Merge-resolution cleanup. The #12506 re-select path kept main's preset_bundle
null check, and both select_machine calls now use effective_agent_id rather than
mixing it with the equal-but-differently-named agent_info.id.
Resolve five conflicts, all of which needed both sides rather than a pick:
- BackgroundSlicingProcess: ours was a pure tabs->spaces reformat of base, so
keep main's per-filament volume/nozzle map read-back (its only change here).
- GUI_App: main's #12506 else-if attached to an `if` this branch deleted;
re-expressed onto the same-agent early-return path (the agent factory caches
per id, so pointer equality is the same predicate).
- MainFrame: both sides relocated Sync Presets independently; keep main's
push_notification plus the branch's Plugins menu items.
- Tab: the "TODO: Orca: Support hybrid" blocks were unchanged base, not a branch
decision; take main's enabled Hybrid to match the already auto-merged siblings.
- test_config: union of both sides' cases (6 plugin + 9 multi-nozzle).
The filament-group golden harness landed with H2C/A2L support (#14685). Its
"FilamentGroup golden regression" / stress_66 case fails intermittently on
Windows x64, on main and on unrelated PRs alike. The test depends on how fast the
runner is.
The k-medoids clustering these goldens exercise is an anytime search bounded by a
3 second wall clock. Every restart is seeded from its own index, so nothing about
it is random. What varies is how many restarts fit in the budget, and the best
cost is a minimum over completed restarts, so a slower runner is never better.
Grading a score produced that way measures the machine as much as the code.
Add a ClusteringBudget struct and let the tests set it. The defaults are the
current 3 seconds and 30 restarts, so slicing behavior is unchanged. A
non-positive timeout removes the wall clock and bounds the search by restart
count alone.
The goldens are then graded under a fixed budget of four restarts, where every
one of them reaches the BambuStudio reference within 3%, so the score becomes a
property of the code. This retires the machine-specific 125103 lock on stress_66.
The default wall-clock path keeps its own test, asserting the grouping is valid
and the search does not run away. It makes no score assertion, because under a
wall clock that number is not a property of the code.
The golden test also checks the run fits in ten times the default wall clock.
Slicing quality depends on how many restarts fit in the budget, so a search an
order of magnitude slower would degrade real groupings while a fixed-budget score
gate stayed green.
The 3% tolerance stays as the parity allowance against the goldens. It also
covers a small spread across standard libraries: the k-medoids search seeds each
restart with std::shuffle, whose algorithm the C++ standard leaves unspecified,
so libstdc++, libc++ and the MSVC STL permute the same seed differently, start
from different medoids, and settle on slightly different groupings, about 3e-4
apart and only on the goldens heavy enough to reach the k-medoids search.
# Description
Adds a --slice (-s) mode to the profile validator that slices a
two-colour cube through every shipped printer, expanding all custom
g-code (change_filament_gcode, machine start/end, etc.). This catches
invalid-placeholder / bad-flow / slicing errors that the static JSON
checks and unit tests can't see.
Included:
- Validator: new -s sweep mode; per-profile error attribution in the
log; resolves the synthetic 2nd-filament nozzle-mapping so multi-nozzle
BBL printers (incl. the Direct-Drive+Bowden X2D) validate cleanly.
- CI, two complementary paths:
- check_profiles.yml — runs the sweep on profile-only PRs (nightly
binary).
- build_all.yml — new parallel slice_check_linux job runs it on
engine/src PRs with the PR-built binary (build_all doesn't trigger on
resources/**, so no overlap). Runs off the build's artifact, so it
doesn't lengthen the build leg.
- Profile fixes surfaced by the sweep: Creality, FLSun, Ginger, Qidi,
RatRig, iQ.
- Engine: whitelist BBL firmware T-opcodes (T1001/T65279/T65535) in the
time estimator (log-only, no g-code change); dedupe a
per-filament/per-layer log flood in get_config_index.
# Screenshots/Recordings/Graphs
<!--
> Please attach relevant screenshots to showcase the UI changes.
> Please attach images that can help explain the changes.
-->
## Tests
<!--
> Please describe the tests that you have conducted to verify the
changes made in this PR.
-->
<!--
> A guide for users on how to download the artifacts from this PR.
