The general idea is to store an array of (type, data) pointers of all
PointerRNA ancestors of the current one.
This will help solving cases in our code where the owner (or sometimes
even the owner of the owner) of a random PointerRNA needs to be
accessed. Current solution mainly relies on linear search from the owner
ID, which is sub-optimal at best, and may not even be possible in case a
same data is shared between different owners.
This lead to refactoring quite a bit of existing PointerRNA creation code.
At a high level (i.e. expected usages outside of RNA internals):
* Add `RNA_pointer_create_with_parent` and
`RNA_pointer_create_id_subdata` to create RNA pointers with
ancestors info.
* `RNA_id_pointer_create` and `RNA_main_pointer_create` remain
unchanged, as they should never have ancestors currently.
* Add `RNA_pointer_create_from_ancestor` to re-create a RNA pointer
from the nth ancestor of another PointerRNA.
* Add basic python API to access this new ancestors data.
* Update internal RNA/bpy code to handle ancestors generation in most
common generic cases.
- The most verbose change here is for collection code, as the owner of the
collection property is now passed around, to allow collection items to get
a valid ancestors chain.
Internally:
* `PointerRNA` now has an array of `AncestorPointerRNA` data to store
the ancestors.
* `PointerRNA` now has constructors that take care of setting its data for
most usual cases, including handling of the ancestor array data.
* Pointer type refining has been fully factorized into a small utils,
`rna_pointer_refine`, that is now used from all code doing that operation.
* `rna_pointer_inherit_refine` has been replaced by
`rna_pointer_create_with_ancestors` as the core function taking care of
creating pointers with valid ancestors info.
- Its usage outside of `rna_access` has been essentially reduced to custom
collection lookup callbacks.
Implements #122431.
--------------
Some notes:
* The goal of this commit is _not_ to fully cover all cases creating
PointerRNA that should also store the ancestors' chain info. It only
tackles the most generic code paths (in bpyrna and RNA itself mainly).
The remaining 'missing cases' can be tackle later, as needs be.
* Performances seem to be only marginally affected currently.
* Currently `AncestorPointerRNA` only stores PointerRNA-like data.
This will help `StructPathFunc` callbacks to more efficiently generate
an RNA paths when calling e.g. `RNA_path_from_ID_to_property`, but will
not be enough info to build these paths without these callbacks. And some
cases may still remain fuzzy. We'd have to add thinks like a `PropertyRNA`
pointer, and for RNA collection ones, an index and string identifier, to store
a complete unambiguous 'RNA path' info. This is probably not needed, nor
worth the extra processing and memory footprint, for now.
Pull Request: https://projects.blender.org/blender/blender/pulls/122427
Some versions of PySide give errors when imported together with
bpy.app.translations, because the latter uses a tp_name that does not
exist. The only other tp_name we have with a dot in it (bgl.Buffer)
does exist.
Pull Request: https://projects.blender.org/blender/blender/pulls/134033
Caused by 59dc67974a
The behavior of `RNA_path_from_ID_to_property` /
`RNA_path_from_ID_to_property_index` is that it only gets the "prefix"
of the struct -- the one we are looking for, the one that is missing
here -- if we feed it the `newptr` since it calls
`RNA_path_from_ID_to_struct(ptr)` to get that prefix.
If we feed it the **ID** pointer it will be empty (makes sense it would
basically be "path from ID to ID"...we need it to be "path from ID to
constraint").
Pull Request: https://projects.blender.org/blender/blender/pulls/133977
This makes the code simpler, aligns it with the builtin translations in
messages.cc, and should improve performance as well. My motivation
for the change is to allow returning StringRef/StringRefNull from the
translation functions instead of char pointers.
To test the change I used the Sun Position addon which includes
translations and toggled different languages in the preferences.
Pull Request: https://projects.blender.org/blender/blender/pulls/133827
Previously, the global storage of these types either used a GHash or a
blender::Map. VectorSet is preferrable to GHash because it's type safe,
clearer, and faster. It's preferrable to Map because the key doesn't
have to be duplicated and because iteration is faster.
