Code review comment for lp:~vanvugt/mir/client-side-vsync

> kdub:
>
> (0) Sure you can keep server-side queueing if you want it. So long as this
> code still has the ability to turn it off. Because having it queueing on will
> reintroduce the bufferbloat-nested-queuing-lag problems that are fixed here.
> So you probably don't want that, but up to you :)

Sure, we need an option to turn it off, it might just be useful to have it around.

> (2) Yes, I expect the logic to move around with Cemil's future changes but
> where it is right now works for what we use right now. As for a public API,
> yes I agree that will be wanted; already mentioned it in the description
> above, but also a public API will be needed by Xmir to implement
> https://www.opengl.org/registry/specs/OML/glx_sync_control.txt

Alright, so we might need some extensions for this to work in the "MirRenderSurface" world, and the driver would manage what BufferStream is managing now.