD.RELOAD` passed the refusal and read the file
on the window thread. Both now resolve the id as the engine does, for every
command that loads a file. The batch row of the protocol said the refusal held
only with the agent cursor shown; it always holds.

No test was run locally, at the owner's request: compile, clippy and fmt only.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* Point clouds: a file load keeps the request's own param_check (#8818)

`file_load::control` takes a command the file loader runs before
`control::handle` does, so that command skipped the request's `param_check`:
`pointcloud.reload` accepted a word that is no level, a `strict` call with an
unknown parameter read the file anyway, and the reply did not say the level.
`doc.open`, the imports and the DXF export went the same way.

The hook now takes the level itself and checks the call before the file is
touched, as a command that switches designs does: the worker's session goes on
to be the design's, which keeps its own level. Both share the same check, and
the reply says the level and any warnings.

No test was run locally, at the owner's request: compile, clippy and fmt only.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

---------

Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
2026-11-06 06:56:52 +00:00
JK3DandClaude Sonnet 5.5 ed30a59795 History: joints as one grouped button, and the History bar folded into groups, both off by default (#8884, #8885) (#8886)
Rebased onto main as one change: the joints group into one button in the timeline, and the History bar folds into component runs, feature folders and STEP inserts. Both views are Preferences > Layout settings and off by default.

Co-authored-by: Claude Sonnet 5.5 <noreply@anthropic.com>
2026-11-06 06:10:37 +00:00
c66752e8d3 Agents and scripts: the render binding commands for library surfaces and optics describe their exact parameters and reply (#11741) (#11742)
* claim: #11741

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* Render bindings: input and output schemas for the surface and optical bind and unbind commands (#11741)

describe_command now gives the exact parameters and reply of the four
commands that bind a library surface or an optical profile to bodies in
a saved render setup, and unbind them, and the warn and strict
parameter checks use them. The handlers already read their parameters
with strict serde types; those types move to the new
schema/render_bindings.rs, gain descriptions and JsonSchema, and become
the contracts. The doc crate's target and setup types have no
JsonSchema, so their schemas are written by hand. The replies are built
from typed values instead of json! literals, with the same JSON, and
the saved setup is serialized before the document changes.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* Render bindings schema: `set` names the setup's fields, and `origin`'s default is the body's centre

`patched_settings` reads a `set` patch onto the saved setup as a
RenderSetup, which refuses unknown fields, so the schema now names the
setup's 14 fields with additionalProperties false (still at most 13 at
once). Without `surface.origin`, render.library.surface.prepare starts
the maps at the centre of the body's bounding box, not at the origin;
the description now says so. A new test covers the field names.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* Render bindings: keep main's CPU and GPU wording and the OpticalMaterial import after the rebase

Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>

---------

Co-authored-by: Claude <noreply@anthropic.com>
Co-authored-by: bherbruck <brennen.herbruck@gmail.com>
2026-11-06 05:47:42 +00:00
PeppaWutzandbherbruck 429520088a Agents and scripts: the view navigation and layout commands describe their exact parameters and reply (#12278) (#12353)
* Agents and scripts: the view navigation and layout commands describe their exact parameters and reply (#12278)

describe_command now gives the exact parameters and reply of view.navigate,
view.layout, view.history_layout, view.graphics_preset and
view.ambient_occlusion, and the warn and strict parameter checks use them.

The handlers still read their parameters from the raw JSON, so the new types
read each key as its handler does and give its message. Keys the handlers
refuse when left out (mode, preset, enabled, camera) are required in the
schema but left to the handlers' messages. The replies are built from typed
values instead of json! literals, with the same JSON as before.

* view.navigate: the contract reads only the keys of the form the handler takes (#12278)

The handler reads a motion's keys when motion is given (each motion its own), else direction, else degrees and action, and never the keys of another form. The contract is now read by hand in the same order, so a key of another form is not checked in warn or strict either, and a call that works with the check off works in both.

---------

Co-authored-by: bherbruck <brennen.herbruck@gmail.com>
2026-11-06 05:39:22 +00:00
Daniel DimitrovandClaude Opus 5.5 25d7a6b0c3 Hole: the same parameter definition typed for several holes uses that parameter (#12472) (#12473)
Single holes on several faces send one solid.hole command per face, each
carrying the same `name = expression` text. The second command was refused
because the name was already taken. A definition typed again exactly as an
existing user parameter was defined (same expression, same unit) now names
that parameter; a different value is still refused.

Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
2026-11-06 04:41:09 +00:00
selune 068c46cf8b Render: restore saved shots, progressive CPU/GPU images and exports (#6201) (#11087)
* Render: restore saved shots, progressive images and exports

* Render: register existing live and coat command tests

* Render: isolate lens saves and retain bounded previews

Save only lens parameters on existing CPU/GPU jobs while keeping unrelated
draft settings dirty. Retain pinned checkpoints through preview and status,
and release host-only snapshots outside the status lock.

Use the admitted PNG bound for registered-executor checkpoint slots. Preserve
executor work credits, final publication and regrade ownership. Add regression
coverage and align help, translations and capability documentation.

* Render: preserve live feedback and address review regressions

* Preserve saved cameras when applying Render library edits

Keep the saved camera when lighting or maps are applied to an existing shot. Fit and centre the measured Render host using the painted picture. Carry forward the upstream CPU and GPU preset and ZoomTool contracts.
2026-11-06 02:47:52 +00:00
708d5a149d Plane at Angle: use axis direction only with a pivot (#12041) (#12131)
* claim: #12041

Co-authored-by: ReDead <ReDead@users.noreply.github.com>

* Plane at Angle: use axis direction only with a pivot (#12041)

Add a direction_only option that projects the axis onto the base and
turns about a pivot (or the base origin), with dialog checkbox and
conditional Pivot pick. Older designs keep turning about the full line.

Co-authored-by: ReDead <ReDead@users.noreply.github.com>

* i18n: add ar, he, vi for Use axis direction only (#12041)

Complete catalogs need every English message; the pending locale file
was missing those three so cargo xtask i18n --check failed.

Co-authored-by: ReDead <ReDead@users.noreply.github.com>

---------

Co-authored-by: Cursor Agent <cursoragent@cursor.com>
Co-authored-by: ReDead <ReDead@users.noreply.github.com>
2026-11-06 01:38:49 +00:00
0c094142d3 Sketch Break and Extend: a cut or stop on a shape's centre ends there, with no second point (#12299) (#12301)
* Sketch Break and Extend: tests for a cut or stop on a center rectangle's centre

Breaking a center rectangle's diagonal at the centre leaves two points there, and extending a line up to the centre ends it on one diagonal only. These tests fail on main.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* Sketch Break and Extend: a cut or stop on a held midpoint ends there, as Trim does

Break dropped the broken line's Midpoint and joined the pieces at a new point held only on the cutter, so a center rectangle's centre came loose and a corner drag pulled the two points apart. It now uses Trim's rule (one helper, hold_cut): the pieces meet at the held point, which moves its hold to the cutter's middle when the sketch already implies that, and otherwise stays on the cutter.

Extend stopped on a point it lays on the curve it meets unless that curve ends there. A point held at another line's middle (a center rectangle's centre) is now a stop too, so the extension joins it.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

---------

Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Co-authored-by: bherbruck <brennen.herbruck@gmail.com>
2026-11-05 23:27:01 +00:00
1939adba6b Sketch Dimension: pick an origin axis as a reference (#2364) (#2373)
* claim: #2364

* Sketch Dimension: an origin axis can be picked as a reference (#2364)

The dimension tool ignored clicks on an origin axis and the engine's
sketch.dimension did not resolve x/y. Axis picks now map to the sketch's
linked axis line, as Symmetry and the other line constraints already do,
in all three entry points (click-then-pick, select-then-D, and the command).

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* Sketch Dimension: tests for a circle centre and a line against the sketch axes (#3137, #3320)

Folds in the cases from #3298: a circle's centre point dimensioned to the
x / y axis (#3137), and a line's angle and an end's distance to the x axis
(#3320). The axis resolution from #2364 already covers them; a plain circle
or arc with a line still measures to its edge (#1803).

Co-authored-by: Phanoob <phanoob@gmail.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* Sketch Dimension: the axis test picks the line off its midpoint, which is a point pick (#4872)

Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>

---------

Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Co-authored-by: Phanoob <phanoob@gmail.com>
Co-authored-by: bherbruck <brennen.herbruck@gmail.com>
2026-11-05 23:13:42 +00:00
Robson PereiraandClaude Opus 5.5 207679e261 Face and edge dialogs name the body of their picks (#7174)
Fillet, Chamfer, Offset Face (and Press Pull on a curved face, which opens
it), Draft, Replace Face and Delete Faces sent only the picked points. The
feature then checked every body's edges for them and evaluated on the body
whose mesh box is nearest the first point (`body_at`):

- In an assembly, a bore or pin inside a housing that comes first in the
  body list was looked for on the housing, which has no face there, and the
  command failed with "no face at [...]".
- In a large design, `body_at` meshed every body (and Fillet/Chamfer worked
  out every body's edges) on each preview and apply: seconds per change.

