Part 2 of #876: one walk over the references a feature holds (the sketches it's built from, the features it names, the bodies it works on) replaces three hand-written matches, with a golden record covering every variant.
Where SolidCraft can replace itself safely, the Update window gets Update now:
- AppImage: the new AppImage is downloaded next to the old one, verified and renamed over
it; Restart now starts it as SolidCraft exits.
- Windows MSI: the verified .msi starts (msiexec /i … /passive) once SolidCraft has closed,
after the usual questions about unsaved designs.
- Windows portable zip / standalone exe: the running solidcraft.exe is renamed to .old, the
new one takes its name (solidcraft-cli.exe too), Restart now; .old goes at the next start.
- macOS: the verified .dmg is opened (the app is ad-hoc signed and not notarized, so it isn't
replaced behind the user's back).
- Flatpak, Snap, deb, rpm, tar.gz: never replaced; the window shows the command.
Every package's SHA-256 must match the release's SHA256SUMS.txt (HTTPS, same release); a
mismatch deletes the download and changes nothing. With PUBLIC_KEY set in update/install.rs,
SHA256SUMS.txt.minisig must also verify (minisign-verify, MIT) with a trusted comment naming
the release. release.yml signs the list when the UPDATE_SIGNING_KEY secret exists and fails
when the app expects signatures but the secret is missing; docs/releasing.md has the steps.
Co-authored-by: Claude Opus 5.5 <[email protected]>
Pieces that share a name are numbered by where their centres lie. Two pieces lying almost level along the sort axis no longer swap numbers between runs because of noise in their centres' last bits, so a saved name keeps pointing at the same piece. Closes#569.
* 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]>
* GPU tessellation: a pipeline the device cannot build falls back to CPU meshing instead of crashing (#1098)
Build the tessellation compute pipeline inside error scopes; TessGpu::new returns an error,
and the viewport logs it and meshes bodies on the CPU. A Windows CI job builds the pipeline
with D3D12's FXC on the WARP adapter.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
* Tessellation shader: basis functions in one-dimensional local arrays, so D3D12's FXC compiles it (#1098)
FXC has to unroll loops that write an array inside a struct (Basis.n0/n1) or a 2-D array
(ndu) by a counter, and cannot with run-time trip counts (X3511). A2.2 basis functions in a
local 1-D array returned whole; derivatives from the degree p-1 functions; the pole step
offsets computed rather than indexed.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
---------
Co-authored-by: Claude Opus 5.5 <[email protected]>
At start, and at most once a day, a background thread asks GitHub for the latest stable
release (one HTTPS GET to releases/latest, nothing else sent; 15 s timeout; silent when
offline; GitHub's rate limit respected). When it is newer than this build, an Update button
appears next to Report a bug. It opens a window with the version, the release notes and
Download (the asset that fits this installation), Skip this version and Remind me later.
- Versions compare as semver; a nightly counts as the release it was built after, so it is
offered only a newer stable release.
- The installation is detected (AppImage, Flatpak, Snap from the store or a file, deb, rpm,
tar.gz, Windows MSI/portable/exe, macOS app); Download picks the fixed-name asset.
Development builds and the Flatpak (no network permission) never check.
- Preferences > General > Updates: Check for updates automatically (on) and Check now.
- HTTP via ureq + rustls with the operating system's certificates (desktop app only).
- docs: Updates and privacy page.
Co-authored-by: Claude Opus 5.5 <[email protected]>
A renderer-independent OpaqueSurface (colour, metallic fraction, roughness, dielectric IOR) with bounded inputs, an original GGX/Smith/Schlick reference response with Lambert diffuse, and exact piecewise sRGB8 decoding. Numerical contract only: no document format, command or UI change.
Loading a design raises next_id above the largest feature, component and occurrence id, so a missing or stale value can't hand a new item an id already in use, and duplicate ids are rejected as a damaged design. Id allocation uses checked_add, so an exhausted id space is an error instead of a wrap or a panic.
Part of #1016 (points 1, 2 and 4). nekomario28's #1061 reached the same fix independently; thanks to both.
Co-authored-by: nekomario28 <[email protected]>
Bodies imported with the same label come in as X, X (2), X (3). Renaming one of them no longer renumbers the others: body.rename now pins the current names of the feature's other bodies when it stores the new one, so later renames, materials, appearances and features still find them.
A tapered extrude now offsets spline and conic curves as well as lines and arcs, so sketch text and any other profile can take a taper, and the join of tapered letters onto a face finds its boolean result.
