This PR cleans up the Stage Interface code generation and use the
correct interpolation qualifier. This is a cleanup/fix before we can
start converting shaders stage interfaces to be compatible with vulkan.
Ref: #111138 for more information about vulkan compatibility
Pull Request: https://projects.blender.org/blender/blender/pulls/111207
A difference was detected between stage interfaces between OpenGL and Vulkan
that are not compatible with our current API.
**OpenGL**
In OpenGL an stage interface struct can have different interpolation qualifiers
per attribute.
```glsl
struct MyStageInterface {
smooth vec4 color;
flat int face_flag;
};
layout(..) MyStageInterface interp;
```
**Vulkan**
In vulkan the interpolation qualifier isn't supported on attribute
level and needs to be added to the struct.
```glsl
struct MyStageInterface {
vec4 color;
};
struct MyStageInterface_flat {
int face_flag;
};
layout(..) smooth MyStageInterface interp;
layout(..) flat MyStageInterface_flat interp_flat;
```
This patch reports shaders that are incompatible with Vulkan so they can be
patched. Report is only done in debug mode and when using the vulkan backend.
After all shaders are patched an error will be raised so developers will
known immediately when incompatibility are created.
Making the shaders compatible and adding the error will be done in future
patches.
**Python**
Via Python gpu module (gpu.types.GPUShaderCreateInfo) it isn't possible
to construct an incompatible shader as instance names cannot be set
via the API. So this isn't a breaking change.
Pull Request: https://projects.blender.org/blender/blender/pulls/111138
This PR fixes an compilation error where sstream used to be included
on a higher level, but has been removed and we need to include them
where they are actually used.
This wasn't detected as most developers don't have the vulkan backend
active.
Pull Request: https://projects.blender.org/blender/blender/pulls/111203
Regression in [0] which caused `execinfo.h` not to be detected by CMake.
Setting a default variable for other platforms prevented the new
variable from being set.
[0]: f197b1a1f1
The `Find*.cmake` modules originally used uppercase commands to match
CMake's own conventions. Since then CMake uses lower-case and even
within our own find modules, using all uppercase wasn't done
consistently. Opt for lowercase everywhere.
Previously FREETYPE_LIBRARIES would sometimes contain
BROTLI libraries, since some configurations set them separately,
declare them separately under all configurations to avoid
changes to one platform breaking others.
Also quiets uninitialized variable warning.
Allow dragging items from Outliner to other windows. Disables EdgePan
operator when entering new window. Canceling the drop redraws target
area and resets all modal cursors. But this does not fix incorrect
mouse cursor during operations on the remote target window.
Pull Request: https://projects.blender.org/blender/blender/pulls/105196
Updates sphinx and the them to the latest versions
This commit also removes 'babel' from the list of requirements, babel is only important for internationalization.
This commit allows both `rna_enum_` and `rna_node_`to be used for enum prefixes
This also fixes one enum by adding `rna_` to the prefix.
Together these changes fix the API documentation generation.
The default vignette effect introduces issues reported several times
on various hardware configurations.
It would still be nice to figure out the issue and fix it in the
vignette effect code. For now, simply use single-color in the
default theme.
As agreed during the 2023-08-15 User Interface Module meeting.
Fixes#93627Closes#106699
Unfortunately, relying only on the 'embedded' flag of the collection ID
pointer to detect the LayerCollection of a Scene embedded collection is
not enough, since this info may not always be available.
So add an extra explicit 'is_master' to the recursive code processing
all LayerCollections ID pointerrs, and double-check with an assert that
this info is coherent with the Collection's embedded flag, when
possible.
In a few cases (IPO conversion, Proxy conversion, ...), versioning
implies creating or removing IDs, and/or needs access to the whole Main
data-base.
So far this was done ad-hoc by adding some code at the end of
`setup_app_data`.
This commit formalizes this process by adding a BLO call
(`BLO_read_do_version_after_setup`) that will encapsulate all such
complex versioning code.
NOTE: This commit does not address the existing issue that this
versioning code is never performmed when linking new data (outside of
the 'opening a blendfile' context). This topic would require its own
design task.
NOTE: This commit does not fix the few current evil cases of ID creation in
regular versioning code. This will be addressed separately.
