A correct fix for this bug likely involves changuing how operators are called in the event loop but such changes better not be made just before the release.
Undo would leave BMEditMesh->me pointer NULL, this would likely crash EDBM_verts_mirror_cache_begin() too.
Rather then restore 'me', remove the pointer altogether and use BMEditMesh->ob->data to save us having to keep track of 2 pointers.
This commit adds some first building blocks for the two operators to work modally based on mouse input. To make their function easier, two hotkeys are introduced, Ctrl+B for bevel and I for inset.
TODO:
After discussion with Campbell, we would like to add scale-style line indicators for the operators. This is already done for transform operators but a new interface for mesh operations may have to be written using pieces from that code since, strictly speaking bevel and inset are not exactly "transform" operators.
Also, a better input method for inset is needed and more options exposed. The method implemented right now uses mouse move for thickness and ctrl-mouse move for depth. These are calculated using the distance of the selection center in screen space and the mouse position. While that may work and prevents abrupt changes in values when switching from thickness tweak mode to depth tweak mode, it limits the magnitude of values that can be put into the tool especially in small or large scale.
Alternatives until a better method is written include:
* use relative offset (works but may give strange results)
* tweak manually after the operation.
This call is still needed when loading editmesh because some operations
are not creating CD layer for new loops (like crating new face) and
to prevent loosing all sculpted data when running disps correct
displacement maps should be allocated on loading edit mesh.
- dont do name based object lookup to find the object for a mesh undo state (possibly object is renamed inbetween undos which would crash)
- remove some todo comments for things that are working as they should.
- remove recently added BM_mesh_select_flush_strip(), functions purpose wasn't clear.
- add BM_mesh_elem_hflag_disable_test(), BM_mesh_elem_hflag_enable_test()
to match existing BM_mesh_elem_hflag_enable/disable_all(), these take a hflag to test before editing each element.
This replaces the need for BM_mesh_select_flush_strip().
from Francisco De La Cruz (xercesblue)
The test to use faces was incorrect since mixed vertex and edge mode would be treated as faces.
made own minor change.
problem was that BMesh had tessellation call when undo pushes were called.
if python called an operator with no undo push, tessfaces would not be created.
fix this by making it the responsibility of each editmesh operator to re-tessellate, as it is with notifiers and depsgraph.
added EDBM_update_generic() function to add notifier, tag for depsgraph update and optionally re-tessellate.
