document.units gets input and output schemas: `units` (mm, cm, m, in or
ft, with the handler's messages for a missing or unknown unit) and its
reply, `units` and `kept`, which the handler now builds from the reply
type. Its line leaves unconverted.txt.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
* Threads: a 12 mm hole takes M12, and a modelled thread narrows a bore that is too wide (#11609)
Automatic sizing of a hole now takes the size whose nominal diameter the hole has (a 12 mm hole
is M12, not M14, the size whose tap drill is 12 mm); a hole near no nominal diameter is still
matched by tap drill. A modelled internal thread in a hole as wide as the thread or wider no
longer refuses with "too wide": the bore is filled in to the thread's minor diameter over the
threaded length and the groove cut there, and the feature's warning says so. A size that does
not fit the hole names the size that does.
A saved thread with no designation keeps the old rule through a pending format step, so an
existing design rebuilds with the size it had.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Threads: keep an old automatic size through edits; kernel-rig cases expect the narrowed bore (#11609)
Editing a saved automatic thread keeps `old_auto_size`, so it does not change size. The five
kernel-rig cases that expected a modelled thread in a hole as wide as the thread to be refused
now expect the narrowed bore, with the volume worked out analytically in each case. The thread
dialog's reopen pattern names the new field.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Thread dialog test: a 5 mm hole is suggested M5, its nominal size (#11609)
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
* Typed schemas for Appearance, Appearance Library and List Appearances (part of #2829)
appearance.assign, appearance.library and appearance.list get contracts
that read what their handlers read, with the handlers' messages, and
reply types the handlers build their replies from. Their lines leave
unconverted.txt.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Appearance library: the opacity is already a number
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Every point on a spline surface returned its basis tables by value, up to
272 bytes several times over. Filling them in place takes about a quarter off
meshing a round extrude's wall, with the same numbers.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
* claim: #6257
* Align: move bodies by frames that follow sketch and model geometry (#6257)
solid.align takes from_frame and to_frame: a joint origin on the model
(a face's or round end's centre, an edge's middle, a corner, found again
by name) or a sketch point or a sketch circle's, arc's or ellipse's
centre. Both are resolved on every evaluation, so the bodies follow
their target when it changes. An optional angle turns the bodies about
the target's direction after the move.
The Align dialog takes sketch points and curves as From and To, offers
the Joint Origin snap and an Angle row.
* Align frames: box the joint origin (clippy large_enum_variant)
* Align frames: every moved body on the frame's component, placed once
* Align frames: a sketch id's length in characters
* Align frames: the dialog test's sketch curve hit carries its point
* Align frames: the dialog test takes Vec2 from solidcore-record
---------
Co-authored-by: MateuszSychXpl <151755914+MateuszSychXpl@users.noreply.github.com>
* New kernel cases go to solidcore's corpus; tests/kernel-cases/ only shrinks
tests/kernel-cases/CASES lists every rig and ladder case there is today. `cargo xtask lists`
fails on a case directory it does not list, saying where a kernel case goes instead
(solidcore's crates/solidcore/tests/kernel_cases/, exported with `cargo xtask kernel-cases
--export` or written by hand), and on a listed case that is gone. AGENTS.md, CONTRIBUTING.md,
the tests skill, the rig README and EXEMPT say the same; a new command gets an EXEMPT line
naming where it is tested.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01DtL5M41G8Nua7HvuFPZkmJ
* CONTRIBUTING: the kernel-cases paragraph ends before the naming corpus one
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01DtL5M41G8Nua7HvuFPZkmJ
---------
Co-authored-by: jyviko <2501648+jyviko@users.noreply.github.com>
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Each path test opens its own device and compiles the path pipeline. On a
cold driver cache that takes minutes and gigabytes per compile, and libtest
ran one per CPU thread side by side: on NVIDIA the solidcraft-gpu tests ran
for hours and committed ~80 GB. Executor::new now holds a process-wide lock
while it compiles; the first compile fills the driver's cache and the
waiting ones get the pipeline from there.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
The CLI's kernel_cases_export.rs still imported Brep from solidcraft_geom::brep after the layer split moved it to solidcore_record::brep. Changes this one import.
The second half of #11626: apps/solidcraft-cli/src/kernel_cases_export.rs (#11568) imported
solidcraft_engine::geom::brep::Brep, which #11496 moved to solidcore_record::brep::Brep.
cargo clippy --workspace --all-targets -D warnings is clean with it.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01DtL5M41G8Nua7HvuFPZkmJ
#11568's trace (crates/smlib/src/trace.rs) named the record as solidcraft_geom::brep::Brep,
and #11496 moved that type to solidcore_record::brep. Each merged green on its own; together
main does not compile (error[E0425]: cannot find type Brep in module b).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01DtL5M41G8Nua7HvuFPZkmJ
* Kernel: the B-rep record and the native body file come from solidcore-record
SolidCraft's portable B-rep record (`solidcraft_geom::brep`: brep.rs, topology, history,
envelope, part, curve2) and the native body file (`solidcraft_kernel::native`: the tagged binary
codec and the container) move to solidcore's `solidcore-record` crate: geometry as data, with
structural validation and no maths (solidcore's docs/layers.md, stage 1). Every caller uses
`solidcore_record::brep` and `solidcore_record::native`; the old copies are deleted.
