A homemade 3D input device's driver app (MAKMOUSE) listens on a loopback
port; with Preferences > Navigation > "Navigate with a device app on this
computer" on (off by default), the desktop app connects, says hello and
reads JSON lines: axis motion (orbit, pan or zoom by mode), zoom steps and
standard views. The motion goes through navigation::device like a 3D
mouse's, only in the focused window; a HID device the app names is left
to it. Documented in docs/3d-mouse.md.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
* claim: #6646
* Control channel: document.revision, a design revision and a document id (#6646)
A program that follows the open design asks document.revision and exports
only when the answer moves: document_id names the open design (kept through
edits, saves and renames, another for each design started or opened, never
shared between two tabs or two runs) and revision moves whenever the design,
what is hidden or an uncaptured component position changes. It stays put on
reads, selections and view changes, unlike the session's view revision.
The revision is read lazily: the session remembers the design it last
reported (held weakly, so Arc::make_mut can never change it in place and its
address can't be reused) and takes a new number from one counter for the
whole run when the design is another one. The control method answers at once,
also while a slow command computes; engine.execute reaches the same command
for scripts and MCP.
* document.revision: hash the run's id prefix with the process's random keys
A clock set before 1970 gave every run the prefix 0, so a restarted app could
answer a new design with the id and revision of the old one. The docs no
longer promise an answer at once: a screenshot waiting for an opened design to
be meshed still holds the requests behind it.
* document.revision: its label in the complete catalogs (ja, zh)
* claim: #6985
* Control channel: control.subscribe and design.changed events (#6985)
A connection that sends control.subscribe gets the active design's id and
revision in the reply (as document.revision answers), then a
{"event":"design.changed","params":{document_id, revision}} line whenever
either changes, so a live link need not poll.
The window publishes the design at the end of each frame's logic and
drawing while a connection is subscribed (one atomic load otherwise). Each
subscribed connection waits on that from a thread of its own and writes
the latest after a 100 ms gap, so a burst of changes is one line. Replies
and events share one lock per connection, each written whole (#6390).
* control.subscribe: the reply before any event, a second one refused
The events thread started before the subscribe reply was written, so a
change right after the window's answer could reach the client before the
reply that named the older revision. The subscription now starts holding
the connection's line lock and writes the reply under it.
A second control.subscribe on a subscribed connection is refused on the
connection itself: its answer could arrive behind a newer event. When no
thread or socket is left for the events, the reply says it did not
subscribe instead of ok.
* control.subscribe: a second one does not lose a finished late reply
The already-subscribed answer came before the busy check, so a request that timed out and had finished since lost its reply: control.pending then said the app ended it without one. It now comes after the busy check and drops a finished late reply as every other request does.
* control_server: declare the subscribe guard at the top of the loop
Clear of the lines #6643 changes (its token check moves to where the declaration was).
* control_server: the subscription's items beside serve
Away from the lines next to send, where #6643 adds token_of.
* control test: New on a blank design reuses its tab, so redo the box first
---------
Co-authored-by: MateuszSychXpl <151755914+MateuszSychXpl@users.noreply.github.com>
* claim: #6389
* perf: write each control reply to the socket in one piece
The control server wrote each reply with writeln!(out, "{reply}") on the unbuffered TcpStream.
Formatting a serde_json::Value hands the writer every string, number and punctuation mark
separately, so each became its own write system call. A reply of a few megabytes (component.list
on a design with thousands of components) took about half a second to send.
Every reply is now serialized to one buffer and written at once. Writing one large Value to a
loopback TcpStream: 1.79 MB took 1,052 ms with writeln! and 4.9 ms with one write_all.
The test counts the write calls for a 2000-component reply (one) and reads it back.
At launch the app adds the solidcraft entry to each installed AI client off the window thread,
names newly connected clients once, and refreshes a stale path. Settings > Assistant has
"Make SolidCraft available to AI apps on this computer" (on by default); off removes the entries it
added. Entries the user changed are left alone; automation runs never touch the files.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
* claim: #5467
* Warn when a design is open elsewhere or changed on disk (#5467)
The desktop app keeps a lock file next to each open design
(.~<file>.lock: computer, user, version, opened and refreshed times),
refreshes it every minute and removes it on close; a lock not
refreshed for three minutes is stale and ignored. Opening a design
that another SolidCraft has open asks: Open Read-Only, Open Anyway or
Cancel. A read-only design saves only to another file.
A design remembers its file's size and modification time when it is
opened or saved. Save over a file another program or computer changed
since asks: Save a Copy (the default), Overwrite or Cancel.
Programmatic calls never ask: doc.open reports open_elsewhere and takes
read_only, and file.save over a changed file returns an error until it
is given overwrite: true. ui.documents answers both questions.
* claim: #2063
* Windows OpenGL: use the system title bar so the picture lines up with clicks (#2063)
On Windows with only OpenGL to start, the window opens with the system
title bar instead of using the application bar as its title bar. The
window without OS decorations keeps its caption style while having no
frame, and some OpenGL drivers (older Intel HD/UHD) show the picture a
caption's height lower than the window, so clicks land above what they
point at. Direct3D 12 and Vulkan keep the custom title bar.
The choice is made before the window opens, from the backends wgpu
starts, so the window never changes its frame while OpenGL draws into it.