A chamfer of a hole corner, where the mitre extension runs into the wall, now takes the local blend and comes out as one bevel instead of a notched cut; where that blend can't build the edge, the cutters that used to succeed are kept (#814).
With a Chinese UI on Windows and macOS, the first CJK fallback was a Japanese font without the simplified characters, so one word mixed glyphs from two faces with different line heights and they sat on different baselines. For zh the Chinese face now joins the fallback chain before the Japanese one, and the chain is rebuilt on the frame the UI language changes; other languages keep the previous order.
sketch.text and sketch.edit_text took at, height, angle and text with a silent default when a value had the wrong kind, so a height of "H" made 5 mm text and at: ["H", 25] put it at the origin. An input that is given (not null) now has to be what it names, or the command is refused with the input's name, and a null at no longer passes as the origin.
The live preview of a join, cut or intersect now sends the tool body's scene as soon as the tool is built, then runs the feature for its error while the dialog keeps its spinner. OK pressed while the preview is still computing the same commands waits for it instead of computing the feature a second time on the UI thread; the session shares its evaluation cache with the preview's scratch copy.
macOS hands files opened from Finder, the Dock or open -a to the application delegate's application:openURLs:, which winit's delegate lacks, so AppKit showed "SolidCraft cannot open files in the SolidCraft design format". open_documents adds that method to the delegate's class before launch finishes and queues the paths for the UI, which opens them like files named on the command line.
This is the one place SolidCraft's own code uses unsafe: the app crate denies unsafe_code instead of forbidding it, and only the two macOS-only functions in open_documents.rs allow it, with the reason documented there and in AGENTS.md. Adapted from VectorCraft (see ATTRIBUTION.md).
Closes#997.
A planar union no longer leaves a sliver where an edge crosses its neighbour, so a sheet whose outline runs on past a flange's edge with a slight bend unfolds again.
The command search asked its text field for focus on every frame. On Wayland that restarted the text edit's held key, so the key kept repeating in other windows after the palette closed. Focus is now requested once, when the box opens, with a test that the field doesn't take focus back while the box stays open.
Fixes#724.
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]>
On Windows the temporary folder often has an 8.3 short name, which the path scope refuses. vfs::long_names spells folders SolidCraft derives from the environment with long names, and the MCP scratch folder and the recovery folder use it, so screenshot without a path and the recovery commands work there again.
Pasting an Import together with later features that name its bodies now points those features at the pasted copies, including bodies whose names carry a number such as X (2).
Since the model's mesh went back to the reference mesher (#940), the GPU viewport drew
opaque bodies from their display plans while hover and selection lit faces with the
reference mesh's triangles. Two triangulations of the same face at equal depth fought, so
a hovered face shimmered in wedges instead of lighting whole.
Faces lit over a body drawn from its plan now use the plan's own triangles
(ModelBody::plan_mesh, evaluated once on the CPU), and the highlight pass gets a small
depth bias toward the eye so f32 rounding between the GPU and CPU evaluations of the same
triangles cannot speckle it either.
Test: every face of a box with a pocket and a round boss, hovered and rendered offscreen
through the viewport's real pipelines on the GPU path, lights the same pixels as on the
CPU path (skipped without a GPU adapter).
Co-authored-by: Claude Opus 5.5 <[email protected]>
The template still carried the 0.9 notes, and nightlies publish it as is. Keep the introduction, the upgrade notes that hold for every version and the install instructions; the release-specific part is written for each release.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
* feat: validate physical material cards with sources and conditions
* feat: query directional elastic modulus from physical material cards
* fix: register physical material validation labels in complete locales
* fix: register directional elasticity labels in complete locales
* Add German and Korean labels for the material property commands
---------
Co-authored-by: bherbruck <[email protected]>
* fix: a modelled thread that would cut nothing is refused (#995)
An M5 modelled thread in a 5 mm hole cut nothing into the nut: the
internal groove runs from the minor diameter out to the major one, so
with the hole as wide as the thread it lay entirely in the hole, and the
faceted cut only left the wall faceted with slivers ("the thread goes the
wrong way"). The fit check takes a 5 mm hole as an M5 hole.
