A design built over MCP is many small commands, and each one did work over the
whole design, so building or showing a 1,300-body design took minutes:
- moved components were transformed and meshed again after every change; world_state
now reuses each placed body whose definition and placement are unchanged
- next_d_name searched the design once per candidate number
- Document::eval evaluated every parameter for every expression; now only when the
expression names one
- param_rows' used_by was quadratic; the desktop no longer gathers completions on every
change while no text box has focus
solidcraft-cli bench-agent builds a generated ~1,300-body bicycle command by command
and times it (241 s -> 16 s here). Eval output is unchanged apart from timings.
Co-authored-by: Claude Opus 5.5 <[email protected]>
Thin Extrude builds walls of a set thickness along sketch curves, open chains and closed loops, with the wall centred, inside, outside or on both sides at two thicknesses. Saved under format 20 (thin on an Extrude); an extrude without thin saves as before. Closes#721.
* Straight edges stay one edge after cuts (#1049)
Co-Authored-By: Claude Opus 5.5 <[email protected]>
* Robustness goldens: cuts join straight edges along untouched walls (#1049)
Co-Authored-By: Claude Opus 5.5 <[email protected]>
* fmt
Co-Authored-By: Claude Opus 5.5 <[email protected]>
* Robustness goldens note: what each changed seed shows when replayed on both builds
Co-Authored-By: Claude Opus 5.5 <[email protected]>
---------
Co-authored-by: Claude Opus 5.5 <[email protected]>
timeline.export_script (and solidcraft-cli script) writes a design's history as a command script that solidcraft-cli eval replays into the same design, so a bug that only shows on one design becomes a one-file repro. timeline.add restores one stored feature with its id, names and references, through the same checks a design file gets when it opens.
Closes#886. SychPL's #1078 approached the same issue independently; thanks to both.
Co-authored-by: SychPL <[email protected]>
The sampled pi-case hole on Body1/F13:side:[email protected] became a real hole with #952 (7.5 s instead of 170 ms); beside the heavy tests on main's push runs it passed the matrix's 90 s action limit in its edit round, reported as the design not loading. The sample now takes every CPU in nextest so nothing runs beside it, and the edit round writes a start line so a hang there is reported as <action>+edit.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
A hole drilled along a panel, into its rim, has no place in the flat pattern
as a cut-out, so every such hole was refused. It is now kept as a bore of that
panel, in flat coordinates, and cut only from that panel's material when the
folded part and its flat pattern are built, so the sheet still unfolds and
folds back with it. A hole that would run into a bend, or cross a panel at an
angle, is still refused. The flat pattern lists the bores.
The action matrix worker takes ACTION_MATRIX_ONLY (just these actions) and
ACTION_MATRIX_UNFOLD (unfold and refold every sheet after each action) for
rechecking one group by hand.
Co-authored-by: Claude Opus 5.5 <[email protected]>
Fillets and chamfers ending on a boss's pieced wall or beside it, on a face that runs along its seam twice, and loops the loop blend refuses
- A straight edge whose end meets one curved surface in several faces (a cylinder a join left
in pieces) ends on that surface: the section is cut where it crosses each seam, which is
shortened there; where no seam is crossed (the wall missing inside the other body) the
section lies on the bared surface and the rest of the side edge becomes a seam.
- Two end faces that aren't both planes meeting in lines: each face takes the section's points
on it, cut where it crosses the edge between them. Not where an end face runs on smoothly
from a side (an earlier round): the corner build takes those.
- Face boundaries close over several connector edges.
- A face on a periodic surface running along its seam twice (as a STEP file gives it) is
rebuilt the same way (#691).
- A curved chain ending square on a face that runs on past its side extends that face's seam.
- A smooth chain stops before an edge between faces that meet smoothly.
- A loop the loop blend refuses for its shape is blended edge by edge; refusals about its size
stand.
- Action matrix: ACTION_MATRIX_ACTIONS and ACTION_MATRIX_VALIDITY for quick local runs; the
fillets and chamfers that pass now recorded.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
A through-all hole's tool is twice the largest body long, and the boolean's tolerances
followed that length: a 6 mm hole through a 920 mm part was traced at nearly 1 mm, wider than
the hole, and the cut came out empty or failed. When every other strategy has failed, the
boolean is now asked again (unshifted, then once shifted) at tolerances scaled to where the
bodies meet: the overlap of their boxes and the thinner operand's thickness. Its result is
checked as the others are and held to the retry guard. It runs last, so whatever an earlier
strategy answered is kept as it was, and only for single-shell operands (bodies in several
pieces are retried a piece at a time, each with this pass of its own).