The AppImage writes a ~/.local/bin/solidcraft-cli launcher from its desktop-integration step
(same opt-outs, never over a file it did not write). The macOS app links the CLI into
/opt/homebrew/bin or /usr/local/bin at launch when one is writable, and Help > Install
solidcraft-cli command does it with an administrator prompt otherwise. Snap and Flatpak are documented.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
A Home picture still being drawn by the detached saved-thumbnail thread
could be on the GPU as the process exited, and the process died with
SIGSEGV inside the graphics driver. capture::close stops new pictures
and waits for those in flight (the app calls it in on_exit), and
dropping SolveApp waits for the Home picture threads it started.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
The taskbar asks a window for its icon once, when the window appears. eframe
sets the icon while the window is still hidden and shows it after the first
frame; when that frame is slow (safe mode's software renderer) or startup kept
the UI thread busy long enough for Windows to mark the window as not
responding (building the viewport pipelines), the request isn't answered in
time, the taskbar falls back to the window class's icon, and winit's class has
none: the button showed Windows' generic icon for as long as the window was
open.
Give the class the exe's embedded icon (resource 1) before the window is shown.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
* claim: #3988 (3D mouse: follow the device that moves, not the first one found)
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* 3D mouse: follow the device that moves, not the first one found (#3988)
On Windows and macOS the reader opened the first 3D-mouse interface it
could and read it until it failed. A 3Dconnexion Universal Receiver with
nothing paired opens fine, has a valid layout and never fails, so it kept
the reader for good. A SpaceMouse on a cable next to it was never read
("Not calibrated (hid 256f:c652)", every axis at zero).
The reader now opens every candidate and follows the one in use. The
first is followed from the start, as before. Another takes over once it
shows use (an axis off zero or a button down) while the followed one has
shown none for STALE_AFTER, so two devices moved together don't trade
the reading. A candidate that fails is dropped; if it was the followed
one the reading resets and the next is read. With one device the reader
behaves as before, down to the 100 ms read timeout. Several take 10 ms
turns, and reports queue in the system meanwhile.
The selection is `follow()`, written against a small `ReportSource`
trait, so it is tested with simulated devices on every platform: an idle
receiver next to a SpaceMouse, two devices in use at once, and devices
failing.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* 3D mouse: a press that makes a device the one in use counts as a press
Since #3825 a HID device's first button report after it connects is its
baseline: a button down in it is held, not pressed. A device taking over
from another one was connected at that moment, so the press that took
over was swallowed. With an idle Universal Receiver found first, that is
the first button press on the SpaceMouse.
follow() now keeps the buttons each candidate last reported while another
was read, and the reader hands them to the reading as the baseline when
that candidate takes over. A device that reported no buttons before it
took over keeps the held-not-pressed rule.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Tyler Chiriboga <18270301+Tyler-Chiriboga@users.noreply.github.com>
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
* claim: #5009
* Assistant: a request timeout per endpoint for slow models (#5009)
An OpenAI-compatible endpoint now stores how long one model request may
take (request_timeout_s, 300 s by default, clamped to 30 s ..= 1 h). It
is set in the Add agent dialog and in the endpoint's entry under Agents
(a running conversation uses a change from its next request), through
ui.agent_chat add_agent, and reported in the chat's state. A request
that runs out of time says the limit and where to raise it instead of
"timeout: global". The desktop HTTP cap goes from 900 s to an hour.
The tool-call timeout stays 330 s: it covers SolidCraft's own tool work
and the user's Allow, never the model.
* Assistant settings: line up an endpoint's Request timeout with Permissions
The label column of an agent's entry is wide enough for "Request timeout",
so the label isn't cut and both rows share one column.
* claim: #4664
* Control: name the token file and spot a stale one when a call is unauthorized (#4664)
When the app refuses a token that the MCP client read from a token file,
the error now names that file, when it was written and how long ago,
without printing the token. The app also writes `<port>.pid` beside the
token file (its process id and, on Linux, its start time) and removes
both at exit; when that process has exited, the error says the token
file is stale. The process check runs only after a refusal, and the
token file itself stays one line, so existing clients keep working.
Document force-quit leftovers and the MSIX AppData redirection.
* Bench assistant: read the token out of the new ClientToken
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
---------
Co-authored-by: bherbruck <brennen.herbruck@gmail.com>
Co-authored-by: Claude Sonnet 5.5 <noreply@anthropic.com>
* wip(render): reconcile completed material and image pipeline source with 0.20 (#5493)
Source-only integration reference for staged review, not a merge-ready aggregate.
Donor: 0xlillith/solidcraft@aea20b0115.
Upstream: 858e5995d6.
Retain original CPU/material contracts and explicit GPU donor provenance in
docs/render-materials-lab.md (#2067, #2191, #2780, #3306). Preserve released
format migrations and current navigation/startup/shell behavior. New saved
records use pending descriptors. The physical-card descriptor follows the
maintainer's material-engineering-cards naming request, with a legacy-lab
format-29 regression. Replace raw new panel hosts with shared containers.
Repository checks, complete workspace/all-target Clippy, and the independent
physical/resource authoring checks pass for this source. The fresh full local
CI invocation has reached test compilation and is not yet complete. Do not
transfer donor native output or prior test totals to this integration.
Generated review screenshots, video, package ZIP and machine logs remain at
the immutable personal evidence pin. Uncommitted geometric displacement is
backed up separately, not silently included. No kernel/vendor change, direct
main write, release or independent contributor-branch rewrite.
