* claim: #9258
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Sketch: paste sketch geometry copied in another design (#9258)
Sketch geometry copied or cut in a sketch (sketch.copy / sketch.cut) now pastes into a sketch of another open design with Ctrl+V or the context menu, at the cursor or at the copied coordinates, with the constraints and dimensions among it. A dimension whose expression names the copied design's parameters keeps its value, in its unit; a text keeps its numbers. SketchClip::for_another_design prepares the clipboard; the app keeps the last sketch copy beside the cross-design body/component copy.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
joint.create's flip text and the assemblies guide said two origins at one
point with the same z stay as they are; a point origin without x takes its
x from z, so a part turned about z turns back. Say so, and point to giving
x on both origins or joint.as_built.
edit.checkpoint and edit.restore_checkpoint now say that restoring or
forgetting uses the checkpoint (and later ones) up and that only the last
16 are kept, and the "no checkpoint N" error says why. The editing and
mechanisms skills no longer read as if a checkpoint can be restored twice.
Co-authored-by: thocsmith <193096300+thocsmith@users.noreply.github.com>
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
The browser asks about every joint of the design on every frame, and
the answer walked the list of hidden joints: a nested STEP assembly
hides thousands of them, so a frame cost grew with the size of the
design.
Session keeps the hidden joints and folders as sets and answers from
them. The sets are built when the view changes and kept until it
changes again: the revision says the design changed, and the lists
say the view changed under it (the view is replaced outside the
engine too, without a revision of its own).
The behaviour is unchanged: the hidden joint stays hidden, the shown
one stays shown, and undo and redo keep the hiding right.
* claim: #9641
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Insert SVG: say why a drawing has nothing to import, and read UTF-16 files (#9641)
An SVG with nothing SolidCraft draws (only an embedded picture, shapes
pulled from another file with <use>, or not an SVG at all) was refused with
"invalid parameters for sketch.insert_svg: the drawing has no geometry
SolidCraft reads". It now names the file and says why, pointing pictures at
Insert Canvas and Canvas to Sketch. SVG files with a byte-order mark or in
UTF-16 now read, and a gzip-compressed (.svgz) file is refused clearly.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Insert SVG: decode UTF-16 with as_chunks (clippy)
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
* claim: #9686
Co-Authored-By: Claude <noreply@anthropic.com>
* Edit Joint: joint.edit takes the two sides
joint.edit accepts a and b in the form joint.create takes them (read by the same origin_param), so a joint can move its attachment without being deleted. A side left out keeps its place. The two sides must stay on different occurrences, checked as joint.create does. Limits, values, expressions and the name are kept; a relation's values start again where its parts are.
joint.edit gets an input contract and reply type (schema/joints.rs), and leaves schema/unconverted.txt.
Co-Authored-By: Claude <noreply@anthropic.com>
* Edit Joint shows both sides and picks them again
The Edit Joint dialog now shows the two sides in the same selection inputs as Joint, each pre-filled with the face or edge of its own component under the origin. OK sends joint.edit a or b only for a side picked again, so an untouched side keeps its origin exactly.
Co-Authored-By: Claude <noreply@anthropic.com>
* Edit Joint: a moved side turns the part as joint.create would
joint.edit kept the joint's old angle when a side changed, so the part could turn by a half turn about its axis when the new origin's x lay the other way. When a side moves and no angle is given, the angle is worked out again as joint.create does (the least turn from where the part is now); a relation's angle is its tilt and stays.
The mechanisms reference says joint.edit takes a and b.
Co-Authored-By: Claude <noreply@anthropic.com>
* Edit Joint: pre-fill sides by name, send the angle only when changed
A side shows the face or edge its snap names, the piece nearest its stored pick when the name has several, else the one nearest that pick; it no longer searches outward from the frame origin, which could show the plate beside a hole for the hole's wall. A point or frame side shows nothing. OK sends the angle only when the user changed it, so a side moved without it turns the part as joint.create would.