- thread_groove measures how deep the groove cuts into the material
(hole wall out to the major radius, or shaft surface in to the root);
under a quarter of the basic depth 5H/8 the modelled thread is refused
with the measured depth, the tap drill (as solid.hole drills it) or
shaft size to use, and the size that fits the face, e.g. "this 5.00 mm
hole is too wide for a modelled M5 thread: it would cut 0.00 mm of the
thread's 0.43 mm depth; drill 4.2 mm for M5, or thread this hole M6".
- Cosmetic threads and their fit check are unchanged.
* test: pin the cut-nothing limit for coarse and fine pitch; show the hole size with show_mm
---------
Co-authored-by: MateuszSychXpl <[email protected]>
Co-authored-by: bherbruck <[email protected]>
The dialog_docked scenario check and the ui_layout width test looked up
the dialog window by the id "sc_dialog", but dialogs are now registered
under ("sc_dialog", <title>). Every lookup failed with "no command
dialog", so the test had to stay ignored and could no longer catch an
oversized dialog.
Read the rect the dialog publishes each frame instead (viewport
dialog_rect in the check, "dialog:outer" in the test), and re-enable
every_dialog_opens_docked_within_its_width_limits. The exploded view
dialog it was ignored for now opens 365 px wide, inside its limits.
Claude-Session: https://claude.ai/code/session_01VbBYY5zfMXgq939AF9Vren
Co-authored-by: PhilippBischoff0 <[email protected]>
Co-authored-by: Claude Opus 5.5 <[email protected]>
Co-authored-by: bherbruck <[email protected]>
Convert to Sheet Metal without bend edges refused every plate whose bounding-box diagonal reached a thousand times its thickness ("no outline on that face"), and just under that limit a hole smaller than the mesh's sag fell out of the outline and the DXF. The outline's tolerance is now at most a tenth of the thickness: a thin plate is meshed again at that tolerance. Nothing new is stored. Tests: long thin strips, sheets thousands of times longer than thick, and a strip keeping holes down to 1 mm.
inputs_mut, inputs and expressions are now built on one exhaustive walk (walk_exprs / walk_exprs_mut) instead of two hand-written matches over FeatureKind. inputs() no longer clones the feature, plane chains are walked with the same nesting cap for inputs as for expressions, and a new variant without a walk arm does not compile. A golden record of every variant's expressions guards the walk.
docs/mcp.md gains a "Camera and close-up recipe" section: the orbit camera's conventions (radians, eye direction for yaw and pitch, orthographic half-height, which side inspect.section keeps) and a seven-step recipe for a close-up of one region of a large STEP model. Documentation only.
* Fillets and chamfers meeting at corners on curved faces, on top of #935
What #935 and #926 leave of #851 and the matrix:
- Corner build for blends meeting on faces of any surface (curvecorner.rs):
the knife-edge tip's round with the arc's or the back edge's, edges
picked in any order, a round ending in an earlier round of its radius
(as a feature of its own), and chamfers meeting at the arc's corner.
Its end curves are fitted by chord length: a uniform parameter
through points clustered at the ends doubled back on itself there,
and the display planner refused the face.
- A straight edge ending at a curved face (the back edge into the
knife-edge tip): rolled, or cut by the face in the planar corner
build (blend_corner.rs).
- A rolling ball turning about a rounded corner of its own radius
(a pivot: a piece of sphere or cone).
- Seams that are an earlier blend's straight contact (fitted curves)
count as straight.
- No spline across a joint where the strip's curvature jumps (a line
into a small round): each strip ends on the joint's section.
- Sections at a chain's vertices on the faces of the edge from it
alone.
- Every new fillet path passes #949's check for openings.
Tests measure volumes through the display plan, as #935's do.
Action matrix: 51 more fillets and chamfers pass (sweep, tangent arc
chain, sheet-metal hem, bracket holes), none fewer. Robustness seed 33
recorded (its last fillet builds differently; display and measured
volumes now agree).
Co-Authored-By: bherbruck <[email protected]>
Co-Authored-By: Claude Opus 5.5 <[email protected]>
Claude-Session: https://claude.ai/code/session_015bV2ntwLfUHV4JnCMPSyRi
* Corner blend tests: scale, seed 33's volumes, tighter engine bounds
From the review of #854:
- seed_33_display_plan_and_measure_agree: seed 33's bodies measure
the same through the display plan and the measure (2e-5), which the
golden's re-record rests on.
- the_arc_rounded_after_the_back_edge_matches_both_together: the order
that still leaves 0.05 mm³ at the tip, as an ignored test.