Co-authored-by: Christian Svensson <blue@cmd.nu>
* test(cli): pin the Compact camera-sample admission diagnostic (#5493)
The bounded request still refuses 1024x1024x64 samples before host initialization and writes no output. Update only the stale expected diagnostic to the existing Compact budget wording; production limits and code are unchanged. Exact-tree CLI regression suite: 14 passed, 0 failed. The full repository gate has been invoked separately and is not yet complete.
* Normalize UI container violation paths across platforms
Report repository-relative paths with forward slashes so diagnostics and the existing nested-package regression assertions agree on Windows and Unix. Keep the container detection rules unchanged.
Validation: official ui-containers gate and all seven focused xtask tests passed; rustfmt and diff checks passed. Only xtask/dependencies were compiled; no application build or heavy tests.
(cherry picked from commit 44b58d8e103a5fbc816ebf7c22b8de768ddf592b)
* fix(render): preserve camera contracts and correct acceptance regressions (#5493)
Preserve Design view.apply semantics with a separate strict Render camera command. Retain existing translations when English-only aliases collide; add material-axis entries and camera-specific pitch labels. Compare complete validated engineering cards and retain job handles before cancellation.
Focused and native acceptance is pinned to tree f8a423b21febd80ba88aaf1a67d5edbe2306314a. Full CI has been invoked separately and is not claimed green. Landing follows the render integration policy in #5493.
* fix(render): resolve showcase licence and upstream resource hygiene (#5493)
Record SychPL's explicit MIT OR Apache-2.0 grant, retaining source/model bytes and the Ubuntu lettering notice. Remove 80 absent attribution rows, personal fork records, dead delivery attributes and three fork/resource-only workflows. Existing xtask assets now performs bounded offline hash checks and rejects orphan render attribution; no parallel draft CI or package installation. Original authoring reproduction remains explicit. Scoped xtask tests, assets, Clippy, formatting and prepared SELENE validation pass; no application or renderer-wide result is inferred.
* fix(render): align saved bounds and progressive sample resolution (#5493)
Address review r4364211669/r4364211677/r4364211681/r4364211691/r4364211695. Persisted light/frame domains respect the existing 1e9 document bound; related LED/rectangle/direction cases use the same bound. Workstation primary resolve shares the 4096 admission cap and film counters use checked conversions. Exact candidate: 485 CPU render tests, 21 render document tests and strict scoped Clippy pass, including positive boundary round trips and 2048/4096 end-to-end checkpoints. No released format/migration edits, relaxed test limits or full-stack acceptance claim.
* wip(render): isolate optional resources and harden sample handling (#5493)
Review work in progress: opt-in encoded render library with reproducible runtime catalog exports; native-only robust dependency through the workspace; checked bloom slices; per-sample point-light diagnostics; independent area-light spatial/film keys. Resource-module binaries in both configurations, generation, Cargo production feature graphs and attribution checks pass. Full engine/render regression and combined acceptance are pending: do not merge this donor as a constituent or treat scoped resource proof as runtime acceptance.
* fix(gpu): preserve donor allocations through the scope-safety repair (#5722)
Port the native-validated #5723 scope-cleanup correction into the experimental donor, retaining its adaptive, volume, aging, photographic and bloom allocations. Same reverse-drain helper and regression cases; the donor combination still needs its own compile/test gate and is not covered retroactively by render-base results.
* claim: #1369
* Reject unknown pending concepts after format folding (#1369)
* test(render): close bounded sampling and bloom review regressions (#5493)
Validate the optional-resource/lighting/dependency donor at its exact tree. Required Vulkan GPU 127/0/3; CPU renderer 487/0; engine Render 72/0; document render 21/0; lean library 7/0; strict scoped Clippy, assets/layers/lists/containers and lean-engine WASM pass. Preserve original failed controller/typing attempts. This is the draft donor/disposition source, not a replacement for constituent render PRs or final main acceptance.
* wip(render): preserve stable aging target validation candidate
Exact staged source from prior review work on donor 3b892eab5e10f4508fa50f15b202630d5adfa594. This is a recovery/review checkpoint, NOT accepted delivery. An actual command-path regression currently fails because the selected executor refuses ordered surface aging. No assertion, capability or admission limit is weakened. Native/UI/full combination acceptance remains pending. Saved/runtime split adds one pending descriptor and retains original file scope; no main/render branch is changed.
* test(render): preserve numeric setup intent at the admitted coordinate bound
Follow up jyviko review on #5494 without weakening the corrected saved-coordinate contract. The positive clicked fixture uses 1e9 instead of inadmissible 2e9; explicit out-of-range cases still refuse, and the full saved setup round-trips through JSON and the native container. Exact-source Render workspace 95/0 and strict all-target UI Clippy pass. This does not incorporate the separately failing stable-aging candidate or claim full current-render acceptance.
* test(render): use the real GPU executor for saved aging
Move successful native cold-reopen Check/Start coverage from the CPU-backed protocol fixture into the production GPU executor tests. Verify capture-index changes preserve the stable target and that enabling aging changes both PNG and raw EXR output. Retain explicit, allocation-free refusal coverage for executors without aging support. Production capabilities and admission checks are unchanged.
* fix(render): persist stable identities for targeted aging (#5493)
Separate saved RenderBodyTarget identities from captured mesh ordinals. Reject legacy ambiguous targets, resolve Check/Start after visibility changes and preserve native reopening, copies and undo. Unsupported executors still refuse; real AMD Vulkan regression completes saved aging after reordered capture.