This PR moves these registered types to VectorSet, just like the
node, node socket, and node tree types. Note that none of these types
use RAII for allocation, so freeing is still done manually.
Testing was manually interacting with each of these systems, including
with addons that register their own types.
Pull Request: https://projects.blender.org/blender/blender/pulls/133778
Similar to `bpy.data.user_map`, it returns a mapping of IDs to all the
filepaths they use.
Fairly basic still, may need to be refined with more options to
control which filepaths are included etc.
Mainly intended to make handling of a production blendfile dependencies
more easy.
Also adds some basic testing of the new feature.
Pull Request: https://projects.blender.org/blender/blender/pulls/127252
When calling bpy.utils.expose_bundled_modules(), these modules are
added to sys.path.
This provides a solution/workaround to two problems:
* Using bpy together with packages like usd-core is problematic. Besides
crashing due to C++ symbol conflicts, it's just impossible to import
different versions of the same module, or to have distinct environment
variables for both. (#127132)
* Blender add-ons using these VFX modules do not currently work with
the bpy module.
This adds about 15MB to the bpy package.
Pull Request: https://projects.blender.org/blender/blender/pulls/133082
Support differentiating between portable & system installations,
useful to properly locate relative paths which would not work
on system installations.
Ref !133143
Clarify the meaning `bpy.app.version_file`, by being more explicit in the
description of `bpy.app.version` and their differences.
### Old:
`bpy.app.version`
: The Blender version as a tuple of 3 numbers. eg. (2, 83, 1)
`bpy.app.version_file`
: The Blender version, as a tuple, last used to save a .blend file, compatible with ``bpy.data.version``. This value should be used for handling compatibility changes between Blender versions
### New:
`bpy.app.version`
: The Blender version as a tuple of 3 numbers (major, minor, micro). eg. (4, 3, 1)
`bpy.app.version_file`
: The Blender File version, as a tuple of 3 numbers (major, minor, file sub-version), that will be used to save a .blend file. The last item in this tuple indicates the file sub-version, which is different from the release micro version (the last item of the `bpy.app.version` tuple). The file sub-version can be incremented multiple times while a Blender version is under development. This value is, and should be, used for handling compatibility changes between Blender versions
Pull Request: https://projects.blender.org/blender/blender/pulls/133142
* In case a subset of IDs is given, it was not checked that all items
are ID data.
* In case of error while processing the subset, there could be some
python data 'leak' (the `user_map` would not be unreferenced).
* Improve performances in case there are many duplicates in given subset
iterable of IDs, by not processing the duplicates.
Discovered in !127252 review.
When using clangd or running clang-tidy on headers there are
currently many errors. These are noisy in IDEs, make auto fixes
impossible, and break features like code completion, refactoring
and navigation.
This makes source/blender headers work by themselves, which is
generally the goal anyway. But #includes and forward declarations
were often incomplete.
* Add #includes and forward declarations
* Add IWYU pragma: export in a few places
* Remove some unused #includes (but there are many more)
* Tweak ShaderCreateInfo macros to work better with clangd
Some types of headers still have errors, these could be fixed or
worked around with more investigation. Mostly preprocessor
template headers like NOD_static_types.h.
Note that that disabling WITH_UNITY_BUILD is required for clangd to
work properly, otherwise compile_commands.json does not contain
the information for the relevant source files.
For more details see the developer docs:
https://developer.blender.org/docs/handbook/tooling/clangd/
Pull Request: https://projects.blender.org/blender/blender/pulls/132608
Rendering a scene with py-drivers would crash after dropping an
extension onto Blender's window then canceling before the update
completed.
This was caused by context.temp_override pushing arguments into the
context but failing to pop them in the case of an error - which would
happen when the popup was closed while the update ran.