- corner_blends_hold_at_any_scale: the cases this branch adds on the
#851 part, at 0.01x and 100x, take off what they take at full size
times k³ (within 5e-6, checked to 1e-4).
- The engine tests on the part as drawn measure through the display
plan; bounds are about twice what each reads (2e-4 to 3e-3), and the
front corner's sets are checked against the kernel's CSG grid.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
Claude-Session: https://claude.ai/code/session_015bV2ntwLfUHV4JnCMPSyRi
---------
Co-authored-by: bherbruck <[email protected]>
Co-authored-by: Claude Opus 5.5 <[email protected]>
press_pull_on_the_enclosure_lid_is_prompt took 61-87 s of wall time on loaded runners against a 60 s bound, for about 9 s of CPU. It now checks the operation's CPU time (under 300 s) and how far it grows the peak memory (under 1 GB; the runaway it guards against passed 24 GB), from /proc/self where the system has it.
Co-Authored-By: Claude Opus 5.5 <[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]>
Adds the read-only material.properties.validate command: schema_version 1 engineering material cards in SI units are checked for numerical consistency and elastic stability, with derived isotropic moduli, missing conditions and sources needing review. Nothing is assigned or saved.
Thanks @0xlillith.
Fixes#971. At the node limit the offset certificate splits the fitted curve where it meets the exact offset midpoint, so curves with different parameter speeds certify within the same 0.002 mm tolerance. Associated offset fit nodes are hidden from display, picking, box selection and snapping until Break Association.
Thanks @Karl-VonRichter.
#970's lenient retry of a face's parameter boundary searches the whole surface as a last resort; on robustness seed 20 it found an edge's points across a cylinder's seam, half a turn apart, and the Cut failed. A lenient boundary whose points jump half the parameter range or more is refused, and the face is left undivided as before (#970's elbows step at most a fifth of the range). Seed 81's volume moved by 0.1 mm3 with #952's hole direction; recorded.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
Fixes#928. Extrude To / From Object and Revolve To Object now resolve their target in the feature's own component frame, so a face in another component whose occurrence is moved is found where it is shown instead of the extrude stopping at a flat plane through the picked point. The point is named after the nearest face, a named point keeps its named face when another component touches it there, and such features re-evaluate when a placement changes, so they follow the target when it moves later.
Co-authored-by: SychPL <[email protected]>
Fillet gets a Curvature (G2) radius type (type: "curvature", alias "g2"): a quintic blend whose size is the equivalent circular setback, for convex and concave straight edges between planar faces with a perpendicular planar end face at each end. Unsupported cases are refused rather than falling back to a circular fillet, and a blend that would cross a loop in its end face is refused. Persisted as a fillet style under format 19; labels in en/de/ja/ko/zh.
Co-authored-by: midasdf <[email protected]>
Touching sews now split 4-use straight edges on extruded walls (along their rulings), not only on planes. Faces round such an edge are ordered by the way they leave it and, where they leave it tangent, by their curvature. Sheets bound the material wedges as before, except where that would make a face touch itself (a hole touching the outline from inside); edges ending there pair the empty wedges instead, so each loop gets its own vertex. Coincident faces are still refused, plain sew still refuses vertex contact, and shell_groups is unchanged.
A Ø10 hole tangent inside a stepped Ø40/Ø20 boss now cuts exactly instead of failing with "wire is not simple". Robustness seeds 42 and 51 are recorded: their touching hole patterns in stepped bodies now cut exactly instead of by a tool nudge.
Co-authored-by: froglemonade2 <[email protected]>
The CPU renderer's line depth test allowed a fixed share of the scene's depth range (2e-4), which on large designs or close views let edges a millimetre or more behind thin sheet show through it. Each pixel now keeps the screen depth slope of the surface it shows, and a line passes where it is no further behind than that slope spans across the line's width, plus f32 rounding. Section caps use their plane's slope.
Edges lying on steep faces now draw their full width; edges behind a face stay hidden. Only crates/render/src/raster.rs changes; the GPU viewport is unchanged.