Source tree 6b84c8ad passes document 23, engine Render 75, GPU effects 5 and Render UI 95 with no failures; scoped strict Clippy, WASM and repository gates pass. Fresh CLI/native reopen and clean/aged GPU exports pass with identical physical totals and inspected actual images. Whole current-render integration is separate.
* fix(render): reconcile combined CPU/GPU contracts on current upstream
Unify captured resources/status and CPU-vs-GPU request schemas after the real two-parent integration; remove obsolete duplicate Vulkan selector and retain combined platform-filter regression. Deduplicate only identical or superseded test definitions, retaining the stronger card metadata and reserved-executor assertions. No assertions, admission limits or feature refusals are weakened.
All targets of render dependencies, GPU, engine, UI, CLI and native app pass cargo check with raytrace-cpu and bundled-render-library at tree c70d00d1. Native executable tests and full current-source acceptance remain pending. This is a private integration rehearsal for subsequent constituent PRs, not an aggregate official render/main merge.
* test(render): retain one complete host lifetime regression set
Deduplicate byte-identical DropProbe/task-lifetime tests and retain the adapted environment/texture ownership regressions with every assertion and complete memory estimate. Keep the registered-geometry admission regression once; only struct field order differed. Replace a one-element diagnostic loop with its equivalent block, preserving all wrap/width assertions.
Test-only integration repair. Formatting and diff checks pass; refreshed executable validation is required. No production behavior, limits, assertions or ignores changed.
* fix(doc): recognize the proven format-29 card descriptor alias
Normalize only the historical format-29 physical-material-cards spelling, and only when material-engineering-cards is registered. Admission and migration de-duplication use the same identity. Unknown concepts, wrong versions and incapable readers still refuse before mutation. Six controlled-history Rust regressions and actual CLI migration/no-clobber acceptance pass. No version, released migration or CI changes. Refs #5793, #1369, #5493.
* test(render): preserve grounding refusal before explicit instance moves
Update the two Render binding fixtures for upstream first-occurrence grounding. Verify grounded movement refuses without changing document/history, then explicitly unground through the public command before the intended move. Retain every target, resource, copy and persistence assertion. Refs #5493.
---------
Co-authored-by: Christian Svensson <blue@cmd.nu>
Co-authored-by: Victor Santos Oliveira <56417388+Vittuu@users.noreply.github.com>
Co-authored-by: cfaulkingham <colin.faulkingham@gmail.com>
* claim: #3083
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* WIP: choose the window system on Linux in Preferences (#3083)
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Preferences: a window system set over the control channel is saved for the next start again
The merge kept main's check in ui.prefs, which only looked at the anti-aliasing.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* Preferences: the window system section drops the restart badge its note already states
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
* Add a small test boat to steer around with the 3D mouse
A little hull in its own camera, drawn by the UI itself and not the CAD
viewport, so a 3D mouse can be tried out without opening a design. It
brings the model (generated by assets/models/make_test_boat.py) and the
tests for how the camera moves it.
Nothing uses it yet; Preferences picks it up with the rest of the 3D mouse
work.
* UI kit: keep a choice row from flipping with the scrollbar
A choice row picks segments or a dropdown by whether its labels fit. A page
scrollbar that comes and goes narrows the row by its own width, so the row
could flip back and forth between the two on a page that sits right at the
edge. It is now judged on the whole row and through a small helper,
fits_steadily, that only changes its answer once the miss is wider than the
scrollbar.
Also tests for that, for typed numbers in a slider's value box, and a
test hook to put up the issue-report input barrier by hand.
* 3D mouse: device model, guided calibration, buttons and a picture
The 3D mouse becomes a device we know something about instead of six raw
numbers.
The reader picks between spacenavd and the direct hidraw driver, falls back
from one to the other, and says so when it can't open the device. Guided
calibration asks whether the device is a spring cap or a trackball, adds a
touch step, and uses a settle filter so a ball let go of stops reading as
motion. Buttons get jobs (keys, held modifiers, commands), saved per
device, with defaults for the SpaceMouse Pro.
Preferences gets a picture of the device that moves as it does, next to the
test boat, and the same picture can sit in the 3D view at Left Viewport,
Right Viewport or Left Window, above the floating History card like the
legend and readouts.
These ride together because the crate's calibration types, the Preferences
page, the Linux reader and the button hook all use each other.
* 3D mouse: offer to calibrate an uncalibrated device
When a 3D mouse shows up that has not been calibrated, a card in the 3D
view offers to calibrate it, to swap drivers when the other one would work
better, to close, or to never ask again. It is made once per device, goes
away after 30 seconds in view, and gives way to a status or computing
toast instead of stacking on it.
The card is the shared toast, which now takes buttons: they group by
purpose, wrap inside the card at narrow widths, and do nothing while an
issue report holds CAD input. The picture of the device in the 3D view
keeps clear of the card.
* 3D mouse: translations and attribution
The 3D mouse strings (ui.3d_mouse_*, the calibration steps and the offer
card) in every shipped language, the Japanese catalog note, and the credit
for the test boat model in ATTRIBUTION.md.
This is last because the code before it only looks keys up: it builds and
runs without these, falling back to English until they land.
* 3D mouse: draw its preferences with the settings kit like the rest of Navigation
The 3D mouse section was a folded header with its own layout. It now
uses the same kit as the sections above it: a settings_section card,
settings_form rows with the shared label column, dividers between the
device, axes, picture and buttons groups, Text::Hint for the helper
lines, and settings_control_width for the dropdowns.