These dialogs now name the body when every pick lies on one, as Shell does
since #4555; picks on several bodies name none, as before. Press Pull passes
a given body on to Offset Face.

Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
2026-11-05 22:38:11 +00:00
PaddleStroke 22f8da2689 View: hold a shortcut to drag and zoom (#11701) (#11706)
* claim: #11701

* Allow assigned shortcuts to drag the view to zoom
2026-11-05 22:08:53 +00:00
51d8842e8a Create Sketch: the flat faces of a cone take an attached sketch (#7476) (#7772)
* claim: #7476

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* Create Sketch: the flat faces of a cone take an attached sketch (#7476)

The cone primitive was evaluated as an unnamed tool, so its faces had
only geometric names, which a face attachment refuses as not exact.
Its faces are now named from the revolve that builds it, as a
cylinder's are: `bottom`, `side` and, on a cut-off cone, `top`. A
sketch on its base or top attaches by that history name and follows
the cone's height.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* Fix compile after the layer split

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

---------

Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Co-authored-by: bherbruck <brennen.herbruck@gmail.com>
2026-11-05 21:54:30 +00:00
PeppaWutzandbherbruck 1444801d96 Agents and scripts: the canvas commands describe their exact parameters and reply (#12275) (#12338)
canvas.insert, canvas.decal, canvas.edit, canvas.calibrate and canvas.delete
now publish typed input and output schemas. Each parameter is read the way
the handler reads it today and refused with the handler's own message, and
the replies are built from the same types the output schemas describe, so the
JSON they return does not change. Their lines leave unconverted.txt.

The plane check of sketch.create's plane reader is split out so the canvas
contract can use it without changing what it accepts.

Part of #2829.

Co-authored-by: bherbruck <brennen.herbruck@gmail.com>
2026-11-05 21:38:07 +00:00
4eb0c6f5ec Agents and scripts: Manage, Assign and List Plastic Rules describe their exact parameters and reply (#12313) (#12314)
* claim: #12313

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* Plastic rules: input and output schemas for Manage, Assign and List Plastic Rules (#12313)

The three plastic rule commands get typed contracts and replies, in a
new `schema/plastic.rs` laid out like the sheet metal rules', and leave
`schema/unconverted.txt`.

- Each key reads as its handler reads it, with the handler's message:
  the rule name as `plastic.manage_rules` trims and limits it, `bodies`
  through `config::NameList` (which mirrors `params::string_list`), the
  ten rule values as `LengthExpr` and `AngleExpr`.
- `name` and `bodies` are optional in the types and required in the
  schemas (a `require` transform), so a call without them still gets
  the handler's own message.
- Keys of the wrong type that the handlers ignore today (a non-boolean
  `active` or `delete`, a rule value that is neither a number nor text,
  a non-text `from`, `material` or `rule`) are refused; in warn the call
  still works, with a warning. A test pins each one.
- The handlers build their replies from the reply types through a small
  `typed_reply`; the JSON is unchanged.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* Assign Plastic Rule: the schema's `bodies` is a name or a non-empty list, not null

The field is `Option<NameList>`, so its published schema allowed `null`
(from the `Option`) and an empty list (from `NameList`), both of which
the handler refuses ("`bodies` must list bodies"). It now has its own
schema: a name, or a list of 1 to 10 000 names. The reader is unchanged,
so strict still gives the handler's message for those calls. The schema
test adds both cases.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

---------

Co-authored-by: Claude <noreply@anthropic.com>
Co-authored-by: bherbruck <brennen.herbruck@gmail.com>
2026-11-05 20:00:52 +00:00
PeppaWutzandbherbruck a07b60cd81 Agents and scripts: the linked-component commands describe their exact parameters and reply (#12277) (#12351)
component.insert_design, update_links, relink, break_link, links, extract,
copies and replace_with_link now publish typed input and output schemas.
Each parameter is read as its handler reads it today, with the same error
messages, and each reply is built from the typed value and serializes to the
same JSON as before. Their lines leave unconverted.txt.