The sampled shortcut also gave up on a slender tool lying across its box (a long hole drilled at
an angle fills little of it): it now samples past the first 2000 points until enough land
inside, so a tool that misses a body leaves that body as it is. Bodies that had enough samples
in the first 2000 are sampled exactly as before.
Baseline: the 90 actions that newly pass, rechecked one by one with the final build.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
Display meshes after opening, the face names read when a feature finds a face by name, and inspect.measure over all bodies are now worked out per body on plain scoped threads (solidcraft_kernel::for_each_on_threads / map_on_threads), in body order and with the same results; the browser build stays serial.
solidcraft --control PORT writes a random per-launch token to a user-only file (<data>/control/PORT.token; 0700/0600 on Unix, icacls on Windows; SOLIDCRAFT_CONTROL_TOKEN overrides) and refuses any connection whose first request lacks it. mcp --connect, the demo and packaging smoke clients send it. MCP execute/batch/open/save/export/screenshot are annotated destructive and open-world, new_design destructive. Docs: control-protocol.md, mcp.md, AGENTS.md, pending release note in releasing.md.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
Agent tool calls (MCP, and bridged control requests carrying allow_paths) reach files only in the open design's folder (strictly inside home or the temporary folder, not in a hidden folder, and never its hidden files), the app's recovery folder, the server's private scratch folder and each --allow-path; --allow-any-path lifts the limit. Paths are resolved first, so .. and symlinks that lead out (including a save through a link, #845) are refused. On Windows, UNC, device, NT, verbatim non-drive, drive-relative, alternate-stream, reserved-device and 8.3 short-name paths are refused before any file system call, and the resolved path must be on a local drive. doc.recovery_discard and doc.recover {discard} delete only real recovery entries. Docs: mcp.md, control-protocol.md, README.
Loading a design (from a worker's start or restart to its first output) gets 300 s; each action keeps 90 s. oracle/41-tube-radial-holes (about 23 s here) timed out loading on a loaded CI runner, and reloads after a hung action were charged to the next action.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
Adds oracle case 81 (a box edge rounded at a clamped cubic B-spline radius) with analytic expected values, and passes spline fillets through recipes.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
truck's face and edge IDs are their addresses, so walking maps keyed by them followed heap layout, and the ribbed lid's side-wall hole passed or failed by chance. The planar boolean's rebuild and glue now take truck's connected components in the shell's order, and the planar boolean's own maps are ordered. Healing's union-find keeps the lesser root and takes a run's ends in order of use. The face classification after an intersection orders each component's faces by the shell first. The lid now builds the same body on every run, with a test that builds it six times in one process. Its 0.5 mm side-wall holes, and one sensor enclosure hole of the same kind, now fail every time and are marked failing in the baseline.
Fixes#707.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
The nearest point on a triangle with a zero-length edge (a cone's tip) is no longer NaN, so an axis or tangent plane on a drill tip works again (#567).
A cut built slab by slab (the exact prism boolean) sewed the result from horizontal
slabs and only joins merged the coplanar pieces back. So a floor vent cut through the
Raspberry Pi case split the long wall into strips at the ports' tops and bottoms
(87 faces instead of 79 at 32 mm), and at 24 mm the end opening split the end wall.
- Prism booleans merge coplanar neighbours for every operation (faces only: edges,
curves and surfaces stay exact), keeping each face's construction source where all
its pieces share it, Split Face lines and regularization protection. Painted or
protected operands keep their authored pieces as before.
- The prism history now tells apart a wall's several edges with its caps (a wall
with openings meets each opening's floor), so cuts beside existing openings keep
exact history and faces keep their names.
- Tests: the Pi case's faces pinned after every feature at 24, at 24 changed to 32,
and built at 32 (each outer side one face); random openings and bosses on a
shelled case never split another side; every corpus design rebuilds a parameter
edit like a fresh build and reverts exactly.
- Action matrix baseline: demo/pi-case re-blessed (fewer, merged faces; no
regressions), demo/pi-case-0.8 recorded, and 88 new passes on
regressions/lid_ribs_flush_with_window.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
Sheet metal rims and bend ends attach by name (<panel>:rim.<edge>, <flange>:rim.bend0/1), as do face profiles from sheet faces. A plain or counterbored Hole square to a panel is machined into the flat pattern. Boss, Lip, Rest and Snap Fit keep the body's history and name their own faces. Hole tools name their faces by the part they cut, and pattern copies as copies. A pattern's repeated fillets keep the body's history. Loft sides are named by their section curves. A piece Offset Face moved off its face's plane attaches by its own qualified name. Sides of a profile grown to overlap a flush body keep their sketch line's name, so walls a Join merged take a sketch (#681). Face sketch 75% -> 90%, offset plane 75% -> 90%, press-pull/extrude 64% -> 85% in the action matrix; baseline blessed.
