Wayland preview

selsinork wrote:

 

Wayland is stuck with a nasty chicken vs egg problem. As it's such a different beast it effectively breaks all the existing X11 apps

Is this true though?  My understanding (which is entirely derivative and may easily have been swayed by overly optimistic Wayland articles) is that X11 compatibility is provided not just grudgingly but as a primary goal.  Indeed, anything less would be stupid, since practically 100% of Unix graphics applications would be abandoned --- not a recipe for success.

As I read it, all the old stuff will still work, some of it a bit slower because of translation overheads, some of it faster because the translation overheads may be less than the new gains.  The only thing that will disappear for a while (until remoting protocols take over) and is an accepted regression is network transparency.  That will definitely impact some people, but it's in the single digit percentages of X11 users or less.

Morgaine Dinova wrote:

 

Is this true though?  My understanding (which is entirely derivative and may easily have been swayed by overly optimistic Wayland articles) is that X11 compatibility is provided not just grudgingly but as a primary goal.  Indeed, anything less would be stupid, since practically 100% of Unix graphics applications would be abandoned --- not a recipe for success.

and therein lies the problem, if you make it a goal to provide X11 compatibility where's the incentive to move away from X11 ?  With no incentive to move and possibily less performance due to the compatibility layer will wayland ever gain the developer acceptance needed for success ?

and IMHO it also needs wide OS, proprietary GPU vendor, toolkit (Qt & GTK) and desktop environment support.  If it ends up being Linux/Weston only it's going nowhere fast.

> if you make it a goal to provide X11 compatibility where's the incentive to move away from X11 ?

They do apparently intend to provide an X11 compatibililty layer.  But how much time and space overhead

that will involve, remains to be seen, as does whether it will "just work".  With the X11 compatibility layer,

you will apparently be able to slide windows smoothly on your display.  I think Simon's GPU driver for X11 also

saw benefit from using the GPU for such large operations.  How much better Wayland is than what can be done

using Simon's approach is unknown. 

The Wayland folks talk about GTK and Qt applications bypassing X11 and going "direct" or "native".

Since Wayland doesn't have a native rendering API, I'm not sure exactly what that means or how much

benefit it provides in rendering speed.  Presumably it involves using the GPU for rendering somehow,

such as perhaps by using the OpenGL ES interface.   So it isn't clear if the X11 compatibility layer is

intended to be temporary until "direct" or "native" solutions are implemented, or whether it is intended

to be a permanent solution.

coder27 wrote:

 

So it isn't clear if the X11 compatibility layer is

intended to be temporary until "direct" or "native" solutions are implemented, or whether it is intended

to be a permanent solution.

I suspect it's going to be permanent.  A little further digging reveals that supposedly QT5 already has wayland support and GTK will get it later this year. Existing QT and GTK appear to have some support, but it read as if you need to do something to enable it and then run every app with a certain command line switch to change the backend.

Qt may be able to do it's own rendering - it already runs on various embedded things that don't use X11, but the choices appear to be either simple framebuffer (i.e software rendering) or OpenGL.

It's also worth keeping at least one eye half open on Sailfish, a mobile device O/S seeking to compete in Android's space by being more open, more plain Linux, compatible with Android apps yet totally different in UI concept, and faster through using Wayland.

That's a tall order, but ambitious ideas got us to the moon in 1969.  The early prototypes look interesting.