Co-authored-by: bherbruck <brennen.herbruck@gmail.com>
2026-11-05 20:00:28 +00:00
8c4caf9372 Insert Part help: say which way point and direction place the part (#11729) (#11755)
Co-authored-by: bherbruck <brennen.herbruck@gmail.com>
Co-authored-by: Claude Sonnet 5.5 <noreply@anthropic.com>
2026-11-05 19:51:56 +00:00
202b775de4 Agents and scripts: Add, Rename, Delete, Comment, Favorite and Move Parameter describe their exact parameters and reply (#11941) (#11943)
* claim: #11941

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* User parameters: input and output schemas for Add, Rename, Delete, Comment, Favorite and Move (#11941)

describe_command now gives the exact parameters and reply of the six
commands that manage a design's user parameters by name, and the warn
and strict parameter checks use them. The handlers read their
parameters from the raw JSON, so the contracts in the new
schema/parameters.rs read each key as its handler does, with its
message; `value` stays as parameters.add's other name for `expression`,
and `comment` is required in the schema but left to the handler's
message in the type. The replies are built from typed values instead of
json! literals, with the same JSON; parameters.add's reply flattens an
untagged AddedValue (a number's value or a text parameter's text).

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* Parameters: the schemas count a `null` expression and Move's neighbour as left out

`parameters.add` reads a `null` `expression` or `value` as left out
(`expr_arg`), so the schema's "expression or value" alternatives now
also need the key's type (number or text): `{"expression": null}` alone
no longer matches, while `{"expression": null, "value": 3}` does.

`parameters.move` needs exactly one of `before` and `after`. The schema
now says so with a `oneOf` whose alternatives each need their key as
text, so a `null` neighbour (read as left out) counts as neither, and
both or neither is refused. `conforms` doesn't read `oneOf`, so the test
pins it and counts the matching alternatives itself.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

---------

Co-authored-by: Claude <noreply@anthropic.com>
Co-authored-by: bherbruck <brennen.herbruck@gmail.com>
2026-11-05 19:51:46 +00:00
146a43f38e Agents and scripts: Motion Study, List and Delete describe their exact parameters and reply (#12155) (#12156)
* claim: #12155

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* Motion studies: input and output schemas for Motion Study, List and Delete (#12155)

describe_command now gives the exact parameters and reply of
motion.study, motion.list and motion.delete, and the warn and strict
parameter checks use them. The handler reads its parameters from the
raw JSON, so the contracts in schema/motion.rs read each key as it
does, with its message (a study by id or name, steps, and keys of
[step, value] points); whether the study, joint and value exist stays
the handler's to check. The replies are built from typed values
instead of json! literals, with the same JSON, through `typed_reply`
(the contact commands' helper, renamed now that it serves both);
motion.list's traces stay as motion_trace.rs builds them.

The parameter audit found two tests in cmd/motion_tests.rs sending
`delete: true` to motion.delete, a key only Edit Contact Set takes; a
`deletion` helper now sends it only there.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* Motion Study: a point's step must be a number in the schema, as the handler reads it

The schema typed both elements of a key's point as "a number or text",
so `["0", 20]` matched it although the handler refuses a step that isn't
a number ("points are [step, value]"). The two are now typed apart with
`prefixItems`: the step a number, the value a number or an expression,
each under 1e9 in size, as `study_key_problem` and the handler read them.
Elements after those two, which the handler ignores, are left free.

The 1e9 limit, written twice in the key check, is now the named
`MAX_POINT_MAGNITUDE`, used by the check and the schema alike. A new test
checks nine point shapes against both the schema and the key check, so
the two can't drift apart.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

---------

Co-authored-by: Claude <noreply@anthropic.com>
Co-authored-by: bherbruck <brennen.herbruck@gmail.com>
2026-11-05 19:51:04 +00:00
e04bc060f6 Web app: clippy for wasm passes with -D warnings (#12222) (#12251)
* claim: #12222

* Web app: clippy for wasm passes with -D warnings (#12222)

cargo clippy -p solidcraft-web --target wasm32-unknown-unknown stopped at
drop(file) in safe_write.rs and showed 12 more warnings only that target
has. Each is fixed where it is:

- safe_write::create_atomic and control_auth::write_token_file_with close
  their File by scope (a block; Result::map's closure) instead of drop(),
  which clippy's drop_non_drop flags where File is a stub.
- Native-only imports, the AHEAD constant and the point-cloud worker's
  Condvar are behind cfg(not(target_arch = "wasm32")); Exported::bodies
  and Start::Later, read or built only natively, allow dead_code there.
- home::ask_present uses ctx in its browser branch; the browser blur stub's
  begin returns a Scope, as the native one does.
- The web app drops three map_err(JsValue::from) on dyn_into's error,
  which is a JsValue already.