Although this commit does modifies slightly some logic in this specific
versioning process, no behavioral changes are expected here.
Pull Request: https://projects.blender.org/blender/blender/pulls/111147
Some functionality of `Layer`s and `LayerGroup`s can be shared. This is why the `TreeNode` class exists. It encapsulates all the common methods both `Layer`s and `LayerGroup`s should have.
Up until now, the usage of `TreeNode` was not consistent and often the layers would just access the base c-struct directly.
This refactor makes it so that all of the functions affecting only the `TreeNode` are moved to the `TreeNode` class and only implemented there.
For example, renaming a layer is really just renaming the node. Therefor there shouldn't even be a `rename_layer` and `rename_layer_group` function. These were combined into a `rename_node` function.
Pull Request: https://projects.blender.org/blender/blender/pulls/111177
Not sure how bad that was in practice, was likely harmless. But
foreach_id logic should never decide to not process an ID pointer
because its value is NULL. Callback code is expected to handle NULL
pointers appropriately.
When old and new names of layers/groups are same, garbage string is
displayed. This is because `oldname` char pointer is referenced to
`node.name`. After function execution, `oldname` memory is freed in
`interface_handler` (see `after.rename_orig`), thus displays garbage
string.
Fix is to copy the content of `oldname` instead of referencing it.
Pull Request: https://projects.blender.org/blender/blender/pulls/111173
Include counts of some headers while making full blender build:
- BLI_color.hh 1771 -> 1718
- BLI_math_color.h 1828 -> 1783
- BLI_math_vector.hh 496 -> 405
- BLI_index_mask.hh 1341 -> 1267
- BLI_task.hh 958 -> 903
- BLI_generic_virtual_array.hh 509 -> 435
- IMB_colormanagement.h 437 -> 130
- GPU_texture.h 806 -> 780
- FN_multi_function.hh 331 -> 257
Note: DNA_node_tree_interface_types.h needs color include only
for the currently unused (but soon to be used) socket_color function.
Future step is to figure out how to include
DNA_node_tree_interface_types.h less.
Pull Request: #111113
The recent change to header copyrights [0] unintentionally changed
"Blender Foundation" to "Blender Authors" for the WIN32 file path
which blender is installed into.
Revert lines changed that aren't related to copyright text.
[0]: e955c94ed3
- Add git_data_canonical_authors module with maps various
"Author <email>" combinations to a single author.
- Add git_data_sha1_override_authors module which
overrides authors for individual commits.
- GitCommit.author now includes the email.
- Move co-author access into GitCommit.co_authors.
Using name & email, then de-duplicating removes 43 duplicates in the
generated credits.
Moving the author & override mapping out of credits_git_gen.py
allows this data to be shared by a utility to update AUTHORS
(not yet committed).
Adapt the `ACTION_OT_select_leftright`, `ACTION_OT_select_box` and `ACTION_OT_select_column` operators to take into account grease pencil layer channels. These operators call `actkeys_select_leftright`, `box_select_elem`, and `markers_selectkeys_between`, respectively. In these three functions, grease pencil layer channels are considered now. This resolves#110522.
Pull Request: https://projects.blender.org/blender/blender/pulls/110939
Including <iostream> or similar headers is quite expensive, since it
also pulls in things like <locale> and so on. In many BLI headers,
iostreams are only used to implement some sort of "debug print",
or an operator<< for ostream.
Change some of the commonly used places to instead include <iosfwd>,
which is the standard way of forward-declaring iostreams related
classes, and move the actual debug-print / operator<< implementations
into .cc files.
This is not done for templated classes though (it would be possible
to provide explicit operator<< instantiations somewhere in the
source file, but that would lead to hard-to-figure-out linker error
whenever someone would add a different template type). There, where
possible, I changed from full <iostream> include to only the needed
<ostream> part.
For Span<T>, I just removed print_as_lines since it's not used by
anything. It could be moved into a .cc file using a similar approach
as above if needed.
Doing full blender build changes include counts this way:
- <iostream> 1986 -> 978
- <sstream> 2880 -> 925
It does not affect the total build time much though, mostly because
towards the end of it there's just several CPU cores finishing
compiling OpenVDB related source files.
Pull Request: https://projects.blender.org/blender/blender/pulls/111046
