Pedro J. Estébanez
Pedro J. Estébanez
Precisely these days I'm trying to hunt down the root trigger of certain operations and the ability to obtain a proper C++ stack trace would come super handy. Regarding the...
> @RandomShaper Yes, just looking at the MacOS implementation, there is definitely a lot of overlap in functionality that could be moved into a separate library to save some duplication....
@matheusmdx's and @Noiros' crashes are due to D3D12 not being able to initialize. The way the situation is reported is cryptic indeed. We need to improve that. You may get...
Regarding the shader compilation crash itself, I've been able to reproduce it and I'm working on a fix.
The mobile shader turned out to trigger the usage of a SPIR-V opcode that was left as a _to do_ in the patch for SPIR-V-to-DXIL dependency. https://github.com/godotengine/godot-nir-static/pull/14 should do the...
A few thoughts: - CoW involves dealing with the atomic refcount, with performance implications. - This looks like an _ad hoc_ fix for an issue due to how read-only arrays...
It may be better to fuse the integrity flags with the mode flags to preserve compatibility.
As much as I'm a fan of trying to overload the current enums, that looks too awkward: `swap + sync = none + swap`
It'd be great that the refactor and the fixes were in separate commits. Do you think you could split the changes like that?
Thanks. I will need more time to review this properly, but so far I can acknowledge this fixes the race condition were an object being released by another thread is...