What computes from coordinates stays with the kernel, in geom's `brep`:
- `brep::check`: the record's geometric checks (orthonormal frames, non-degenerate planes,
unit revolution axes, non-singular placements). `check::validate` and
`check::validate_built` are the record's validation and these together, and every caller of
`Brep::validate` uses them, so what is accepted is unchanged;
- `brep::label`: a surface's natural ranges (a cone's apex), `curve_interval`,
`edge_curve_interval` and `form_turns`;
- `brep::pcurve::eval`: a pcurve's point;
- `geometry::check`, `placement::moved` and `split_edges` become functions on the record.
The kernel's native module keeps the conversion between a body and its saved record
(metadata.rs) and checks the decoded record's geometry. `native/analytic.rs`, which no module
declared, is deleted.
solidcore-record is pinned to its rev on solidcore's `layer-record` branch.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01NvoxKMbMFSQU8tzRtoJhtT
* Geometry: Vec2, Vec3, Plane and the profile types come from solidcore-record
The plain data of vectors and points, sketch planes and planar profiles (`Seg2`, `Loop2`,
`Region2`) moves to `solidcore-record`, with its constants and component-wise operators only
(owner decision, solidcore layer split). Their geometry stays in geom as traits:
`Vec2Geom`, `Vec3Geom`, `PlaneGeom`, `Seg2Geom`, `Loop2Geom`, `Region2Geom`, `Seg2ArcFit`,
`Loop2ArcFit` and `Region2ArcFit`, in scope with `use solidcraft_geom::prelude::*`. Every caller
names the types from `solidcore_record`; geom no longer defines or re-exports them.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01NvoxKMbMFSQU8tzRtoJhtT
* Kernel: pin solidcore-record to solidcore main f5e8e9a29
solidcore-record is merged on solidcore main (tangency-labs/solidcore#2519); its crate there is
identical to the rev pinned before but for its clippy configuration. The lock is regenerated
against the merged rev.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01NvoxKMbMFSQU8tzRtoJhtT
---------
Co-authored-by: jyviko <2501648+jyviko@users.noreply.github.com>
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
It fails with status 1001 since #11360 moved SolidCraft onto solidcore alone, as the
kernel test variable_radius_on_a_box_edge already listed in tests/KNOWN_FAILING_TESTS.
The oracle keeps its own list, so main's Reference tests have been red since.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01DtL5M41G8Nua7HvuFPZkmJ
The web app is now offered at /app/ (latest release), /app-rc/ (newest release candidate) and /app-edge/ (current main), each with its own browser storage. /app/ is built from the latest release's web zip instead of main.
- Picks run on a copy of the operand: a pick builds query caches on the body it
reads, so the traced body stays as an untraced run leaves it.
- An untrim and an extension record their picks with the pickers they run
(`face_at`, `nearest_edge`).
- A final body matches the newest export of its record that an operation made.
- A stopped run is the kernel's when the last failed operation is followed by
copies only.
- A re-export removes each case's old entry first; fixes.json is written whole.
