running BM_CHECK_ELEMENT was taking over 75% of the time to subdivide a mesh, since this only reports errors, and is so slow - only run this on non-release builds.
garbage strips for each tile
Promoted the behaviour of combine channels to node operation so that all
nodes use the same implementation. (CombineChannel had a better
implementation)
When the scene is updated Cycles resets the renderer device, cancelling
all existing tasks. The main thread would wait for all running tasks to
finish before continuing. This is ok when tasks can actually cancel in a
timely fashion. For OSL however, this does not work, since the OSL
shader group optimization takes quite a bit of time and can not be
easily be cancelled once running (on my crappy machine in full debug
mode: ~0.12 seconds for simple node trees). This would lead to very
laggy UI behavior and make it difficult to accurately control elements
such as sliders.
This patch removes the wait condition from the device->task_cancel
method. Instead it just sets the do_cancel flag and returns. To avoid
backlog in the task pool of the device it will return early from the
BlenderSession::sync function while the reset is going on (tested in
Session::resetting). Once all existing tasks have finished the do_cancel
flag is finally cleared again (checked in TaskPool::num_decrease).
Care has to be taken to avoid race conditions on the do_cancel flag,
since it can now be modified outside the TaskPool::cancel function
itself. For this purpose the scope of the TaskPool::num_mutex locks has
been extended, in most cases the mutex is now locked by the TaskPool
itself before calling TaskScheduler methods, instead of only locking
inside the num_increase/num_decrease functions themselves. The only
occurrence of a lock outside of the TaskPool methods is in
TaskScheduler::thread_run.
This patch is most useful in combination with the OSL renderer mode, so
it can probably wait until after the 2.64 release. SVM tasks tend to be
cancelled quickly, so the effect is less noticeable.
calling BKE_ptcache_ids_from_object() could recursively call BKE_object_handle_update(), when there is no reason to update object transforms at all. Add option not to do any updates and just return an object list.
patch by Nicholas Bishop, modified to avoid looping over vert-loops one extra time.
added BM_iter_as_arrayN(), returns an iterator as an array without knowing the length before calling.
Without this - there wasnt a good way to ensure addons were aware of new data being loading in new files.
requested by request by liquidape and maccuno.
This way we can do predictable key-shortcut-chaining. Eg.
Shift+A, M, O --- adds a metaball cone.
In editmode
Ctrl+V, X, A --- Assign new vertex group.
* Removed unused lib and include directories (also one bad 'src' include that wouldn't work anyway)
* Copied cmake library finding to APPLE and WIN32 sections. These may need some adjustment for their respective OS.
use an enum typedef for button types. it was quite annoying debugging UI code since the defines are bit-shifted. GDB would show but->type as 13824 and blender define was (27 << 9).
Now but->type shows as a humanly readable names.
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.
instead of checking if the mouse is over another button to exit the popup.
Just check if the mouse is outside the rect-union between the button and the popup.
This fixes [#32399] VSE doesn't show last 3 frames of Quicktime movie.
Some decoders store frames internally until EOF.
So one has to feed the decoding engine with empty packets after EOF
until all frames could be extracted properly.