Co-Authored-By: Claude <noreply@anthropic.com>
* Edit Joint: a moved side leaves an angle expression alone (#9686)
Co-Authored-By: Claude <noreply@anthropic.com>
---------
Co-authored-by: Claude <noreply@anthropic.com>
* claim: #9465
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Face attachments: a piece of a face its certificate finds the other pieces of keeps that certificate (#9465)
The nightly action matrix has refused a sketch and an offset plane on the
sensor enclosure lid's underside (`examples/sensor_enclosure :: Lid/F2:top`)
since #8278. The lid's lip, joined below it, parts the underside into a ring
and a panel; the ring rests on the case and is named after the case's top,
so the lid's cap certificate (via certified_pieces) finds only the panel.
#8278 then refused the certificate because it doesn't find the picked face,
and the history-name fallback can't take a name made by another body's
feature.
A face that is no pattern or mirror copy and lies on the same oriented plane
as the pieces the certificate finds is another piece of that face: the
certificate stands for it, as before #8278. Copies (the #8278 case) still
fall back to their own history name.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
* claim: #9062
* Sketch drag: a line is never shrunk to a point to follow the cursor (#9062)
A line's end on a horizontal construction line through the origin, dragged
up off it, made the construction line vanish. A horizontal row is
b.y - a.y, which a line a hair long meets at any tilt, so the drag's held
solves met the point-on-line by shrinking the construction line: to 1e-5 mm
on the first frame, 1e-7 mm by the tenth, and the end went up.
A drag step (held solve, the tangent-ends-held solve, and each cursor
correction pose) that shrinks a line or circle from over 1e-3 to under
1e-4 now counts as failed, as a step that wraps an arc does. The end then
stays on the construction line, which keeps its length.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
* claim: #9142
* Exploded View: grey it out until the design has two components (#9142)
On a part made of plain bodies (the sample plate) Exploded View opened its
dialog and failed on OK with "an exploded view needs two or more
occurrences". The command is now disabled until the design has two or more
occurrences, and its tooltip says why and which commands make components.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
* claim: #9731
* Sketch Extend: only an end on a circle counts as held (#9731)
grow_line's on-circle test was true for every point inside the circle; test distance to the circle instead. Also pin the new end point of a held-end extension onto the curve it stops on, as the free-end path does.
* Fix regression test harness usage and held-end setup (#9731)
---------
Co-authored-by: Dahak Builds <dahakbuilds@users.noreply.github.com>
construction_visible checked the item's own hide and its folder, but not
the placement its component is drawn at, so hiding Part:1 left Part's
planes, axes and points on screen. It now asks feature_placement_visible,
the check sketches already made; the placement path is computed in one
place for both.
* claim: #9741
* motion.list gives study keys in motion.study units
Angle keys came back in radians while motion.study takes degrees, so an
agent that listed a study and wrote it back changed every angle by a
factor of 57.3, silently. Convert key values on the way out like
joint.list does (degrees for angles, mm for distances), give each key a
unit and its joint's name, and add a round-trip regression test.
---------
Co-authored-by: Dahak Builds <dahakbuilds@users.noreply.github.com>
* claim: #8776
* Crash reports: a window opened after startup no longer labels the run "during startup" (#8776)
egui-wgpu reports "window surface" and "surface configure" for every native
window, also the ones opened after startup, such as the agent chat in its own
window. Each one overwrote the journal's "running" phase, so a later unclean
exit was reported as "during startup (surface configure, N s after launch)".
crash::startup_step records a startup step only until the app runs, and the
app's egui-wgpu hook calls it.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
While Frame shows the picture's frame, the preview projects into that frame
rather than the viewport, but Fit Product and the fit on entering Render
framed the product for the viewport's free area, so a picture taller than
wide cut the product at its sides. The fit now uses the viewport layout
seen from the frame (the frame as the viewport, the free area inside it
with the usual 16-point margin).
Entering Render fitted before the saved setup replaced the draft, so a cold
workspace with a portrait setup fitted for the old draft's aspect; the fit
now runs once the setup in effect is loaded.
render.camera.fit takes the picture's aspect (0.000001 to 1000000) through
Camera::fit_aspect, which publishes only a valid fitted camera: an aspect
that would divide the distance into infinity is refused and the previous
camera, document and undo stay as they were.
