What's wrong with wayland?
daniel at fooishbar.org
Fri Apr 22 03:33:48 PDT 2011
On Thu, Apr 21, 2011 at 09:15:24PM +0200, Enrico Weigelt wrote:
> * Sam Spilsbury <smspillaz at gmail.com> schrieb:
> > since the majority of clients will do all the surface painting Client-side
> > (cairo, Qt) or direct to the hardware (OpenGL).
> That's the main problem, IMHO. It leads to really a lot of unnecessary
> code and data duplication. (btw: I'm glad to have all my systems free
> of the Qt bloat ;-o).
> Alternative: let the clients just _describe_ their window contents
> as vector graphics, that only get manipulated when needed (and NOT
> repainted over and over by the client itself!). Most of the common
> widget toolkit's code (and runtime data) would be simply obsolete
> this way.
Which leads to massive and unpredictable latency since you've now thrown
a full vector rasteriser into your display server, so any window
management, cursor movement, etc, operations will be blocked behind
rendering which may run for an unbounded amount of time.
IOW, exactly what Wayland was (quite rightly) designed to avoid.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Size: 198 bytes
Desc: Digital signature
More information about the wayland-devel