Native builds behave the same.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

---------

Co-authored-by: thocsmith <193096300+thocsmith@users.noreply.github.com>
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Co-authored-by: bherbruck <brennen.herbruck@gmail.com>
2026-11-05 19:50:25 +00:00
PeppaWutzandbherbruck 464da90238 Agents and scripts: the Browser folder commands and Rename Body describe their exact parameters and reply (#12276) (#12339)
browser.collapse_all, browser.expand_all, browser.group, browser.move,
browser.order, browser.rename_group, browser.ungroup, browser.visibility
and body.rename now publish input and output schemas, so agents and
scripts can read exactly which parameters they take and what they
return, and the warn and strict parameter checks cover them.

Each parameter is read the way the handler reads it today, with the same
error messages. The replies are built from typed values and serialize to
the same JSON as before. Part of #2829.

Co-authored-by: bherbruck <brennen.herbruck@gmail.com>
2026-11-05 19:49:12 +00:00
SS-ENandClaude Opus 5.5 0b32d3619d Extract a component whose body was built in another component (#12153)
* Extract a component whose body was built in another component

component.extract now takes along the outside features that built, changed or moved
in a body the component owns, when they touch nothing that stays outside. Otherwise
it names the feature that blocks it. Fixes #12142.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* Kernel case: extracting a component with a body moved in now succeeds

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

---------

Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
2026-11-05 19:48:19 +00:00
austin62jakeandClaude Opus 5.5 eb10552c61 Agents and scripts: the native studio and quality profile commands describe their exact parameters and reply (#11766) (#11768)
* claim: #11766

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* Render presets: input and output schemas for the native studio and quality profile commands (#11766)

describe_command now gives the exact parameters and reply of
render.preset.resolve and render.preset.apply (review and apply a
native studio composition) and render.quality.prepare (review a fixed
sampling budget), and the warn and strict parameter checks use them.
The handlers already read their parameters with strict serde types;
those types move to the new schema/render_presets.rs, gain descriptions
and JsonSchema, and become the contracts. The replies are built from
typed StudioComposition and QualityPlan values instead of json!
literals, with the same JSON.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* Render presets schema: `set` names the setup's fields

As on #11762 and #11742: `patched_settings` reads a `set` patch onto
the saved setup as a RenderSetup, which refuses unknown fields, so the
schema now names the setup's 14 fields with additionalProperties false
(still at most 13 at once). A new test covers it.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

---------

Co-authored-by: Claude <noreply@anthropic.com>
2026-11-05 19:45:18 +00:00
austin62jakeandClaude Opus 5.5 1fdc56835b Agents and scripts: New, Edit, Inspect and Delete Physical Material describe their exact parameters and reply (#11862) (#11865)
* claim: #11862

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* Physical materials: input and output schemas for New, Edit, Inspect and Delete Physical Material (#11862)

describe_command now gives the exact parameters and reply of
material.create, material.edit, material.inspect and material.delete,
and the warn and strict parameter checks use them. The handlers read
their parameters from the raw JSON, so the contracts in the new
schema/materials.rs read each key as its handler does, with its
message: `null` keeps its meaning (`color: null` and `physical_card:
null` clear), the look keys reuse the appearance commands' types, and
the card keeps the handler's own check, with its schema from
schema/material_properties.rs (whose card_schema becomes pub(super)).
The replies are built from typed values instead of json! literals, with
the same JSON.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* Colour schema: a colour's text form allows what color_param reads

inspect::color_param trims the text and drops any number of leading
`#`, so "  #ff00ff  " and "##ff00ff" are colours to appearance.assign
and the material commands, but ColorParam's published pattern refused
them. The pattern is now built from schema::types' whitespace class (as
AxisDir's is), which becomes pub(crate), and a test pins it next to the
reader's answers.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* Colour pattern: a byte may be `+` and one hex digit, as the reader takes it

`inspect::color_param` parses each of the three bytes with
`u8::from_str_radix`, which accepts a leading `+`, so "+fff00" is read as
[15, 255, 0] today. The schema's pattern asked for six hex digits and
would have made strict refuse a colour the handler accepts. The pattern
now allows either two hex digits or `+` and one digit per byte, and the
test pins both the reader's answer and the pattern's text, since
`conforms` doesn't read patterns.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

---------

Co-authored-by: Claude <noreply@anthropic.com>
2026-11-05 19:42:20 +00:00
austin62jakeandClaude Opus 5.5 3fde538320 Agents and scripts: Inspect HDR Source and Save Render Settings with HDR describe their exact parameters and reply (#11761) (#11762)
* claim: #11761

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* Render HDR: input and output schemas for Inspect HDR Source and Save Render Settings with HDR (#11761)

describe_command now gives render.hdr.inspect's and
render.settings.capture_hdr's exact parameters and reply, and the warn
and strict parameter checks use them. The handlers already read their
parameters with strict serde types; those types move to the new
schema/render_hdr.rs, gain descriptions and JsonSchema, and become the
contracts, so the check is no stricter than the handlers. The replies
are built from typed HdrInspected and HdrCaptured values instead of
json! literals, with the same JSON. `expected_sha256` is the doc crate's
HdrId, whose schema is written by hand (64 lowercase hex digits or null).