* Browser: nest groups and reorder components
Groups can hold other groups (up to 8 levels) and be reordered with the
items around them. Sub-components can be reordered too, through a new
`components` list in `browser_order` that is not drawn as a folder.
Document: `BrowserGroup` gains `parent`; group items and `browser_order`
lists hold `BrowserEntry` values (a number is a group, a string an item).
One pending format file; older designs read unchanged. Loading and
refreshing drop what the lists name that is gone, and cut cycles and
over-deep nesting to the top level.
Engine: `browser.group` takes `parent`; `browser.move`, `browser.order`
and `browser.ungroup` handle groups. A change to only the browser's
groups and order no longer re-evaluates the model.
UI: nested rows, dragging a group row (top of a row: before it, the rest:
into it), New Group inside a group, and dropping a placement on the top
of a sibling to reorder it.
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
* Fixed formatting
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
* Fixed unintended side effect in render scene browser test
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
* Preserve group order in render workspace
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
* Increased test coverage
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
* Updated clipboard test from upstream main for group item type
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Sonnet 5.5 <noreply@anthropic.com>
* claim: #9281
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Sketch text: a number parameter in the words shows its value (#9281)
A text's words expression can now name a number parameter: `'W' + Width`
reads "W85 mm" (lengths in the design's units, angles in degrees, up to
three decimals). `number(Width)` drops the unit and `number(Width, 2)`
fixes the decimals. The words and the cut follow a parameter edit in the
same rebuild, a rename reaches inside number(), and deleting the parameter
is refused while a text shows it.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Sketch text: a Text starting with = builds its words from parameters (#9281)
In Edit Text, a Text that starts with = is the words' expression
(='W' + number(Width)): OK keeps it, the dialog shows it again with its =,
and an untouched OK no longer turns it into typed words. Typed words still
replace it, and a saved text that happens to start with = stays typed
until it is edited. The field's tooltip says how.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Sketch text: the formula hint in every complete catalog (#9281)
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
* claim: #9170
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_019FAWPuFoh9qxn2uS2j53QZ
* Model cache: a design with mesh bodies keeps its saved model (#9170)
One mesh body (an inserted STL or 3MF, or a STEP part shown as a mesh of
its faces) left the model cache out for the whole design, so opening it
evaluated every feature again. The doc records refused mesh bodies, and
the native B-rep codec the io cache stores bodies with refuses them too.
Kernel work is paused, so the kernel is left alone: the io cache stores a
mesh body as its own slice kind (SCMESH01, a JSON header with the counts,
colour and paint, then the vertices as f64 and triangles as u32), read
back through mesh_body, with_color and with_paint. A mesh body is kept
only when its slice reads back as the same vertices bit for bit, the same
triangles, colour and paint. The decoder checks the counts against the
bytes before allocating. The picture shown while opening draws a mesh
body with its own triangles (native_preview_mesh refuses mesh bodies).
Cache version 4 -> 5; docs/design-container.md describes the mesh slice.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_019FAWPuFoh9qxn2uS2j53QZ
* Model cache: decode mesh slices with as_chunks, as clippy asks
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>
* claim: #9324
* Sketch: Project takes a construction axis, not only X, Y and Z (#9324)
sketch.project with {"axis": "Axis1"} upper-cased the name, and the axis link knew only the
origin axes, so a construction axis failed with "unknown axis `AXIS1`". X, Y and Z (any case)
are still the origin axes; any other name must be a construction axis, kept as its feature names
it. The link finds the axis where the timeline before the sketch puts it (construction_axis_in,
as Revolve does) and projects it as an origin axis: its line in the sketch plane, or a point
when the axis stands square to it. Renaming the axis renames the link, as for a construction
plane's projection.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Claude <noreply@anthropic.com>
* claim: #6262
* Inspect Sketch: the ribbon button opens a Sketch report window (#6262)
sketch.inspect gets a place on the Sketch tab (Inspect panel). ui.start opens a
read-only Sketch report: solver status, degrees of freedom left, counts of
curves, points, constraints and profiles, the curves and points that are not
fully constrained, and Copy JSON. engine.execute still only returns the JSON.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Inspect Sketch: a Sketch Report button on the Sketch tab while a sketch is open (#6262)
SKETCH/INSPECT placed it in the Inspect tab's Surface Quality drop-down, which
is hidden while a sketch is edited. It now has its own Inspect panel on the
Sketch tab, labelled Sketch Report; the test clicks the ribbon button.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Sketch Report: strings in a pending locale file, help for sketch.inspect (#6262)
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Sketch Report: the Sketch tab fits at 1366 px; the width test closes the report (#6262)
The panels left of the Finish group keep the gap before it when they are fitted,
so they collapse before they reach it.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Co-authored-by: Frank <plexi32@gmail.com>
* claim: #9001
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Sketch: a projected face follows a resize that moves its corners far (#9001)
A sketch on the face of an angle kept its projected face boundary (and
reported "projected face boundary identity is ambiguous") when the wall
thickness changed: the inner corner moved by more than a tenth of the face,
past the nearest-point match, and the same-order fallback refused faces.