- The manifest names no commit for a copy without git, never an empty one.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01DtL5M41G8Nua7HvuFPZkmJ
An entry keeps what each operand body is and names <case>.bodies.json.gz, which
holds the records (gzip with no timestamp, so the same records give the same
bytes); the entry itself is pretty JSON. The corpus of 1202 cases is 33 MB.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01DtL5M41G8Nua7HvuFPZkmJ
An entry's operation is expressible when the kernel accounts for its result; the
values only SolidCraft grades (continuity, meshes, files, command answers, merged
counts, SolidCraft's wording of a refusal) are listed under `ungraded` instead of
making the whole case inexpressible.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01DtL5M41G8Nua7HvuFPZkmJ
`cargo xtask kernel-cases [--ladder] --export DIR` runs the rig with the kernel's
operations traced (`smlib::trace`) and writes one entry per case: the operation, or
chain of operations, that made the bodies the case ends with; each operand that came
in as a record, as that record; the parameters, with every pick by points also given
as the topology index it picked; the fix set; the case's expected values and the
rig's result. A case no kernel operation alone accounts for is written too, marked
not expressible with its reasons. DIR/manifest.json counts them per area.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01DtL5M41G8Nua7HvuFPZkmJ
The packaging action can't read a hyphenated version from the tarball's name, so the job gives it a name without the hyphen and renames the package back. A hyphenated version no longer fails the build if its Arch job fails; a plain version still does.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Bumps the version to 0.40.0-rc.1 and folds the pending format changes into format 49. The default-layout header test now expects the room the version really takes.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
06-threads/018-external-m10-both-hands makes 606 edges and 404 vertices (486 and 324 before),
with the same 204 faces, area and volume (to 1e-11), in two runs on the box; every other ladder
case is as the golden records it.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01DtL5M41G8Nua7HvuFPZkmJ
Five tests fail at the new pin and are listed with their causes: the variable round along a box
edge (status 1001, twice), the shell of a fully rounded box past its rounds' radius
(SM_ERR_INVALID_INPUT), the chamfer that would grow the part (no face instead of the
too-large refusal) and the box-edge round's wire record. The tabbed plate's outer loop with a
partial selection passes and leaves the list; issue2422's vertical edge now fails as a smooth
edge.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01DtL5M41G8Nua7HvuFPZkmJ
851-fillet-arc-then-top-front and 851-fillet-top-front-then-upright build (9524.3437, 10 faces,
solidcore#2431); 5158-fillet-round-just-past-a-10mm-face, the curvature pipe on a plate's
order-2 section, the concave-arc and front-top rounds (r5, r7 and default) and the peg's front
and concave rounds and chamfers build and pass their checks.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01DtL5M41G8Nua7HvuFPZkmJ
The rim alone was pinned to 7.958674274526857, the marched end rib's value; its closed form,
the spandrel (1 - pi/4) r^2 revolved about the disc's axis outside |x| < 4, is
7.958672911836132, which the rim's End (fillet_blend_corner) builds to 5e-8. With the corner
arc's Pappus volume the two take off 8.117808752; solidcore 8ddce9016 removes 8.117808800.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01DtL5M41G8Nua7HvuFPZkmJ
At solidcore 8ddce9016 (pinned) the r2, r2, r3 corner at (20, 0, 20) is one patch: 10 faces,
7928.743041456134 mm³, the r3 round and the patch the second fillet's. The test is pinned to
that and leaves the known-failing list. The fork's setback corner (15 faces, 7928.142563938)
waits for solidcore#2310; solidcore#2382, not merged, moves the case to 7927.724282512234.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01DtL5M41G8Nua7HvuFPZkmJ
smlib's run() catches a solidcore panic and replaces the kernel (kept: the app does not die
natively), returning Error::Internal; the kernel turned it into KernelError::Failed and the
engine into Other, so hostile_params_never_panic and the fuzzer no longer saw kernel panics.
Now the kernel's wrapper errors go through KernelError::smlib, which keeps Internal; a body the
wrapper could not take because of a panic is Internal too. A feature whose evaluation hit it
records the result as internal (not cached: the kernel was replaced), and the command that
made it, or any command whose kernel call returns it, fails with EngineError::Internal under
its own id. Crash reports mark a panic inside a solidcore call recovered (kernel::guarded), and
the crash test's kernel mode panics inside a solidcore call.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01DtL5M41G8Nua7HvuFPZkmJ
The kernel crash test's command must fail with EngineError::Internal naming
diagnostics.crash_test, as a panic in the command itself does: the panic detectors
(hostile_params_never_panic, the fuzzer) look for Internal. Today it is EngineError::Other.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01DtL5M41G8Nua7HvuFPZkmJ
solidcore's algorithms recurse (subdivision, intersection, filleting) and its README asks for
8 MiB, a native main thread's stack; Rust starts other threads with 2 MiB, and only the web
build set more. kernel::KERNEL_STACK is 8 MiB and kernel::kernel_thread a thread builder with it.
The kernel's parallel map runs on a pool of its own with that stack (on a worker of a pool
already, as in map_with_local_pools, it stays on that pool as before), the one-thread mesh pools
and map_on_threads' helpers have it, and so do the engine's interference worker and the
evaluator's plan-mesh thread. A test recurses 5 MiB deep on each.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01DtL5M41G8Nua7HvuFPZkmJ
While a dump folder was set, every call serialised each input body before it ran, so the rig's
timings (taken with a folder set) measured the serialiser as much as the kernel. The callers now
hand run_on the bodies the call reads; they keep them until its result is known, and only a
failing call serialises them, into the folder as before. A test counts the serialisations: an
answered volume and union none, a failing call one per input.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01DtL5M41G8Nua7HvuFPZkmJ
The check had become one-sided after the C++ kernel's removal: a case that started passing,
or a change in the count, went unnoticed. As on main, a case that now passes fails the test
until it is added to the list, one that no longer passes is a regression, and the count is 111.
At solidcore 8ddce9016 solidcore passes exactly the 111.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01DtL5M41G8Nua7HvuFPZkmJ