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* Render HDR schema: `set` names the setup's fields, and its job may be the saved one's

`patched_settings` puts a `set` patch's keys onto the saved setup and
reads the result as a RenderSetup, which refuses unknown fields. The
schema now names the setup's 14 fields with additionalProperties false
(still at most 13 at once), so `set: {"widht": 10}` is refused, as the
handler refuses it; the values stay the handler's to check. The
description now says the setup the changes make (a saved one keeps the
fields left out) needs a cpu_raytrace or gpu_path job, not that every
patch must give one. A new test covers the field names.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

---------

Co-authored-by: Claude <noreply@anthropic.com>
2026-11-05 19:41:27 +00:00
SilentWolf75andCursor 283ccd0b21 Web: tab title follows a renamed design (#11375) (#11381)
* claim: #11375

Co-authored-by: Cursor <cursoragent@cursor.com>

* Web: tab title follows a renamed design (#11375).

When an OPFS design is renamed, open tabs keep the new path and a borrowed design name so the tab matches the file list.

Co-authored-by: Cursor <cursoragent@cursor.com>

---------

Co-authored-by: Cursor <cursoragent@cursor.com>
2026-11-05 19:41:04 +00:00
cronos 21a1269d5c Sketch Trim: keep the stored offset distance after deleting its dimension (#12186) 2026-11-05 19:39:21 +00:00
SS-ENandClaude Opus 5.5 1b52186524 Timeline: reordering a feature says when nothing moved and returns its new index (#12171)
* timeline.reorder: say when nothing moved, and return the new index (#12140)

A position past the end was capped to the last slot; when the feature was
already there the command reported success with no sign nothing moved. The
reply now gives position (index now), from and moved, with a note that says
why nothing moved or that the last slot was used. A move that puts a feature
ahead of what it uses says so, and the command refuses before changing the
design if the feature would not land at the asked position.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* timeline.reorder: word a refused move by its direction; note a marker-only change

A feature moved later can fail for reasons other than order, such as a cut that
removes its body, so the move-earlier hint is given only when it moved earlier.
With `active`, a feature left in place gets a note saying only the history
marker moved, or that nothing moved.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

---------

Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
2026-11-05 19:38:38 +00:00
Homero ThompsonandHermes Agent 51f4074506 Split and Trim errors name the command a script can run (#4712) (#11205)
* Sketch modify: name the public command in the split and trim errors (#4712)

Part of #4712.

The split inside the sketch modify tools refused in the name of the
internal `split` step, and the free-form trim in the name of the
internal `trim` one: `invalid parameters for `split`: a circle needs
two points to split at` names a command nobody can write, so an agent
or script reading the message has nowhere to go.

Both now name the public command the caller used: `sketch.modify` for
the split shared by trim and break, `sketch.trim` for the trim. The
descriptions are unchanged.

Co-authored-by: Hermes Agent <noreply@anthropic.com>

* Sketch modify: the split names the command the caller used (#4712)

`sketch.modify` is not a command a caller can write: the registered
sketch commands are `sketch.trim`, `sketch.trim.path` and `sketch.break`.
The split shared by trim and break therefore refused in the name of a
command that does not exist, and a script reading the message had
nowhere to go.

`split_at` takes the command of its caller and refuses in that name:
`sketch.trim` when the trim splits the curve, `sketch.break` when the
break does. Both callers already had the command in hand.

The test now states the decision: the refusal of `sketch.break` carries
`sketch.break`, and the rendered text names no other command.

Co-authored-by: Hermes Agent <noreply@anthropic.com>

---------

Co-authored-by: Hermes Agent <noreply@anthropic.com>
2026-11-05 19:28:22 +00:00
FrankandClaude Opus 5.5 9cdee588f2 Extrude Two Sides: fix the side and missing value box of Distance 2 (#11203)
* Extrude: Two Sides keeps each distance on the side Direction gives it

Distance 2 stays on the side Direction gives it, whatever the sign of Distance 1, and its value box shows on the canvas. Rebased onto main as one change.

Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>

* Two Sides extrude: keep older designs' sides by a marker, and the full profile search

- A saved two-sided extrude is marked `second_follows_first` instead of having a literal
  negative distance rewritten: expressions such as `-width` keep the old sides when their
  parameters change, `-0 mm` keeps its span, and a feature's named inputs keep their values.
- Live sketch drags and the restored model cache use the full profile search evaluation uses,
  so a sketch whose only closed boundary is projected keeps its profiles.
- The canvas Distance 2 box stays while its text does not evaluate.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* Tests: give the extrude added on main the second_follows_first field

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

---------

Co-authored-by: Claude Sonnet 5.5 <noreply@anthropic.com>
2026-11-05 18:39:09 +00:00
Nathan Dane 90d9c6488a Panels: add columns, stacks and tabs with exact drag previews (#11305) (#11608)
* claim: #11305

* Panels: add columns, stacks and tabs with exact drag previews (#11305)
2026-11-05 18:38:54 +00:00
b2f5da6ca7 joints: support custom names and joint type-dependent default names (#10340) (#10165)
* joints: support custom names in dialog and type-based default names

Allow users to specify a custom name in the Joint dialog and generate
type-prefixed default names (e.g. "Planar Joint1", "Revolute Joint1")
instead of generic "Joint{n}".

- Add `name` field to `JointForm` and display a text row in the Joint
  dialog with a dynamic placeholder matching the chosen type.
- Update `new_joint_name` in `cmd/joints.rs` to take `JointKind` and
  prefix auto-generated names with the joint kind label.
- Prioritize non-empty custom names and pass them through `joint.create`,
  `joint.as_built`, and `joint.edit`.
- Pre-populate the existing joint name in `edit_joint` to allow renaming
  directly from the Edit Joint dialog.

* joints: add tests for custom and type-based joint naming

Verify that the joint dialog and engine command layer correctly generate
type-prefixed default names and respect user-specified custom names.

- Add engine-layer test `new_joints_generate_type_prefixed_names_or_keep_custom_name`
  to verify `joint.create` and `joint.as_built` naming logic.
- Add UI-layer test `joint_dialog_applies_custom_name_or_falls_back_to_type_default`
  to ensure the `JointForm` captures and submits names correctly.

* Format dialogs_assembly_tests.rs (trailing blank line)

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

---------

Co-authored-by: bherbruck <brennen.herbruck@gmail.com>
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
2026-11-05 18:31:43 +00:00
Steven WatersandClaude f1159a5d51 Added mesh import units for STL and OBJ (#10406)
Co-authored-by: Claude <noreply@anthropic.com>
2026-11-05 18:31:26 +00:00
6500c42c9f Agents and scripts: sketch Copy, Cut and Paste describe their exact parameters and reply (#10963) (#10964)
* claim: #10963

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* Sketch Copy, Cut and Paste: input and output schemas (#10963)

sketch.copy, sketch.cut and sketch.paste get typed contracts and reply
types in schema/sketch_clipboard.rs and leave unconverted.txt. `at` and
`translate` read as the paste reads them, with its message; the replies
are built from Copied and Pasted, with the same JSON as before.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* Sketch clipboard: import Vec2 from solidcore_record after the layer split

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

---------

Co-authored-by: Claude <noreply@anthropic.com>
Co-authored-by: bherbruck <brennen.herbruck@gmail.com>
2026-11-05 16:14:13 +00:00
9f2e117934 Revolve follows the highlighted sketch edge (#7452) (#7479)
* Revolve follows the highlighted sketch edge (#7452)

Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>

* Attribute the #7452 screenshots and recording

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* Fix needless borrow in the revolve axis test

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

---------

Co-authored-by: bherbruck <brennen.herbruck@gmail.com>
Co-authored-by: Claude Sonnet 5.5 <noreply@anthropic.com>
2026-11-05 15:31:53 +00:00
abdf195132 Sketch: sketch.move_points moves several points, each to its own target (#5961) (#5978)
* claim: #5961

* Sketch: sketch.move_points moves several points, each to its own target (#5961)

A script editing a traced outline could move one sketch point per call,
and only by its id: an agent that had to move two points of a
490-segment polyline redrew the whole sketch instead, breaking every
dimension and feature built on it.

sketch.move_points {moves: [{point | near, to}], tolerance?} drags each
point to its own target in one edit (one undo step), as sketch.move_point
drags one, and the sketch is solved. `near` picks the sketch point
closest to a coordinate, refused when none lies within `tolerance`
(default 0.1 mm) or two are as close. Every point is found before any
moves. The reply gives each point's id, its target, where it ended up and
whether it got there, so a fixed or constrained point is reported rather
than silently left in place.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* Move the new command label to a pending locale file

Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>

* Fix the new command for the sketch store and geometry split

Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>

* Carry the new command label in every complete catalog

Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>

* sketch.move_points reads the sketch through sketch_store

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

---------

Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Co-authored-by: bherbruck <brennen.herbruck@gmail.com>
2026-11-05 12:20:00 +00:00
bherbruckandClaude Opus 5.5 7e002f182d Opening a design from a newer SolidCraft: it opens with a notice instead of being refused (#12166)
* Open a design from a newer SolidCraft with a warning instead of refusing it

A newer format number or an unknown change no longer blocks opening. What
this version can read opens; a feature it can't read stays on the timeline
with an error, and fields, sections and container parts it doesn't know are
named in the reply and written back by Save, under the file's own number.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* Saving a newer design from the CLI, and editing a feature this version can't read

`run --save` writes a design holding a feature only a newer build can build;
`--out` still stops, since it would export the part without that feature.
Edit Feature on such a feature says why instead of opening the palette.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* Translate the newer-design notice for the complete catalogs; clippy

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

---------

Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
2026-11-05 12:18:25 +00:00
dc33c4a24d Saving: designs with many threaded holes keep all saved states (#10601)
* claim: tap marking records once in the model cache

* Model cache: write each tap marking record once and share it between states

Problem: every state of a saved model wrote the cosmetic tap markings of all the features
before it, so the records grew with the square of the features. A design of 400 marked
cylinders wrote 34 of its 37 MB manifest that way: past the 32 MiB limit the save thinned the
cache (590 of 806 states left out, opening evaluates them again) and wrote the manifest
several times while choosing what to keep.

Solution: the records keep a list of markings, each written once (keyed by its text, as the
sheet bodies are); a state names its markings as runs of places, like its bodies and
sketches. A marking at the same place in the state written just before, equal to the bit, is
found without writing it as text. A thinned model keeps the whole list and counts it once;
the reader counts the markings with what a state holds. The cache format is now /6: an older
cache is a miss and the design is evaluated as before. The design document is unchanged.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* Saved model: count each tap marking by its text against the held limit

A state clones every marking it lists, so a run repeating one large marking could hold far more than MAX_HELD items of memory.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

---------

Co-authored-by: Markus Kaltenbrunner <kaltenbrunnermarkus@users.noreply.github.com>
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Co-authored-by: bherbruck <brennen.herbruck@gmail.com>
2026-11-05 12:11:07 +00:00
de07905f78 Sketch Offset: an inward offset past a narrow neck is refused, not looped (#11055)
* Sketch: an outline drawn inside a shape past its neck is refused, not looped

An inward offset of a closed outline wider than half a neck of the
shape would pinch it into two; one chain can't be that, and its pieces,
each joined only to its neighbours, ran back over each other in a loop.
When the outline itself doesn't cross itself and its offset would, the
offset is refused like other offsets the shape can't take. An outline
that already crosses itself (a bow tie) keeps its crossing offset.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* Loop guard: the arc's direction comes with its source, not with the offset piece

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* Loop guard: a touch counts, lines along each other count, tolerances from the outline's size

An outline split where it crosses itself only touches itself there and
was taken for simple, so its offset was refused: between pieces that
aren't neighbours a touch now counts as crossing. A neck exactly twice
the distance gives two lines lying along each other, which cross at no
point: such an overlap now counts too. The tolerances come from the
outline's own size plus the coordinates' rounding, not from how far it
lies from the origin.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* Sketch offset: pass the arc direction from the outline after the dog bone change

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

---------

Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Co-authored-by: bherbruck <brennen.herbruck@gmail.com>
2026-11-05 11:34:02 +00:00
1cf7ae8e5c Sketch: dragging one of several selected sides no longer runs ahead of the pointer (#11324)
* Sketch: a selection held back in part follows the pointer, not past it

Dragging a selection part of which cannot move (a rectangle with a corner
on the origin, all four sides selected) took the dragged side nearly twice
as far as the pointer. `sketch.drag` reported the grab where the average of
all dragged points went; the held corner kept that behind the side, and the
next frame's catch-up overshot. It now reports how the dragged point that
got furthest toward the pointer moved, so what can follow moves 1:1.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* Sketch drag test: unwrap the world state after the engine change

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

---------

Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Co-authored-by: bherbruck <brennen.herbruck@gmail.com>
2026-11-05 11:33:48 +00:00