ContextTempOverride objects were not freeing the reference to their
arguments when the context managers __exit__ didn't run.
This would happen if __enter__ failed with an unsupported configuration
or if object was created and not used.
Resolve by making ContextTempOverride a container type with GC support.
This also allows the ContextTempOverride to be reused since it's
arguments are now cleared when it's deallocated instead of __exit__.
This caused build errors on the docs builder, I can't seem to reproduce
locally, so revert for now and have another look at some point in the
future.
Sadly as these changes usually go, this took 5c515e26bb and
2f0fc7fc9f with it as well.
Pull Request: https://projects.blender.org/blender/blender/pulls/132559
1) Typing fixes for vector properties (similar to !131855)
2) More specific type for PointerProperty.poll since it's only called if type is ID.
3) Added a note to PointerProperty.poll that it's only checked when property is assigned from the UI.
Pull Request: https://projects.blender.org/blender/blender/pulls/131925
Give the `IDWALK_CB_…` enum an explicit name:
`LibraryForeachIDCallbackFlag`. This way the flags are type-safe, and
it's known where values come from. This is much preferred (at least by
me) to just having `int flags`.
Uses of `0` have been replaced with `IDWALK_CB_NOP` as that has the same
value and is of the right type.
One invalid use of `IDWALK_NOP` was detected by this change, and is
replaced by `IDWALK_CB_NOP`.
This change might be incomplete; I gave the enum a name, fixed the
compiler errors, and then also updated assignments like `int cb_flag =
cb_data->cb_flag`. I might have missed some assignments to `int` though.
No functional changes.
Pull Request: https://projects.blender.org/blender/blender/pulls/131865
Give the `IDWALK_…` enum an explicit name: `LibraryForeachIDFlag`. This way
the flags are type-safe, and it's known where values come from. This is
much preferred (at least by me) to just having `int flags`.
Uses of `0` have been replaced with `IDWALK_NOP` as that has the same value
and is of the right type.
One invalid use of `IDWALK_CB_NOP` was detected by this change, and is
replaced by `IDWALK_NOP`.
This change might be incomplete; I gave the enum a name, and then fixed
the compiler errors.
No functional changes.
Pull Request: https://projects.blender.org/blender/blender/pulls/131865
Two commits that basically do the same thing for two `enum`s: give
them a name.
- the `IDWALK_…` enum → `LibraryForeachIDFlag`.
- the `IDWALK_CB_…` enum → `LibraryForeachIDCallbackFlag`.
This way the flags are type-safe, and it's known where values come
from. This is much preferred (at least by me) to just having `int
flags`.
Uses of `0` have been replaced with `IDWALK_NOP` and `IDWALK_CB_NOP`,
as those have the same value and are of the right type.
One invalid use of `IDWALK_NOP` was detected by this change, and is
replaced by `IDWALK_CB_NOP`. And another one in the opposite
direction.
This change might be incomplete; I gave the enum a name, fixed the
compiler errors, and then also updated assignments like `int cb_flag =
cb_data->cb_flag`. I might have missed some assignments to `int`
though.
No functional changes.
----------
I intend to land this PR as its two separate commits. I just put them in the same PR so the buildbot can handle them in one go, and we don't have a stack of highly relatled PRs.
In the future this could also apply to the `IDWALK_RET_…` enum. This one I left out, though, because a proper cleanup there would also have to include their ambiguity on whether they are bitflags (like the enums in this PR) or not. Their values and the code in `BKE_lib_query_foreachid_process()` implies they are bitflags, but in practice they are never or'ed together and just used as discrete values.
Pull Request: https://projects.blender.org/blender/blender/pulls/131803
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).
Use StringRefNull for all function arguments and return types.
Not a StringRef but StringRefNull since there is still large
interaction with C api so null-termination usually necessary.
If string is expected to be not only empty but also a null then
optional is used. This change depends on #130935.
Pull Request: https://projects.blender.org/blender/blender/pulls/131204
