The real story behind Wayland and X

Daniel Stone explains the history, architectural limitations, and performance bottlenecks of the X Window System and why Wayland was created.

Presenter(s): Daniel Stone URL: lca2013.linux.org.au/schedule/30256/view_talk (Or, 'Why Everything You've Read in LWN and Phoronix Comments is Untrue'.) Wayland is possibly the most impressive free software project to date in terms of the amount of controversy generated to the number of people who've actually ever used it. The X Window System has served UNIX and its derivatives admirably for longer than the presenter has been alive. Born in the age of monochrome displays, X's extensibility has allowed it to be dragged kicking and screaming into the twenty-first century. But this extensibility has come at a cost: the average X server now has at least two wholly unused rendering models, four input stacks, four display management extensions, et al. Not only that, but the internal design and implementation of the X server (which could generously be described as 'awful') is a serious hinderance to achieving anything. One thing as constant as X's dominance is new rival window systems which aim to replace it. DirectFB, Berlin/Fresco, QWS, the Y Window System and more, have all aimed at one time or another to unseat X, and failed. But over the past few years, since the revival of X under the X.Org banner, a great deal of its development has been directed at making it easier for others to replace X. The ongoing development of DRM/KMS, Mesa/EGL, libpciaccess, libxkbcommon and others, has actually made it practical to build a competing window system for the first time. So where do we stand now? What are the actual benefits and downsides of each? (If you said that X's network transparency is a benefit, please come see this to hear about how that's totally wrong.) Does X's extension model really mean it'll live to see 50? Will Wayland replace X next week? This talk will try to explain the design of both X and Wayland, how they actually operate in practice (which is often totally unrelated to the design), and how they actually differ. If time allows, there might well be bonus information, speculation and pontification on the future development of each. This will hopefully be rounded off with the obligatory demo of the true mark of a useful and complete window system: semi-transparent rotated windows. lca2013.linux.org.au/ - linux.org.au CC BY-SA - creativecommons.org/licenses/by-sa/4.0/legalcode.txt