The section is open like every other one (the header still folds it),
so the device badge on the folded header is gone; the calibration row
already names the device. Every setting, string and key is unchanged.
Tests that unfolded the section no longer click it, and the hint lines
are looked up as hints instead of notes.
* 3D mouse: show each axis's live tilt in Preferences, like main's axis rows
Each view axis row is main's axis row again: a meter that follows the tilt of the motion driving it, with the dead zone marked, next to Invert, Speed and the "driven by" menu. The meter reads the calibrated travel when the device has a calibration and the raw reading over full scale when it does not, and asks for a frame whenever the reading changes. The helpers are back as they are on main, and the calibration state is a badge by the enable toggle.
* 3D mouse: line its preferences up with the rest of Navigation
- The two checkboxes sit side by side in the same two-column grid as Wheel and trackpad.
- Row labels stop repeating "3D mouse" and use sentence case: Calibration, Driver, Restore defaults, Reset view, Left viewport, Right viewport, Left window.
- Hints line up under the control column instead of the left edge.
- The six view axes are named Pan X, Pan Y, Zoom, Orbit X, Orbit Y and Roll so they fit the name column.
- Buttons are a divider and rows in a form like the other groups, without the bold sub-heading.
- The driver, dead zone, axes, picture and buttons sit behind a collapsed Advanced section, since most people never touch them.
* 3D mouse: tidy the Advanced test for clippy
Rebased onto main as one commit (was 120d26f7, df488f9a, db9d1073 plus
merges of main).
- A request the app reports as still running (#1618) is waited for on the
same connection through `control.pending`; past the wait the error says it
was not cancelled and will still apply (or only write its file), later calls
are not sent behind it, and the new read-only `pending` tool returns its
result once it lands.
- A failed batch says what really happened: `rolled_back` only when the
design is back where the batch started (as #3288 / #3242 on main: the app's
`rolled_back` is believed; for an older app, every completed step undone;
only a rollback asked for is reported as one), plus `applied`, `not_run` and
`failed.still_running`.
- A batch step may name a tool (set_parameter, list_commands,
describe_command, execute, measure, inspect_design, undo, redo, export, save,
open, new_design, crash_reports); screenshot, body_topology, batch and
pending are refused before anything is sent.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
The app opens its localhost control channel on port 27878 (or the next free
one) unless Preferences > Assistant turns it off; --control and
SOLIDCRAFT_CONTROL_PORT still win. A current.port discovery file lets
solidcraft-cli mcp find the running app with no arguments.
Co-authored-by: Claude Sonnet 5.5 <noreply@anthropic.com>
each_frame_hands_the_view_its_motion_through_the_drivers_mapping dropped
the output of a frame it ran (`let _ = ctx.run_ui(...)`). epaint panics
in debug builds when a TexturesDelta is dropped with deltas left ("Dropped
TexturesDelta with 1 unapplied deltas"), so the test failed under a plain
`cargo test` on every platform while the ci profile passed. It now clears
them, as the other UI tests do since #60.
Co-authored-by: Tyler Chiriboga <18270301+Tyler-Chiriboga@users.noreply.github.com>
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
#5187 gave the scene its own offscreen target with the app's sample count,
which skipped the OpenGL clamp to 1x from 6dd202d. The scene's count now
goes through effective_msaa_samples as well.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
* Viewport: show the picture it drew last when nothing changed (#4820)
A frame redrew the whole scene whenever egui asked for a frame, even when
the camera, scene, highlight and overlays were identical: pointer moves over
empty space and UI animations over a still view paid for the full scene.
The renderer now hashes everything the picture depends on (scene keys,
camera, size, section, images, analysis, accessibility map, occlusion) and
blits the last offscreen picture when the hash matches. The first still
frame filters a supersampled picture down once and later frames copy it as
it is. A frame with the same content costs a blit instead of a scene draw.
Co-Authored-By: Claude Code <noreply@anthropic.com>
* Viewport: draw the scene offscreen, into a 1x window pass (#4820)
The scene keeps the app's depth, stencil and MSAA, but egui's window pass now
has one sample and no depth. Whenever the two targets differ, the scene
(product previews too) is drawn offscreen and blitted, so the UI pass no
longer clears and resolves a multisampled window.
Over 8 s of pointer moves on the 450E rotor head, the app's GPU intervals fell
from 1325 ms to 1006 ms and egui_render from 892 ms to 509 ms; idle stays at
0%. A GPU test checks the split path's picture matches the direct one.
Co-Authored-By: Claude Code <noreply@anthropic.com>
* 3D mouse: a repeated report at rest asks for no frame (#4820)
A device may repeat its latest report while the cap rests. Publishing it
again asked egui for a frame although nothing changed. Only a report that
changes the reading now asks for one; while the cap moves, the view asks for
the next frame itself.
Co-Authored-By: Claude Code <noreply@anthropic.com>
---------
Co-authored-by: Claude Code <noreply@anthropic.com>
* Add the format registry: one table of formats, extensions and options in crates/io.
The hosts' Open-dialog filters and the web accept list derive from it, and
registry tests pin it to the path predicates and the command docs.
* Fix format-registry tests that compared FormatSpec without Eq.
assert_eq on Option<&FormatSpec> does not compile; the none cases
only need is_none, matching the other rows that compare ids.
