Christian Hergert

This is obviously a 4 year old issue, and probably not of much use anymore to you, but in case others come by and want to see an example of...

Okay, fixed a coordinate issue with the rectangle which I was finally able to see now that I added a second commit, which passes damage regions across from the producer...

> would we be able to use memory_order_acqure / memory_order release for the first and second atomic_thread_fence calls instead of memory_order_seq_cst for both? I too don't have much experience with...

> Note that I will need to submit the change to our internal Perforce depot before merging on GitHub And no worries on this, I know the drill with regards...

Turns out `_Atomic` is not what we want actually causes issues (at least with GTK 4) so reverted back to `volatile`.

> Could you elaborate? volatile generally isn't meant for inter-thread synchronization purposes from what I've read. I don't believe we're actually relying on `volatile` here either given how it's using...

> I was asking what issues _Atomic causes and how you determined it isn't what we want here. I can't answer that for you, I've never used these features from...

Honestly, it doesn't matter too much to me anymore because the damage thread is already broken by design. The driver constantly locks up on Wayland and so while adding a...

> I was asking what issues _Atomic causes and how you determined it isn't what we want here. From what I've read from https://en.cppreference.com/w/c/language/atomic, it would look like using _Atomic...

If there is interest in automatically handling concatenated files, that would certainly be nice. GTK does this, which is what I would want to use this for when working on...