Inside TaskDialogPython::clearForm() the interpreter may execute a lot of things where at the some point the GIL must be hold.
This very likely fixes https://github.com/FreeCAD/FreeCAD/issues/22863
@thyssentishman
Please check this branch
* FEM: Extract postprocessing data only if available. Fixes#22193
* [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>
The current solution checked shapes before allowing boolean operation to
happend. That meant that even small error with model in an unrelated
place could prevent the boolean from being computed, even if it would
compute fine and create valid shape. This commit changes that behaviour
so the tool at least tries to compute the results.
With TNP mitigation implementation handling of TopAbs_FACE was changed
in the method computing 2D offset of shape. It used to be very simple
iteration over face wires and it was instead changed to much more
complicated splitWires method. The intent of that change was to ensure
stability of the result, but it breaks existing models.
The problem is caused by OCC bug within offseting mechanism that is
dependent on wire order. Here the outer wire is moved to the end which
can cause issues.
Another issue is that for some reason, if circle is enclosed by other
shape the offset does not create proper 2D offset but moves the circle
in Z direction.
* TechDraw: create a transaction when finished dragging a view
* TechDraw: drag projection group when dragging a subview in AutoDistribute is enabled
* TechDraw: avoid creating 'Drag View' transaction if the document is already in a transaction
* TechDraw: Apply suggestions from code review
---------
Co-authored-by: Benjamin Nauck <benjamin@nauck.se>
This unifies deletion behavior for all PartDesign features ensuring that
sketches that were used to create these features show after deletion and
that the tip is properly transfered.
* Add epic for fixing user pain points
I didn't add any detail page for this, because the pain points are already described in this document. I just think there should be an epic for this, since we are prioritizing changes that fit within an epic
* Fix typo
* Start: Add caching for performance for thumbnails on start page
So currently we can have a problem where we are trying to load whole
image as a thumbnail, which can result in over 256MB~ of internal buffer
memory. Also, even if we load smaller size - every now and then start
page gets refreshed, so to check if any file got modified and refresh it
in recent files. This is okay, but with large files it loads them over
and over, resulting in start page lagging.
Solution for that is first - load image thumbnails as scaled, small
images instead of full image.
Second - for performance, use caching by using `path:modtime:size` key.
If the item fits this key, it means it didn't change, so just proceed
further and get this item from the cache. If the key is different, it
means it has been changed on the disk, so reload it.
* Start: Deactivate Start page if it loses focus to stop receiving events
As the title says. This prevents Start page from processing unnecessary
events (mostly paint ones that were causing extreme lag previously) if
it is not opened.
* Start: Preserve aspect ratio of the image for the thumbnail
* Start: use brace initialization when returning QString
* [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>
* Revert #22941
* [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>
Simple fix for such a simple error I didn't see. Basically,
`valueChanged` signal can result in EditableDatumLabel being deleted
(for example, when we are on the last OVP and all of the OVPs have been
accepted). This could result in crashes as reported in previous issues,
because just after emitting this signal we were trying to use `this` pointer,
which wasn't available anymore as we deleted EditableDatumLabel obj that
was attached to it, so in essence we were dereferencing some random
address.
Fix for that is simple - move the check for `hasFinishedEditing` flag to
check locked appearance in `validateAndFinish` lambda, as it already
does validation and this way we will avoid referencing internals (which
could be deallocated) after emitting this signal.