* claim: #4172
* GPU: a lost device stops drawing instead of crashing (#4172)
After the driver resets the GPU (Windows TDR), the first queue write to meet the
lost device panicked inside wgpu-core (PendingWrites::activate unwrapping
DeviceError::Lost). Nothing watched the device for its loss.
- egui-wgpu: RenderState::device_lost (DeviceLost) is set by the device's lost
callback. The painter draws nothing on a lost device, and before a frame's
uploads it submits nothing (DeviceLost::probe), so a loss the driver reports
is found by that submission instead of the first write.
- Pictures (thumbnails of saves, ui.render) watch the same flag: on a lost
device the capturer refuses with GpuError::DeviceLost without touching it,
and the picture comes from the CPU rasterizer. Headless capturers watch
their own devices.
- The app, seeing the loss, closes keeping unsaved changes for recovery (as a
forced quit does) and starts again; the new start says what happened. A
second loss within ten minutes starts on the software renderer. The loss is
noted in the crash journal.
* claim: #2614 (3D mouse: find devices on Windows without asking every HID device for its strings)
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* 3D mouse: find devices on Windows without asking every HID device for its strings (#2614)
On Windows the 3D mouse reader rescanned every two seconds while no 3D
mouse was plugged in, with hidapi's refresh_devices(). hidapi's Windows
enumeration opens every HID interface and reads its serial number,
manufacturer and product strings: on USB, three control requests to
every keyboard, mouse and touchscreen, forever, on a machine without a
3D mouse. That fits the report of an idle machine whose USB keyboard
drops off and returns and whose cursor jumps.
Windows now asks Plug and Play which present HID interfaces are
multi-axis controllers (hardware ID HID_DEVICE_UP:0001_U:0008), which
sends nothing to any device, and opens only those with open_path().
hidapi's own discovery is turned off before the reader makes its one
context, so making it touches no device either. macOS keeps hidapi's
enumeration (IOKit's registry); Linux is unchanged.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* AGENTS.md: list the Plug and Play scan among the files allowed unsafe code (#2614)
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Tyler Chiriboga <18270301+Tyler-Chiriboga@users.noreply.github.com>
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Co-authored-by: bherbruck <brennen.herbruck@gmail.com>
* 3D mouse calibration: find each axis from a guided recording (#1463)
A recording of the cap at rest and moved along each axis both ways gives which raw axis is which,
which way it runs, where it rests, how much it wanders at rest and how far it goes each way.
Hands aren't exact: on our SpaceMouse Pro, sliding the cap right moved Ry further than X. So each
raw axis is picked by the one-to-one assignment that best explains all six motions together, each
raw axis scored against its own range so a wider one doesn't win through cross-talk. Each
direction is scaled on its own, so a device that goes further one way still reaches full scale
both ways; a direction under a twelfth of the furthest axis counts as not having responded.
A small state machine walks the user through it and moves on by itself once the cap has been held
out for 0.4 s in one go and let go. "Held out" is the least the solver accepts, twice an axis's own
rest wander, so whatever the solver would accept can be recorded and a noisy axis doesn't hold up a
quiet one.
A calibration also carries a dead zone the user sets, a share of full deflection taken off the
bottom of the travel, with full deflection still reading full scale.
Serde support is behind an optional feature, so the crate still needs nothing by default.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* 3D mouse calibration in Preferences, saved per device (#1463)
Preferences > Navigation gets a 3D mouse calibration row: whether the device being read is
calibrated, Calibrate... and Forget calibration. The window asks for each motion in turn and
saves what it found under the device (hid, web, spacenavd or 3dxware, with vendor and product ids), so
switching devices or turning spacenavd on doesn't apply the wrong mapping. A calibration is saved
for the device it started with, and ends if another one is read instead.
navigate_frame puts the reading through the device's calibration before the driver mapping, so
the per-axis Invert and speed and everything after them are unchanged. While the window is open
the reading goes to it and the view stays put; it closes with Preferences. Restore Defaults on
the Navigation page keeps calibrations, since they describe the device.
The web app hands over the gamepad's readings as they are, so calibration sees where the cap
rests; navigate_frame now leaves a reading inside the dead zone alone after calibration, so a
cap resting just off zero still doesn't keep redrawing. Whether the cap is moving is asked of the
calibrated reading there, in the web page's idle waker and in both wheel filters. Those run before
the frame, so they keep the last frame's calibration with its device and use it only for that
device. On the desktop, an axis with nothing fresh (timed out, or not reported since the device
connected) reads as the device's calibrated rest, so a device resting off zero doesn't drift.
A saved calibration that can't be read or used (a hand edit) is dropped on its own when the
preferences load, with a log line.
The window shows the device's raw readings live and has a dead zone slider, a share of full
travel on top of the measured noise; it's saved with the calibration and adjustable in Preferences
afterwards.
New strings are in all 14 catalogs.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* 3D mouse: choose which motion drives each axis (#1463)
After calibrating, which motion of the cap should drive which view axis is partly a matter of
taste: rolling and twisting swapped, say, or Y and Z. Each axis row in Preferences > Navigation
now has a "driven by" choice next to Invert and speed. Picking a motion another axis uses swaps
the two, so each motion always drives exactly one axis, and Reset axes puts them back.
The choice applies after the device's calibration and before the driver mapping, so Invert and
speed stay with the view axis rather than moving with the motion. It's a preference, not part
of a device's calibration, and stays in force when recalibrating; the calibration window says
so while one is set. A saved assignment that doesn't drive each axis once loads as the default.