Fixes#494, fixes#495, fixes#560, fixes#561, fixes#563, fixes#564, fixes#681.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
Joining ribs flush with a lid's window left a zero-thickness slit far from the rib (the planar fallback glued back only one of the faces its pieces shared) and displaced the tool on the other joins. Gluing now joins every shared face pair, coincident-face separation checks there is material behind a pushed face for every join, and any join or cut that needed a retry must be closed, keep volume monotonic and leave the part unchanged outside the tool's box, or it is refused with an error. Feature command replies carry the feature's warning.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
A single edge's blend no longer needs a plane square to the edge at each end. Straight edges: each rail ends where it reaches the end face's surface, the side edges there are shortened or continued along their line or circle, and the blend ends where its cylinder or chamfer plane meets that surface (a fitted curve checked against both). A chamfer may end across two faces. Curved edges: at a slanted or curved end the rolling ball runs on past the end and the strip is cut back where it meets that face. A straight edge running smoothly into a bend or rounded corner is blended along its whole smooth chain. Rolling-ball fixes: smooth chains follow edges stored against their curve, the ball touches only the faces of the edges it rolls along, the seam normal at a joint comes from the face there, and each strip is fitted with neighbouring sections at its own spacing.
Action matrix: edge.fillet 83.6% -> 93.0%, edge.chamfer 82.6% -> 93.3% (697 new passes, no regressions); reference tests 76/76.
Fixes#556
Co-Authored-By: Claude Opus 5.5 <[email protected]>
Body::new no longer turns a long thin solid (a through-all hole's tool) inside out on the strength of a coarse mesh volume: it checks again on a fine mesh first. 79 more face.hole actions pass in the action matrix; the remaining no-result cuts stay under #559.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
A flat face that no construction certificate names on its own now attaches by its exact history name (a new lineage face certificate): cut walls and floors, chamfer faces, each pattern or mirror copy, side walls a Join merged, and the pieces a later feature split a face into. Construction certificates still come first; a seed keeps its certificate when it is patterned. Two flush extrudes of one outline joined end to end keep their exact history in the prism boolean. Face sketch 52% -> 71%, offset plane 52% -> 71%, press-pull/extrude from face 41% -> 60% in the action matrix; baseline blessed.
Fixes#491, fixes#492, fixes#493, fixes#557.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
Construction axes, points and planes resolve their picks against the model as their component sees it (in the root, every placement, pasted instances too), and a Hole picked on another component's body cuts that body. components-arm 355 -> 502/594 and joints-hinge 304/390 -> 371/408 in the action matrix.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
apps/solidcraft-cli/tests/action_matrix.rs tries sketch, offset/angle/mid planes, press pull, extrude from a face, hole, draft, shell, fillet, chamfer, axes, points, planes, projection, move, copy, mirror and pattern on the final bodies of 104 designs (the old design files, the examples, the 76 reference recipes, the enclosure demos, the Pi case built through the UI, eight typical parts), through the commands and picks the desktop sends, each design in its own process, with an edit round on a sample of passes. tests/action_matrix/baseline.json is a ratchet: an action that passed there and fails now fails the test; newly passing ones are reported for cargo xtask action-matrix --bless. Every CI run checks a fixed sample; the nightly Action matrix job runs the whole matrix in four parts. 30,567 of 35,576 actions pass (85.9%); results and causes in docs/action-matrix.md, one ignored reproduction per open cause (#556-#568).
Co-Authored-By: Claude Opus 5.5 <[email protected]>
solid.move takes copy: true to keep the source bodies and transform a new copy of each, with a Create Copy box in the Move dialog. Designs without it save and load exactly as before.
A profile traced over a face sketch's projected boundary left the extrude's caps and sides without an exact planar source, so a sketch or offset plane on the new top face was refused. The drawn curve is now the edge's source where it covers a projection. Clearer fixed-plane message; regression tests in the engine, UI harness and design corpus, plus a corpus-wide flat-face attachment check (known gaps under #491-#495).
Fixes#489
Co-Authored-By: Claude Opus 5.5 <[email protected]>
A parameter change can break a feature without failing the command that
made it; parameters.change reports it only in its `errors`. `run` then
exited 0 and exported the bodies as they were before the failing feature:
a shrunk sheet-metal tray whose unfold failed came out folded, without
the holes and cut-outs made after the unfold.
`run` now checks the evaluated timeline before writing anything. Any
feature with an error fails the run with status 1, lists the failing
features on stderr and writes none of the --out and --save files, as a
failing command already does. --allow-errors writes them anyway. `eval`
is unchanged and still reports the errors in its timeline.
Co-authored-by: Claude Opus 5.5 <[email protected]>
