Python wheels from extensions were not being removed after
install/uninstall in some cases - although installing an extension
afterwards that used wheels would recalculate deps & remove them.
- Installing an extension didn't include the extension in the
compatibility-cache, causing uninstalling not to remove deps.
- Uninstalling an extension wasn't re-calculating the deps,
leaving them as-is.
Always write the compatibility-cache after installing and uninstalling
so changes are detected & handled.
This PR fixes a latent issue arising from invalid use of `accept_any_intersection(true)` when performing SSS ray-stepping with MetalRT. The comment incorrectly states that "we can optimize and accept the first hit", but to guarantee correct behaviour in future we need to request the closest hit.
Python scripts could perform actions that created notifiers
which would not be handled until the script was complete.
In the case of adding & removing objects a notifier would be created
for adding the object, then cleared when the ID was removed.
This lead to the notifier queue filling up with cleared notifiers
which were included in the search whenever an ID was removed.
The result of this was that adding and removing objects from a script
would become increasingly slower & use more memory.
Resolve by storing the current notifier being handed which isn't freed
(only cleared). The notifier handling loop detects cleared notifiers
and frees them after use.
Remove a workaround for #23871 which manipulated the module
to prevent classes defined in the text editor from having their
name-space cleared.
This caused the "multiprocessing" module to store the "__main__" module
as "__mp_main__" for later use.
Accessing attributes from this module would then attempt to read from
with a null "mp_dict" which crashed. This happened when showing the
extensions preferences but would have occurred if "__mp_main__" was
accessed from elsewhere too.
Resolve by removing the workaround since it has not been needed
since Python 3.2.
When the compositor is canceled mid-execution, the image node will get
corrupt in future executions in new CPU execution mode. That's because
the image node treated cached images as an intermediate results that can
be passed through, while they should be treated as external resources
since they are managed by the static cache manager.
This patch fixes that by adding another variant of wrap_external that
takes a Result and wraps its data, then we use that in the Image node.
Draine phase function sampling internally use Henyey-Greenstein and
Rayleigh sampling for degenerated cases, but the sampling pattern was
different between Draine and Rayleigh. The commit effectively replace
`rand` with `1 - rand` in Rayleigh sampling.
Pull Request: https://projects.blender.org/blender/blender/pulls/129261
When `g == 0`, the Draine phase function from
https://doi.org/10.1051/0004-6361/202142437 simplifies to
\[\Phi(\theta)=\frac{3}{4\pi(3+\alpha)}(1+\alpha\cos^2\theta).\]
Similar as Rayleigh sampling in https://doi.org/10.1364/JOSAA.28.002436,
The solution to the CDF of the marginal density function is
\[\cos^3\theta+a\cos\theta+b=0,\]
with
\[a=\frac{3}{\alpha},\quad b=\frac{3+\alpha}{\alpha}(2\xi_1-1),\]
which has only one real root since \(\alpha > 0\),
resulting in the sample technique
\[\cos\theta=u-\frac{1}{\alpha u}.\]
Pull Request: https://projects.blender.org/blender/blender/pulls/129259
Meta-data are missing on Cryptomatte layers in the GPU compositor, so
they do not get saved using the File Output node. This is due to a use
after free error where a temporary string is used in the meta-data
population logic. This is fixed by assigning the string to a temporary
variable instead.
Thanks to Jorn Visser for finding the cause of the issue.