-->
[How to Download Pull Requests Artifacts for
Testing](https://www.orcaslicer.com/wiki/how_to_download_pr_artifacts)
The Unit Tests job sparse-checks-out only .github/scripts/tests, so the
baked-in absolute PROFILES_DIR was missing at runtime; the shipped-profile
test then read a non-existent JSON and null-dereferenced in opt_string.
Check out resources/ in the unit-test job, and guard the test helper to
skip when the profile is absent and require the key before dereferencing.
query_topology: indexing a body face-by-face was quadratic — face_by_index
re-walks the explorer and edge_by_index rebuilds the whole indexed map on every
single call. On a 15.7k-face / 25.6k-edge imported solid this blew past the
MCP 15 s main-thread timeout with the UI frozen throughout. GeometryEngine
gains faces_of()/edges_of(), which enumerate once in the very same order (ids
stay interchangeable with the _by_index accessors, so fillet/up_to_face targets
are unaffected). Measured on that body: 15 s timeout -> 0.46 s.
Feature ops: every commit-time m_doc.recompute() (fillet, cut, shell, boolean,
extrude, ...) now goes through recompute_guarded(), which runs the rebuild on a
worker thread. Live-preview/drag paths stay inline on purpose — yielding inside
a drag would be worse than the stall.
run_off_ui_thread(): the progress dialog is now created only after 300 ms, so a
fast op does not flash a dialog, while input stays blocked (wxWindowDisabler)
for the whole operation since the worker owns the document.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01BVzKmX6Y1aEteit1HTXG4Q
Viewport: GLCanvas3D::on_idle() cleared m_dirty even when
_refresh_if_shown_on_screen() rendered nothing because the canvas was not on
screen yet — a frame requested while the notebook was still showing a page got
silently swallowed and the viewport stayed blank until some later event dirtied
it again. _refresh_if_shown_on_screen() now reports whether it rendered, and
on_idle keeps the canvas dirty when it did not. Covers Design/Prepare/Preview.
Freeze: a big STEP (17.8 MB) spent ~40-50 s inside OCCT on the UI thread
(read_step_solids + recompute), so the window stopped repainting and the
compositor marked the app unresponsive. Both now run on a worker thread behind
an app-modal pulsing progress dialog: the UI keeps painting and the document
cannot be touched while the worker owns it. OCCT's Standard_Failure is not a
std::exception, so the worker catches it explicitly — an escaping exception
would terminate the process.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01BVzKmX6Y1aEteit1HTXG4Q
request_repaint() only invalidates the canvas (Refresh()) on the hardware-GL
path and relies on a wxEVT_PAINT to follow. When the notebook re-shows the
Design page, that invalidation is issued mid-show and dropped: no paint event
arrives, the canvas never renders, and the pane stays blank until another tab
switch forces an expose. Software GL renders directly, so it never showed there.
force_repaint() defers past the show, then Refresh() + Update() for a
synchronous paint. Called from on_tab_shown().
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01BVzKmX6Y1aEteit1HTXG4Q
The new "Prime-tower visits..." test from #14685 (H2C/A2L support) throws
"Coordinate outside allowed range" at random on CI, on both Windows arm64 and
Linux x86_64. It's an uninitialized read of WipeTowerData::height.
#10780 (H2D/H2S) added a second wipe tower path, generate_new(), that fills in
depth, bbx, brim_width and rib_offset but not height. The older generate() sets
height, and clear() never did, so on the generate_new path it stays garbage.
first_layer_wipe_tower_corners() passes height to get_wipe_tower_cone_base() as
R = tan(cone_angle/2) * height. The stray bytes are usually zero, so R is zero
and the slice is fine, which is why it passes most runs on every platform. When
they aren't zero the cone radius runs past ClipperLib's limit and the slice
throws. Nothing selects for it, so it just flakes around.
#14685's test is the first to exercise this path, so that's when it started
showing up.
Initializing height in clear() fixes it, same as the m_origin fix in #13712. The
BBL generate_new path has no stabilization cone, so height = 0 is right.
Keep action identity and display metadata
constructor-set so registry keys cannot drift
from the objects they index.
Require sources to publish unique ownership while
the registry retains shared keepalive for runs.
Preserve opaque IDs and persisted state keys.
Lifecycle callbacks can be registered while
another thread dispatches them, risking races
and iterator invalidation.
Snapshot shared callback objects under a mutex,
then invoke them unlocked to preserve reentrancy
and mutable callback state.
SoftFever