Currently each node's position is stored in the coordinate space of
its parent. To find the location of a node on the canvas, we have to
apply the translation of each of its parents. Also, nodes have hidden
"offset" values used while transforming frame nodes. Together,
those made the system much more complicated than necessary,
and they made the Python API ineffective.
This commit removes usage of the offset values and moves nodes
to be stored in the "global" space of the node canvas. It also resolves
some weird behavior when resizing frame nodes, and fixes a few bugs.
The change is forward compatible, so we still write files with nodes in
the old parent-space format. In 5.0 the conversion when writing can be
removed. The existing Python API also stays the same. A new
"location_absolute" property gives node locations in global space,
and changing the old property also moves the child nodes of frames.
Resolves#92458, #72904.
Pull Request: https://projects.blender.org/blender/blender/pulls/131335
In Blender 4.4 (since commit 00a8d006fe), polyline shaders stopped
using geometry shaders and now rely on SSBOs.
In C++, workarounds allow these shaders to function as before, albeit
with some limitations.
To address these known limitations, error messages are now raised when
a polyline shader is used with unsupported configurations.
Pull Request: https://projects.blender.org/blender/blender/pulls/131055
In Blender 4.4 (since commit 00a8d006fe), polyline shaders stopped
using geometry shaders and now rely on SSBOs.
In C++, workarounds allow these shaders to function as before, albeit
with some limitations.
However, this change broke the `batch_for_shader` function in Python,
as `GPUShader.attrs_info_get()` only reads attributes and does not
support SSBOs.
To address this, the method now treats polyline shaders differently,
accessing SSBO inputs instead of attributes.
fix
This was caused by a bad state where a texture (likely the viewport)
texture, could be bound as texture by the viewport blit function
and kept bound as texture input during the subsequent redraw.
This would provoke a feedback loop which is UB.
It is still unknown why this is not catched by our GL layer
and why it isn't an issue in Vulkan or Metal nor other GL
implementations.
Add back all missing calls from previous `DRW_state_reset` to
avoid more regressions.
In the File Browser the "Cancel" button always has a tooltip hint of
"Cancel loading of selected file", even when you are saving. This PR
changes that to a more generic "Cancel file operation". Similarly the
save/load button always shows "Execute selected file", even when
loading. This PR instead uses the description of the operator that is
called when that button is pressed. When opening a file it shows "Open
a Blender file", saving shows "Save the current file in the desired
location", etc.
Pull Request: https://projects.blender.org/blender/blender/pulls/131703
This enables moving elements form one vector to another.
Usually this is being doing by extending a vector with the content
from the secondary vector and then clearing the secondary vector.
However sometimes this being performed to transfer ownership of managed elements,
if elements are copied from the secondary vector, but not cleared, this could
lead to 2 vectors to share ownership of objects.
Pull Request: https://projects.blender.org/blender/blender/pulls/131560
Mistake in f6a4d01703. This refactored the lookup for a named string context
member to avoid code duplication.
Intention was to construct a StringRef referencing the string stored in
context. But the refactored lookup returned an unintentional string copy, so
the reference would be constructed from a temporary string object that would
get destructed immediately.
Renaming attributes in the property editor didn't take into account
the required domains and types for builtin attributes. It might be
possible to trigger crashes by using a builtin name with the wrong
metadata. At the very least it would be confusing.
This commit extends the attribute provider check for builtin
attributes to give the required domain and type. We check that in
the attribute rename function to make sure the new name is valid.
Pull Request: https://projects.blender.org/blender/blender/pulls/131702
When releasing the pen the pressure would always go back to zero,
undoing the effect of the brush. To fix this, don't take pressure into
account for the strength, and don't display it in the UI. The grab brush
in mesh sculpt mode works the same way.
Pull Request: https://projects.blender.org/blender/blender/pulls/131744
The default wasn't in the list of static enum items which means the
value was reset to 0 instead of the proper default. Even though the
list is dynamically generated, the possible enum items should be
part of the static list anyway.