Accept the same-order layout when, for every point, the point read in that
order is also clearly the nearest one on the same loop with lines the same
way. Rims and holes, and rotated readings of a rectangle, stay refused.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
* claim: #9341
* Insert Part: seat a fastener in several holes at once (#9341)
parts.insert / parts.fastener take `at` as one pick or a list of picks.
The part is built once as one component; each hole gets its own
placement and rigid seat joint. All placements are made before any is
seated, since joint.create takes the part's seating point where the
placement is. The insert now runs on a scratch session like
parts.resize, so it is one undo step, and a pick off every body fails
it before anything is made.
The Insert Part dialog's hole input is multi-select ("Holes
(optional)") and OK passes every pick.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Insert Part: picks on one hole's rim seat one part (#9341)
A hole's rim is drawn as two half-circle edges, and one click on it
picks both, so four clicks gave eight picks and two screws per hole.
Picks whose wall point lies level on the same hole axis, on the same
body placement, are one hole.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
* Point clouds: colour and filter by any property of a PLY file, on range tracks (#5151)
A PLY file's other vertex properties are kept as named numbers on the
points (whole-numbered for integer types; a comment `property: 0 name,
1 name` names a label's values). A cloud is coloured by any of them, a
ramp or a palette for whole numbers, and filtered by up to eight, each a
range or a set of named values. The GPU colours and filters the points
itself, so editing a range uploads nothing.
In the Edit Point Cloud card each range is a track, as in an audio
editor: a histogram of the property over what the other filters keep,
stacked by colour, under an overview strip. Ranges are drawn, moved,
snapped to colour changes, typed (Start, End, Length, relative values)
and played (Space or Play). The automatic colour range is exactly the
values the property takes over what the filters keep.
pointcloud.insert and pointcloud.edit take color_by: <property>,
color_range and filters; pointcloud.list reports them with the
properties. docs/point-clouds.md describes it all.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Point clouds: note the saved properties, colouring and filters as a pending format change (#8816)
A design now saves a cloud's per-point columns and their value names, colouring by a property and
filters; an older release must refuse it instead of dropping them.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Problem: every state of a saved model wrote the evaluation key of each sketch before it, so a
design of many sketches and many features after them repeated the same map over and over. In
the SELENE rover (5,484 features, 2,116 bodies) that was 27 of the 33.5 MB manifest: the
manifest hit its 32 MiB limit, the save thinned the cache (#3220) to 2,140 of 5,484 states, and
every open evaluated the 3,344 left-out features again. Reading the kept states one after
another took a few seconds more.
Solution: as for sheet bodies (#6152), the records keep a list of sketch-key sets, each distinct
set written once, as the keys it sets and drops relative to the set written before it; a state
names its set by place. A thinned model keeps the whole list and counts it once against its
budget. The reader counts every set's keys, and each state's, towards MAX_HELD before it copies
them (a short record can stand for a large set), then builds the states on threads
(kernel::map_on_threads) once the shared records are read. The cache format is now /5: an older
cache is a miss and the design is evaluated.
SELENE's manifest is 15.9 MB with every state kept; opening it evaluates nothing. Parse and
evaluate on opening, release build, median of 5 on a shared machine: 7.36 s -> 1.14 s. The
evaluated bodies are identical (inspect.measure of all 2,116 bodies).