Co-authored-by: NOWN-DEV <[email protected]>
* Boolean nudge tried both ways: no sliver where a cut's end touches a round side (#897, #827)
Robustness seed 14 cuts a cylinder whose end lies exactly on the apex of the
plate's arc side. Since the display tessellation plans (#805) the generator sees
the arc's true extent, so the cut's base lands on -10 instead of -10.000000000000002
and the end is tangent to the arc. The direct boolean fails there and the
last-resort tool nudge pushed the end into the arc, leaving a planar sliver a
tenth of a millimetre wide; shell then picked that sliver and refused it. The
13 October merges were not the cause (the bisect lands on #805).
The nudge is now tried both ways and the simpler result (fewer faces) is kept,
so the opposite nudge parts the touching faces instead. The same change joins
the hinge barrel in #827, whose top lies tangent to the lid's underside where
the mounting leaf ends. Each result still passes the same checks as before.
Goldens: seeds 14, 26 and 85 now build to the end, 31 and 78 come out with
fewer faces. Seeds 6, 43 and 73 still stop at shell: their joins now succeed and
shell reaches corners it can't offset (a cusp where a box face touches an arc,
coplanar faces facing opposite ways), which is unchanged.
Fixes#897Fixes#827
Co-Authored-By: Claude Opus 5.5 <[email protected]>
* Nudge the other way only when the first nudge fails or leaves a sliver
Trying both nudges and keeping the one with fewer faces could prefer a result
that keeps two pieces joined by a hair over one that separates them (the
centred coplanar revolve cut). The opposite nudge now runs only when the first
one failed or left a face narrower than 1/200 of the part, and it is taken
when it works with fewer of those.
Goldens re-recorded on current main: seeds 14, 26 and 85 build to the end,
31 and 78 with fewer faces. The other changed seeds are main's own since its
last full record, identical with and without this change.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
* Nudge the other way only for slivers the nudge made
On the sensor enclosure's lid the first nudge left thin faces the operands
already had (thin walls, small rounds), so the opposite nudge ran a second
boolean for nothing and the press/pull took 79 s against main's 55 s. The
opposite nudge now runs only when the first one failed or left more slivers
than the operands brought; the press/pull test takes 36 s against 42 s on main
(same machine, one after the other).
Goldens re-recorded on top of #940.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
* Goldens on current main: seeds 31, 43 and 78 with the two-way nudge
Co-Authored-By: Claude Opus 5.5 <[email protected]>
* Keep the hinge and tangent-cut tests beside the #871 fillet test
Co-Authored-By: Claude Opus 5.5 <[email protected]>
* Goldens: seeds 33 and 91 as main's fillet changes build them
Co-Authored-By: Claude Opus 5.5 <[email protected]>
* Goldens: keep main's #950 record
Co-Authored-By: Claude Opus 5.5 <[email protected]>
---------
Co-authored-by: Claude Opus 5.5 <[email protected]>
The state probe opened every cache it listed; when a worker deleted one in between, the open made it again, empty, and the next wait timed out. The probe now reads each cache with caches.match over the builds' files and marks, which never makes a cache (3 failures in 20 runs before, 0 in 30 after). The worker only ever starts a complete build.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
Trimming the base at an edge flange's setback snaps corners within tolerance onto the trim line and merges away the zero-width run a ledge as tall as the setback left along it, so the ledge goes with the strip as a lower one does. Splitting outline edges at other cells' corners now skips points within the model's tolerance of an end, so a ledge a little taller keeps its sliver and the bend runs under it, also off the axes. Both used to fail with "sew: a degenerate edge". Nothing new is stored.
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]>
* Holes on round faces from the app, one hole per pick, and Flip for holes at sketch points (#967)
The Hole dialog takes any face: a hole on a round face drills against the face's normal where it
was picked, and its arrow shows that. A round face split at its seam is picked whole, but makes
one hole where it was clicked. Holes at sketch points get Flip (saved as flip in the hole's points,
under format 18), in the command and in the dialog.
Co-Authored-By: Claude Opus 5.5 <[email protected]>
* Holes at sketch points show their arrow: into the part, or away with Flip (#967)
Co-Authored-By: Claude Opus 5.5 <[email protected]>
---------
Co-authored-by: Claude Opus 5.5 <[email protected]>
Sketch Mirror, rectangular and circular Pattern, and Offset now show a green live preview that follows the selection and settings; OK commits, Cancel leaves the sketch unchanged. The shared dialogs show the current selection step and source counts, sources can be toggled with Ctrl/Cmd-click, cleared and reselected.
Offsets keep editable driving distance dimensions (One Side, Two Sides, Symmetric) that update the derived curves without moving the source, with undo/redo and save/reopen. Format 19: older files keep their fixed offsets.
New dialog labels in en, zh, ja, de and ko.
Fixes#696. Fixes#517.
JIJINBEI