The property actually had a default, but defined via
`RNA_def_property_float_default` [**RNA** defaults handle highlevel (UI)
stuff like "Reset to Default", but is not in effect when creating new
data/editors].
For this to work, we now define the default in **DNA**
(Also corrects the mismatch in value in this particular case)
NOTE: other editors also handle "defaults" in `SpaceLink` creation functions
in their respective space_XXX.cc files, so that is another system to look at.
NOTE: there might be many more cases that could benefit from moving the
default from RNA (or `SpaceLink` creation functions) to DNA, or possibly even
doing this via other alternatives like supporting C++ in-class member initializers
in makesdna, goal is the same: having a single "source of truth".
But going over these will be for another time...
Pull Request: https://projects.blender.org/blender/blender/pulls/131735
Pass the session UID to the identity comparison callback directly,
instead of passing the ID pointer and accessing the session UID inside
the function. This way the dangling ID pointer is avoided.
Makes use of the support for passing named integers via context, which
was added in the previous commit.
Pull Request: https://projects.blender.org/blender/blender/pulls/131676
Similar to 2fbf206491, but for integers (not strings). Essentially this
allows passing named integer values through UI abstractions in a clean
way.
I used 64 bit integers here, since space isn't an issue here, and it
fits common integer types (so we don't have to add APIs for multiple
integer types).
Planned to be used to fix#111463 (following commit).
The Keying node leaks memory when only the edges output is used. That's
because tweaked matte is computed in that case, but is never used nor
released because the other two outputs are not needed. To fix this,
release the tweaked matte if it is not used.
The GPU and new CPU compositors leak memory when an output is connected
to multiple inputs inside the same pixel operation. That's because nodes
do not know that their multiple outgoing links are in fact going to the
same operation, so their initial reference count is more than the actual
reference count.
To fix this, we keep track of the reference count of pixel operation
inputs and release the inputs based on that reference count.
Currently the realize instances code (which is also used by the join
geometry node) always creates generic attributes instead of vertex
groups. The expectation was that vertex groups would be replaced
by some form of generic attribute sooner rather than later. However,
it's clear that won't happen for some time, and this issue causes users
a lot of trouble. This commit preserves the vertex groups through the
operation. Any attribute name that was a vertex group on any of the
input meshes will become a vertex group in the result.
In the code this is a simple change because the attribute writer
abstraction allows writing to vertex groups as if the were just like
other contiguous arrays. In the future we could optimize the code
specifically for vertex groups.
This resolves the "bug" part of #99197 where the nodes remove vertex
groups. However, this doesn't change the fact that generating
primitive meshes in geometry nodes won't create vertex groups.
In general the property editor settings on the original mesh have
no effect on meshes created from scratch in geometry nodes.
Pull Request: https://projects.blender.org/blender/blender/pulls/131692
`.release()` is technically fine but semantically wrong here. First,
release is meant to take ownership from the unique_ptr, meaning
at best it's "weird" to not do anything with the pointer afterwards.
Second, the runtime struct with the unique_ptr was just default
constructed, which means the unique_ptr will be empty anyway.
When pressing Tab to move to the next input in the same block (and
Alt-Tab to move backward), we currently skip buttons that have the
UI_BUT_DISABLED flag. This PR makes these operations also skip buttons
that have UI_HIDDEN. This makes tabbing work better in places like the
color picker where we show/hide the RGB and HSV buttons depending on
mode.
Pull Request: https://projects.blender.org/blender/blender/pulls/131698
The tooltip used to say that the center is always the objects origin.
However, in the case of text-boxes, it's actually the center of the text box.
There might be valid use-cases to move the text box center to the
object's center, but it seems like the same can be achieved with offsets.
Also that would be a new feature that's out of scope of a bug report.
Pull Request: https://projects.blender.org/blender/blender/pulls/131693