Claude-Session: https://claude.ai/code/session_013WUFq6QFBqmZjxwaKy2J1M
Co-authored-by: Markus Kaltenbrunner <kaltenbrunnermarkus@users.noreply.github.com>
* claim: #8429
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Print face: Send to Slicer lays a body on its marked face (#8429)
A body can have one flat print face (Document::print_faces, held by the
certified target a sketch on a face holds). body.print_face {body, point?}
marks or clears it. Send to Slicer, Save Mesh with a slicer and EasyPrint
turn and move the exported copy so the face lies on the bed at Z = 0 with
the centre of mass over the origin; a face that can't be found is a
warning and that body is exported as it lies. Move/Copy and Mirror give a
copy its source's print face. One pending format file; the command
has its own icon (print_face).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Print face: Ctrl+B, the face menu and the Send to Slicer row (#8429)
Ctrl+B and the face context menu mark the selected flat face as its
body's print face, or clear it. Send to Slicer shows the print face of
the one body being sent in a Print face input, takes a flat face picked
in the viewport and clears it when emptied.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Print face: a light tint in the viewport and its strength in Preferences (#8429)
Each visible body's print face is tinted towards the theme's section
colour and outlined, at the strength set under Preferences > General >
Slicer (print_face_highlight, 0 to 1, 20 % by default; 0 hides it). The
user guide describes the print face, and the known limitation after a
shell of a shelled body is listed.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
`schema::issues(id, params)` looks up the command's typed contract and
lists what is wrong with the parameters (none without a contract or for
an unknown id). The batch preflight, the MCP bridge's pre-send check and
the app's check of a design-switching command each did that lookup
themselves; they now call it. No behaviour change.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
sketch_consumed knew Extrude, Revolve and the sheet metal flanges but not
a Hole drilled at a sketch's points, so that sketch stayed drawn inside the
holes. A Multiple hole now counts as using its sketch; Edit Feature shows
its points while the dialog is open. Single holes keep their sketch shown,
since a hidden sketch offers no points to snap to.
A sketch that failed (a dimension set negative by a parameter) left every
feature built on it with "unknown sketch 1 (it must come earlier in the
timeline)", which points at the timeline order. The evaluation now notes
the features that failed earlier in the pass, and a feature whose input
sketch is one of them fails with "sketch `Sketch1` failed: <its error>".
That result is never cached, since the cache key holds neither the
sketch's name nor its error: a second failing value or a rename shows
up at once.
* claim: #9184
* Control channel: a request's own param_check level, applied and confirmed by the app (#9184)
`engine.execute`, `engine.batch` and `engine.script` take an optional
`param_check: "off"|"warn"|"strict"` in their params. The app runs the
request at that level through the same `run_checked` path and batch
preflight a headless session uses, puts its own level back, and every
reply says the level it ran at as `param_check` beside `ok`. A command
that switches designs keeps the app's level; any other word is refused.
Readings and exports answered on a copy off the window thread keep the
level too.
`control.capabilities`, answered at once on the control-server thread,
lists what the channel does beyond the methods every version has:
`{"param_check": ["off", "warn", "strict"]}`. An older app answers
"unknown method".
The MCP bridge asks it once per connection. When the app lists the
level, every call and batch goes with it and the app checks against its
own commands; otherwise the bridge checks before sending, as #9073 does.
