linux.conf.au 2013 videos online [LWN.net]

linux.conf.au 2013 videos online

Posted Feb 13, 2013 18:43 UTC (Wed) by Wol (guest, #4433)
In reply to: linux.conf.au 2013 videos online by dowdle
Parent article: linux.conf.au 2013 videos online

Basically, the guy who forked Xorg off from XFree (because he was the only guy really doing any development) is himself 100% behind Wayland.

And Wayland has been designed, aiui, so that you can either run X over wayland, OR wayland over X, whatever floats your boat. So that when X does sink (as seems inevitable) Wayland will be a seamless transition.

Yes there's a lot of things X can do that Wayland can't. But Wayland was designed to be able to replace X so if somebody wants to code that stuff, there's nothing seriously impeding them from doing so.

Cheers,
Wol

(Log in to post comments)

linux.conf.au 2013 videos online

Posted Feb 20, 2013 23:49 UTC (Wed) by mmarq (guest, #2332) [Link]

Why not put the X protocol in Wayland itself ?

That way all controversy would disappear

As is, if i'm not wrong, it is a choise between Wayland apps or X apps. The idea of a server on top of another server, seems more clumsy and ugly hack than most things that can be said about xorg "codebase".

If it is a mass "transition", means the most severe fork of anytime, it will take A LOT of years to transition everything to Wayland...

And articles like "next year is the year of desktop Linux" will still keep floating every year never questioning the real why...

So liking or not, better or not, a "fork" to all the immense app base has been created... an "app developer" should target X and maintain what is familiar and most of "compatibility" to most of systems for a lot of years... or should he try Wayland ? ...

whatever it is, he would risk bugs and complains about speed and or jerkiness, never seen before, because of that server on top of server model...

Usually "start all over again" can improve things no doubt, Wayland can induce that no doubt... but there is a huge risk in that, besides screwing all development scheduling...

Comparing to now and concerning mass adoption... this can be the kind of "mistake" that can take a lot of years to recover... if ever... (THINK ABOUT IT).

< sometimes i think of what a friend of mine of Merrill Lynch said... "sometimes the best business of all is not be on any business at all"... wait & see... >

linux.conf.au 2013 videos online

Posted Feb 21, 2013 0:05 UTC (Thu) by neilbrown (subscriber, #359) [Link]

> Why not put the X protocol in Wayland itself ?

Because a big part of the point of this is to get rid of the legacy cruft that is in the X11 protocol.

> As is, if i'm not wrong, it is a choise between Wayland apps or X apps.

You are wrong. You can have an X server which renders to Wayland.

> The idea of a server on top of another server, seems more clumsy and ugly hack than most things that can be said about xorg "codebase".

Allowing multiple processes to work in concert to provide some service is a fine and potentially elegant idea. An X11 server which works with the Wayland server is not importantly different from a window manager which works with the X11 server. In each case, different tasks are taken on by different processes.

Did you know that some apache web servers serve content that is stored in an NFS filesystem? This can be quite effective and again is not inherently different from an Xorg/Wayland combination.

Certainly there is a large risk in developing wayland if the goal is to replace X11 everywhere. However if the goal is to have fun, learn stuff, and create something that might be useful, then the risk isn't all that great.

linux.conf.au 2013 videos online

Posted Feb 21, 2013 3:25 UTC (Thu) by dlang (✭ supporter ✭, #313) [Link]

in addition to the other objections people will raise, having the X protocol in Wayland in addition to the Wayland protocol, this wouldn't eliminate the problem that many of us fear, which is GUI toolkit authors deciding to drop support for X and only support the Wayland protocol (with it's current lack of network transparency)

linux.conf.au 2013 videos online

Posted Feb 21, 2013 4:15 UTC (Thu) by neilbrown (subscriber, #359) [Link]

Lack of network transparency in Wayland in roughly matched by the lack of network transparency in 'ls'. Yet I can run "ls" on my NFS filesystem without trouble.

If you want network transparency, you get to use two protocol converters. One that runs on the same machine as your applications, appears to be a Wayland server, and flings images over the network in some way. The other runs on the same machine as you display, acts as a Wayland client, and hands the images over to be displayed. Similarly input events are passed back.

I can easily imaging 'ssh' growing functionality to start these two protocol converters for you (if so configured) - just as it currently will transparently set up X forwarding over the secure connection, with X-authentication handling. Then you should "ssh myhost may-gui-app" and it would "just work" just like it does now.

So lack of network transparency in Wayland isn't a problem - it is an aspect of the well-structured design.

And fear that developers of some project might drop some functionality that you find valuable certainly has some validity (see also Gnome-3) but tying that fear to the functionality of some other project seem hard to support - peoples reasons for breaking things are usually quite different to the reasons that you expect.