To me there should be a very strong compelling reason when we're adding a new library, yet alone new technology. The big missing point in the comparison list is the familiarity of other developers…
should I create a separate PR or add the tests to this one?
I don't have strong opinion. It you think you'd need extra time for curves tests, then landing what you already have I good. And…
@mont29 it is not specific to OSX builds, there are areas in Blender that use OpenMP regardless of the platform. Also, external libraries might use OpenMP. So I am not sure how practical it could…
I am concerned about a situation where hiprt is compiled with one version and a user building blender from source compiles the kernels with a different version of the SDK
I see. Does the…
@ZedDB Yep. looks good on the code side. Just ensure the tests are fine, and land it!
@jorn Thanks for the PR!
For the macOS I am not sure i fully understand. We ship the same exact libraries versions on Linux and macOS. So there shouldn't be any discrepancy between platforms.
…
Looks reasonable, but I didn't do accurate line-by-line verification.
I think this approach is better than having a more unpredictable behavior when an addon appears to be working fine, until some operation on Blender side is performed.
I'd point out somehow nicely "please unrotate" in release notes.
That's probably the least of the messy approaches. I'd also suggest adding this to the commit message.
Have you browsed…
I can not reproduce the issue with the space mouse (3Dconnexion Space Navigator 3D Mouse) i have available here. Ubuntu 24.04, spacenavd from the repos.
The code is almost. The only thing is that seems we can get rid of anim_index_entry::_pad;
Seems we never read or write the entire struct, and always do it on the per-field basis? If so, we can remove the _pad from the struct.