Most of the code treats RESOLUTION_FACTOR as "number of clipper units in
a single step" -- a linear distance unit. However, there are a few
locations in which it is treated as unitless, multiplied by area
constants. This commit folds the current value of RESOLUTION_FACTOR into
these other contents in preparation for declaring it to have distance
units and increasing its value.
Prevent crash (CAMTests) in ToolBitRecomputeObserver when the toolbit object has been deleted/missing
by catching ReferenceError before accessing its Document attribute. This ensures slotRecomputedDocument
exits gracefully if the object is no longer valid.
src/Mod/CAM/Path/Tool/toolbit/models/base.py
- Wrapped access to self.toolbit_proxy.obj.Document in try/except to handle ReferenceError
if object is deleted/missing, preventing crash during document recompute.
Introduced FLOAT_EPSILON for robust floating point comparisons, resolving cases
where Z axis retract moves were missing after geometry recompute in 3D Surface
rotational scan operations. This prevents precision errors from causing the
cutter to skip intended lifts between rings or scan lines.
src/Mod/CAM/Path/Op/Surface.py
- Added FLOAT_EPSILON constant
- Updated Z move comparison logic to use tolerance
A bug in the ToolBit model caused an infinite recompute loop and UI freeze when properties such as Diameter, Flutes, or CuttingEdgeHeight were set to expressions. The visual representation update was being triggered during document recompute, which could recursively trigger further recomputes. This fix defers visual updates by queuing them and processing only after the document recompute completes, using a document observer. The observer is cleaned up after use and on object deletion, preventing memory leaks and repeated recompute cycles.
src/Mod/CAM/Path/Tool/toolbit/models/base.py:
- ToolBitRecomputeObserver: Document observer class that triggers queued visual updates after recompute completes via slotRecomputedDocument.
- _queue_visual_update: Queues a visual update to be processed after document recompute.
- _setup_recompute_observer: Registers the document observer for recompute completion.
- _process_queued_visual_update: Processes the queued visual update and cleans up the observer.
- onChanged: Now queues visual updates instead of calling them directly.
- onDelete: Cleans up any pending document observer before object removal.
- Added --rigid-tap argument to linuxcnc_post.py to enable G33.1 rigid tapping cycle output.
- Implemented logic to skip G80, G98, G99 commands with tapping annotation when rigid tapping is enabled.
- Enhanced handling of G84/G74 tapping cycles for rigid tapping, including pitch (K), depth (Z), dwell (G04 P), spindle reversal (M3/M4/M5), and reverse-out moves.
- Updated Tapping.py to annotate G98/G99 and G80 commands with {"operation": "tapping"} for improved post processor handling.
When a geometry element is selected and then hidden using the visibility
checkbox in the Elements panel, it remains in the selection.
Subsequently, if another geometry element is selected and deleted, both
the visible element AND the hidden element are deleted.
Root cause of that is basically that `changeLayer()` functions change
the geometry's visual layer, but never clear the geometry from
`Gui::Selection()`.
So, this patch adds the handling to collect every sub-element name of
the hidden geometry and then formats it and calls remove selection on
the geometry during layer change transaction.
* [ArchStairs] Fix StringerOverlap Base at Angle
Github Issue Discussion
- https://github.com/FreeCAD/FreeCAD/issues/24321#issuecomment-3492459309
"... error when Stringer Overlap is set above certain threshold combined with a non-baseless stair which does not have orthogonal direction - this means that the base line or sketch is rotated to a degree, which is not 0, 90, 180 or 270 degrees."
Further
Fix#24321
* [pre-commit.ci] auto fixes from pre-commit.com hooks
for more information, see https://pre-commit.ci
---------
Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
Fixes#25084.
Instead of a brep file a point list is now used to generate the figure.
Additionally:
Two points have been added to make the head of the figure less 'pointy'.
It would make sense to move the `get_human_figure` function to a different file in the future.
With 1.7m the figure is not very tall BTW.
