| |||||||||||||
LCA: The ways of WaylandLCA: The ways of WaylandPosted Feb 14, 2013 9:14 UTC (Thu) by michaeljt (subscriber, #39183)Parent article: LCA: The ways of Wayland
> The root of the trouble, Stone said, was that—thanks to politics and an excessive commitment to maintaining backward compatibility even with ancient toolkits—no one was allowed to touch the core protocol or the X server core, even as the needs of the window system evolved and diverged.
This is a rather common problem, and Wayland seems to me to be a textbook solution to it - keep what works, drop the outdated bits and provide a transparent but optional compatibility layer. Presumably if that layer is slightly less efficient than the original it won't matter much, as improvements in hardware will compensate. I often wonder whether this model could be applied to the Linux kernel with its need to maintain ABI compatibility with all previous versions (which is pretty much useless for any application which has dependencies on anything in user space, notably shared libraries).
(Log in to post comments)
LCA: The ways of Wayland Posted Feb 14, 2013 9:21 UTC (Thu) by khim (subscriber, #9252) [Link] It's just not needed in Linux because it carefully kept it's ABI very narrow. Which means that implementations of old, obsolete and no-longer-useful interfaces comprises only tiny piece of kernel (most of these are transparently translated to new capabilities). Most programs still use open(2) and read(2) even today, you know!
LCA: The ways of Wayland Posted Feb 14, 2013 9:27 UTC (Thu) by michaeljt (subscriber, #39183) [Link]
> Which means that implementations of old, obsolete and no-longer-useful interfaces comprises only tiny piece of kernel (most of these are transparently translated to new capabilities).
So in other words an in-kernel compatibility layer. (And indeed, I was just poking about in the fsnotify/inotify code last week after I and some users hit a bug in Ubuntu's kernel.)
LCA: The ways of Wayland Posted Feb 14, 2013 9:54 UTC (Thu) by daniels (subscriber, #16193) [Link]
Actually, it wasn't 100% because of backwards compatibility; a lot of the problem was that only the X Consortium were allowed to modify the core server or specs, so XFree86 never touched it. And neither did the X Consortium, since they were pretty much dead.
|
|||||||||||||