New strings are in all 14 catalogs.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* 3D mouse: axis choice review follow-ups (#1463)
- An axis's speed, like its Invert, stays with the view axis when a motion is reassigned; tested.
- Calibrating records the device's own reading, before the axis choice, so a saved swap isn't
baked into a new calibration and then applied twice; tested through a whole calibration.
- A settings file without the axis choice, or with one of the wrong length, loads with the
default and keeps every other setting; tested.
- Each axis row says "driven by" in sight before its choice, which otherwise only explained
itself on hover (no help on a touchscreen or the web).
- "Reset axes" is now "Reset axis assignment", so it doesn't read as forgetting the
calibration just below it; the "driven by" tip reads more clearly in English.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* 3D mouse: name axis rows by what they do, and motions without a device in mind (#1463)
The axis rows and the "driven by" choices showed the same names, "Push left / right" and so
on, which read as one thing mapped to itself, and they described a cap that springs back to
centre, which a trackball or a knob that stays put isn't.
Each row is now named for what it does to the view: Pan left / right, Zoom, Pan up / down,
Orbit up / down, Orbit left / right or Roll, or Off. The name comes from the driver mapping
navigation used last frame, so it stays right where the mapping swaps axes (object mode swaps
the last two). Before navigation has run there's no mapping to read, and a row is named for
its own motion instead of a guessed effect. The choices are the device's calibrated motions:
Translation X / Y / Z and Rotation X / Y / Z, under a header row, and the calibration window
names a motion that barely moved the same way. This renames #1814's existing rows, which had
no translations; every new name is in all 14 catalogs.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* 3D mouse: rows never fall back to a motion name (#1463)
Without a driver mapping, a row was named for its own calibrated motion,
which is also what its dropdown lists. It now takes the default mapping's
view axis, so the two sides never share a name.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* 3D mouse: grey out Invert and speed on rows the driver turns off (#1463)
A row the driver profile turns off says so on hover, and its Invert and
speed are greyed out, since they do nothing there. Its motion choice stays
live so a motion can still be swapped out. The column header is shorter.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* 3D mouse: unmapped rows say Axis 1-6, and the off-row tip explains the swap (#1463)
Without a driver mapping, a row is now just "Axis n" rather than the
default mapping's view axis, which would be a guess there. The tip on a row
the driver turns off says that a motion swapped into it stops moving the
view, and a test swaps one in and back out.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* 3D mouse calibration: a device that stops short of rest after a release (#2778)
Through spacenavd, a relative-axis device like the Space Navigator never
reports its return to zero, so the reading stays at its last small report
after every release. The wizard never saw the cap let go, and once
calibrated, the stuck value crept the view.
A reading that stays exactly the same, close to rest, now counts as let go,
and where the cap settled after each step goes into the calibration: such a
device rests at the middle of where it settled, with a dead zone out to the
furthest of them. A device that comes back to rest calibrates as before.
* 3D mouse calibration: a steady hold through spacenavd stays a hold (#2778)
* 3D mouse calibration: stopping short of rest only counts from spacenavd and 3DxWare (#2778)
Recentring on where the cap settled kept the rest stage's wander radius but
not its offset, so a rest extreme could fall outside the dead zone; the
radius now grows by how far the centre moved. A reading that stops changing
close to rest only ends a step from a source that sends nothing at rest, so
HID and web devices calibrate exactly as before.
* 3D mouse calibration: settling off-centre counts on any device (#2778)
A drifting stick settles off its rest after a release just as spacenavd's
stuck reading does, but it keeps reporting, noisily. A step now ends once
the reading stays steady, each raw axis within twice its rest wander (so
exactly still from a silent source), close to rest, on any device. The
dead zone takes that wander on top of the furthest settle point.
* 3D mouse calibration: steadiness by the run's spread, settled on the step's own axis (#2778)
An axis that read flat at rest may dither once off-centre, so steady
readings may spread at least 2 units and a 50th of the furthest any step
went. Steadiness is the spread of the whole run, not distance from its
first reading. Close to rest is judged on the axis that set the step's
peak, since a device's axes may have different ranges, and the settle point
is the middle of the steady run.
* 3D mouse calibration: readings close to rest never count as held (#2778)
* 3D mouse calibration: the dead zone reaches every reading of each release (#2778)
Saving one reading of each release and padding it by the rest wander could
miss a reading seen in the same steady run. Each release now keeps the whole
run, and the dead zone is centred on and reaches everything read at rest
and after the releases.
* 3D mouse calibration: settle for half a second before a step counts as let go (#2778)
A hand held still close to rest is less likely to end a step early; a
device that comes back to rest never waits for it.
* 3D mouse calibration: show the dead zones as a game controller tester would (#2778)
Once calibrated, the window, and Preferences under the Dead zone slider,
show three pads: slide, press and twist, and tilt. Each has a circle for
the full travel, the dead zone shaded inside it with each direction its
own size and the user's dead zone on top, and a dot where the cap is now,
lit once the view moves.
* 3D mouse calibration: settle through cross-talk, per-axis steadiness floor (#2778)
The axis that went furthest in a step needn't be the one asked for:
cross-talk can move another raw axis further. A step now counts as let go
once the furthest axis is at least halfway back and some axis that moved
about as far is within a quarter of its own peak; a hand still holding the
push keeps the furthest axis out. The steadiness floor follows each axis's
own reach, and each stage's steadiness starts with its own readings.