At strict, a reply that doesn't say the app checked the call is an error,
and the next calls are checked by the bridge. Also: `add_param_warnings`
takes the reply by `&mut`, and `with_param_check` sits beside
`Remote::new`.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Control channel: a design-switching command is checked at the request's level
Review fixes for #9184. `component.edit_in_place`, `component.finish_edit`,
`view.next_tab` and `view.previous_tab` ran at the app's own level and
echoed `off`, so a misspelt call went through at strict. They now have
their parameters checked at the request's level before they run (and
still run at the app's own level, as each design keeps its own), and the
reply echoes the requested level. The MCP bridge at strict now requires
the echo to match its level, not just to be there. `param_check` is read
only by `engine.execute`, `engine.batch` and `engine.script`; other
methods ignore it. The off-thread test now shows the design change
computing off the window thread before it is refused.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
* Sketch corners: a round that would use up a whole arc is refused
A quarter arc of radius 5 running into a line along it, rounded with
radius 5, put the round's tangent point at the arc's other end and left
an arc of no length. round_corner now skips a round whose tangent point
falls on either curve's other end; with none left it says that radius
does not fit, as for two lines (found by a local trim fuzzer).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Sketch corners: a round where the curves meet smoothly is refused
A line running on from an arc's end along its tangent (off only by rounding)
has no corner, but the round's tangent points landed within 1e-7 mm of it and
the round came out as an arc of no length (found by a local trim fuzzer). The
corner test's line met its arc this way too and was checking that output; it
now meets the arc at a corner.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
* claim: #9179
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_019FAWPuFoh9qxn2uS2j53QZ
* sketch.inspect: each profile loop names a curve once, not once per segment (#9179)
A profile records one curve id per piece its loop walks, and a conic or
spline is walked as many short pieces, so sketch.inspect listed a single
Bezier curve 50 times in a profile's `curves` and as often in a hole loop.
The reply now names each curve of a loop once, in the order the loop walks
them (solidcraft_sketch::distinct_curves, as delete and profile references
already do). The solved profiles keep their per-piece lists.
The lists are now a profile pick as they stand: a whole_boundaries pick
refuses a curve named twice, and the regression test extrudes the wall
from the reply's lists.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_019FAWPuFoh9qxn2uS2j53QZ
---------
Co-authored-by: Claude <noreply@anthropic.com>
* claim: #2829 (construct.* contracts)
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Typed contracts for the construction axes, points and planes
The 19 construct.* commands get input contracts and FeatureResult as
their output, with shared types for references, planes, a plane or
face base, an axis line and a construction plane's name, each calling
the reader its handler uses (split out of geo, plane_param, the face
capture and named_axis_line).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
* claim: #9086 (edge lengths)
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Measure: an exact circle's or arc's length agrees with its circumference (#9086)
Measure gave a circle's or an arc's length_mm as the length of the
polyline drawn for it, which runs inside the curve, while the same reply
gave its circumference or arc_length from the exact circle. When the edge
is known exactly (the kernel's circle, or a sketch's own circle or arc),
length_mm is now its radius times its angle. A circle fitted to a mesh
body's polyline keeps the polyline's length.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Claude <noreply@anthropic.com>
* Release 0.27.1
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Render job test: wait for the cancelled job's reservation to be released
The worker drops a job's reservation just after the job turns terminal, so
checking admitted bytes right after wait_timeout raced (10 of 40 local runs
failed). Poll with a deadline, as dielectric_tests does.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
* claim: #9072
* MCP: --param-check checks calls to the running app (#9072)
`mcp --connect` (and `mcp` once it bridges to an app) took
`--param-check` and forwarded every call unchecked, while the app's own
session is always off. `Remote` now checks each `engine.execute` and
`engine.batch` against this server's command contracts before sending:
`strict` refuses a call, or a batch with such a step, without sending
anything, and `warn` adds `param_warnings` to the app's reply (each
step's result in a batch). A command with no contract here goes to the
app unchecked. `Auto` carries the headless session's level over when it
bridges to a late app. `batch_preflight::contracts` is the contract half
of the existing preflight, for a runner in another process.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
The check for a constraint that shrinks a curve to a point took under 1e-6
for a point. One solve could squeeze a 49 mm side to about 1.2e-6 mm and
the next take it the rest of the way, so neither solve tripped it (two
rectangles' corners made coincident, found by a local sketch fuzzer). Under
1e-5 now counts: in 4000 random sketches the constraints that still
collapsed a line went from 9 to 0, and every refusal rate stayed the same.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
* claim: #9015
Co-authored-by: Cursor <cursoragent@cursor.com>
* Evaluate bare length offsets in the design's units (#9015)
Feature expressions like baseLength - 1 were subtracting millimetres because val() parsed with mm defaults and only applied design units to a final unit-less result.
Co-authored-by: Cursor <cursoragent@cursor.com>
---------
Co-authored-by: Cursor <cursoragent@cursor.com>