The thing i'm really starting to hate is the requirement to specify both
operation code and node type. Seems to be duplicated enums without real
need for that.
Was using some threaded queue on top of task pool, tssk...
Now using properly task pool directly to crunch chunks of smooth fans.
No noticable changes in speed.
Tried to completely get rid of the 'no threading with few loops' code,
but even just creating/freeing the task pool, without actually pushing
any task, is enough to make code 50% slower in worst case scenario (i.e.
few thousands of simple cube objects).
The root of the issue was in custom normal code, so far it assumed that
we could only have one cyclic smooth fan around each vertex, which is...
blatantly wrong (again, the two cones sharing same vertex tip e.g.).
This required a rather deep change in how smooth fans/clnor spaces are processed,
took me some time to find a 'good' solution.
Note that new code is slightly slower than previous one (maybe about 5%),
not much to be done here, am afraid.
Tested against all older report files I could find, seems OK.
This is to be used from the Outliner, when dragging and dropping
collections from the Active Render Layer
It also includes a cleanup on the outliner so it calls the new
functions. Note: the outliner still needs fix to allow all the
functionality here exposed.
But this will be tackled by Julian Eisel later.
This feature was adding extra complexity to task scheduling
which required yet extra variables to be worried about to be
modified in atomic manner, which resulted in following issues:
- More complex code to maintain, which increases risks of
something going wrong when we modify the code.
- Extra barriers and/or locks during task scheduling, which
causes extra threading overhead.
- Unable to use some other implementation (such as TBB) even for
the comparison tests.
Notes about other changes.
There are two places where we really had to use that limit.
One of them is the single threaded dependency graph. This will
now construct a single-threaded scheduler at evaluation time.
This shouldn't be a problem because it only happens when using
debugging command line arguments and the code simply don't
run in regular Blender operation.
The code seems a bit duplicated here across old and new
depsgraph, but think it's OK since the old depsgraph is already
gone in 2.8 branch and i don't see where else we might want
to use such a single-threaded scheduler.
When/if we'll want to do so, we can move it to a centralized
single-threaded scheduler in threads.c.
OpenGL render was a bit more tricky to port, but basically we
are using conditional variables to wait background thread to
do all the job.
Can't say enough how much I hate those proxies... their duality (sharing
some aspects of both direct *and* indirect users) is a nightmare to handle. :(
It wasn't using old immediate mode, but was using
- client vertex arrays (obsolete)
- quads (obsolete)
- state attrib stack (obsolete)
- polygon mode (still allowed, but gross)
Nothing happen yet when it's supposed to insert the collection into
another one, that part will be handled by @dfelinto.
See gif for demo of how it works UI wise: {F500337}
Also fixed off-by-one error in utility function.
There are now only referenced in:
* drawobject.c
* particle_edit.c
* space_image.c (a single case to be handled on workspace branch)
* rigidbody_constraint.c (to be handled in the following commit)
Not really happy of per-pool threads limit, need to find better
approach to that. But at least it's possible to get rid of half
of the nastyness here by removing getter which was only used in
an assert statement.
That piece of code was already well-tested and this code becomes
obsolete in the new depsgraph and does no longer exists in blender
2.8 branch.
There were some issues with how we store outliner tree elements:
Apparently the only removable elements have been data-blocks so far.
When recreating the TreeElements, their TreeStoreElem instances were
mainly identified by their ID pointer. However non-data-blocks mostly
depend on an index. For collections, such an index isn't a reliable
measure though if we want to allow removing items. Depending on it for
identifying the TreeStoreElem instance would cause some quite noticeable
glitches (wrong highlights, two elements sharing highlight, etc).
For now I've solved that by actually removing the TreeStoreElem that
represents the removed element. A little limitation of this is that
after undoing the removal, some information might get lost, like
flags to store selection, or opened/closed state.
A better solution that would also fix this issue would be having a real
unique identifier for each non-data-block element, like an idname or even
its data-pointer. Not sure if we can get those to work reliable with
file read/write though, would have to investigate...
Also added a general Outliner tree traversal utility.