* 3D mouse calibration: steady readings may dither ±2 on any axis (#2778)
* space-mouse: make the settle check's near band match the steadiness spread, test the 4-unit floor
near() measured distance from the settled run's middle against the whole tolerance,
so it accepted twice the band the steadiness check does. Use half. Add a short-range
test where only the .max(4.0) floor lets a +-2 dither settle, and one where dither
the spread check rejects is also not skipped as untouched.
* space-mouse: settle on the middle of the steady run so a lopsided dither isnt movement
near() allows half the steadiness spread around the settled point, but the point was the run mean. A dither that is mostly one value with an occasional far one (10, 10, 10, 14) is steady, yet the far reading sat 3 from the mean and started a hold in the next step. Use the middle of the run range instead.
Adds a lopsided-dither test and a test pinning the documented cross-talk limit ([100,300] then [5,90] stays open).
* Judge an untouched cap by the settled run's spread, not a rounded midpoint
Rounding the midpoint to i16 could leave one end of an odd-width run more
than half the tolerance away, so the cap read as touched while sitting at
the end of the run it settled on. Keep the run itself and ask whether one
more reading keeps its spread within tolerance.
* space-mouse: pin the wider skip band of a narrow settled run and a single-reading settle
* space-mouse: keep a step held while a lesser axis stays out as the larger one relaxes
Settling needed one axis furthest and returned, and one axis about as far and settled, which a single relaxed cross-talk axis satisfied alone. Every axis that moved at least half as far as the furthest now has to be back by RETURNED_SHARE of its own peak too.
* 3D mouse: the calibration tests run in a focused window, as the focus check now requires
* 3D mouse: the axis tests run in a focused window, as the focus check now requires
* 3D mouse: the calibrated-rest test runs in a focused window, as the focus check now requires
* Settle a spacenavd reading that is nearly at rest (#3523)
spacenavd sends a record only when the values change and nothing when the cap springs back, so the last small leftover of a release was held forever and the view kept drifting.
A held reading now reads as rest once every axis is within 150 raw units of the device rest and no record has arrived for 0.5 s. A firmer push stays as long as it lasts, and so does a small value next to a large one. 150 is the smallest round number above the worst leftover in the issue (129 on one axis).
* Choose which 3D mouse driver to read on Linux
When both spacenavd and the device itself (hidraw) are there, Preferences now has a "3D mouse driver" choice: Automatic, spacenavd or Direct. Automatic is what it always was, spacenavd when it runs and the device otherwise, so nobody's setup changes. If the chosen one isn't available the other is read and the log says so once. Each driver keeps its own calibration, as before.
To switch without a restart both reads now wake up when idle and look at the choice: spacenavd with a 250 ms read timeout, hidraw opened non-blocking and polled every 5 ms when it has nothing. A record that arrives is read as before, and one split across a timeout is kept. The reader also publishes which of the two exist, which is what shows the combo box; with only one it says which one is in use.
* Build the 3D mouse row header from existing labels
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
Co-authored-by: bherbruck <brennen.herbruck@gmail.com>
* claim: #3654
* Control channel: control.running lists what the app is running, across connections (#3654)
control.pending answers only for the connection that asks, so a second client,
the MCP bridge after a reconnect or a script saw {"pending": null} while the app
was busy. control.running answers for the whole app, the same on every
connection and at once (the connection's own thread answers it, also while it
is busy with a timed-out request): every request the app has been sent and not
answered yet, oldest first, with its connection, id, method, command id (or a
batch's step count), seconds since it arrived and whether it still waits its
turn. A timed-out request stays listed until the app answers it, also after its
connection closed. Parameters are never shown.
The window marks a request started when it takes it (ControlRequest.started),
which tells queued from running. The MCP server gets a read-only `running` tool.
---------
Co-authored-by: MateuszSychXpl <151755914+MateuszSychXpl@users.noreply.github.com>
Save As and Open use the browser's file dialogs where the File System Access API exists; Save writes back to the same file, its handle kept in IndexedDB across reloads. File > Download design added. Other browsers keep storing designs in OPFS.
Save ran on the window's thread: about 130 ms on the Prusa Core One design,
and more than a second once saves keep the evaluated model. The toolbar's
Save and Save As now run the command on a scratch copy of the session on a
worker; the design waits for it as it waits for a file picker (commands are
refused meanwhile), and the outcome is taken up as running the command here
would have left it: the name and links the save gave the design, its file,
its saved state (`Session::adopt_save`), the echo, the error status, the
recent files and the home thumbnail. A tab closing or the app quitting after
Save waits for the write, as does the app's exit. Scripts, the control
channel and the browser build still save in place.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
A PDF goes whole to agents that take embedded documents; to the others, or past 12 MB, the message names it with its type, size and place. Any file that is not a picture, text or CAD file is attached by reference instead of refused. Chips show the type's extension; the file dialog offers All files.
* claim: #1801
* Insert SVG: the file picker opens on SVG files (#1801)
Every Open went through one host picker whose filters start with designs, CAD
files and meshes, so Insert SVG (and Insert DXF) in a sketch opened on STEP,
IGES and meshes, and the SVG file had to be found under another filter.
The host's pick_open now takes the extensions the command reads and offers
them as the picker's first, default filter. Insert SVG and Insert DXF pass
their type; every other Open passes none and keeps the list it had. The
browser host ignores them (it shows its own file panel).
---------
Co-authored-by: MateuszSychXpl <151755914+MateuszSychXpl@users.noreply.github.com>