<div class="gmail_quote">On Wed, Feb 22, 2012 at 5:13 PM, Daniel Stone <<a href="mailto:daniel@fooishbar.org">daniel@fooishbar.org</a>> wrote: <blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"> Hi, <div class="im"> On 22 February 2012 21:34, David Jackson <<a href="mailto:djackson452@gmail.com">djackson452@gmail.com</a>> wrote: > As I mentioned, counting on applications is really not a reliable method, > some programmers are sloppy and wont do it. The capability really needs to > be somewhere in wayland to sort of be able to count on it. </div>Toolkits deal with these kinds of issues so applications don't need to. If you decide not to use a toolkit and write your application badly, then yes, it won't work well. That's not something Wayland can or should attempt to solve. <div class="im"> </div></blockquote><div> I completely disagree. Toolkit developers themselves can be sloppy. Someone will develop a toolkit, perhaps not a very well known one, and simply not include any way to switch its target to X11 at runtime. Or, an application developer may not even decide to use a toolkit and will use their own junk. People do it, You may think its dumb, crazy etc, but the fact is, people do some things we think are dumb. So we cant rely on programmers of applications, toolkits etc to implement this feature. Wayland needs to provide it in some way. Despite my misgivings on wayland and my being skeptical of it, I would be happy enough and placated if you can implement this one feature for your users: All I need is the ability to display a single wayland application to an X server as an xclient connection over X protocol, with network transparency, with a command line, run time, per process argument to tell it the target X server to use. I dont care as much about the particulars of how this is done, just that it can be done in some way. Here is where I am coming from as well: I am happy with X window system. I heavily use network transparency. Many people do. Most users, including newbies, I show them what can be done with network transparency, they love it,. Its always been a strength of Linux. Since I am happy with X Windows, Wayland does not solve a problem for me, in fact, it will create problems due to the inevitability of some apps becoming wayland only. X's flexibility is the anti-MS-Windows. Its a flexible system that allows a great deal of control for users, including being able to swap window managers on the fly. Sadly what i see with wayland is an inflexible, rigid and for me useless MS Windows type thing. Linux also has a strength in running on older hardware. Again, Wayland i worry may be inferior here as well since a lot of drivers for old hardware are a part of the X server. I am not very fond of wayland, I like X, it works wonderfully for me. I thought the Linux desktop was great with X and all of the flexibility it offered. And X is backwards compatable so I can use older X applications without problems. So, I have been very skeptical of this idea to create this incompatable window system, Wayland, that i am worried does not support old hardware, is inflexible, difficult to use and so on. Although wayland does not do anything I need to do, Whatever wayland does could had been done with an X extension I am sure and with maintaining backwards and forwards compatability. The big concern fro me is that will, there will be a lot of wayland only applications, and I basically want to keep using X, my X window manager, my old hardware, my network transparecy, my ability to have multiple X sessions, etc, I dont want to have to end up using this wayland desktop very much just to be able to see applications. and I can gaurantee knowing how people and programmers operate, applications hard tied to Wayland API and only useable with it. Its gonna happen. What problem does wayland solve for me? Nothing. What problems does it create for me? Everything. Its hard to beleive, but Wayland basically ruins the Linux desktop as is, the best things about it, which is support for old hardware and network transparacy, and the compatability, you can run any X application and not worry about if its compatable with your choice of window manager. X windows seperated the X server from the window manager, meaning the window manager could be changed on the fly on a running desktop, and since, the applications do not depend on the window manager. This was good design. It allowed people to use any window manager they wanted and make their desktop work how they wanted without worrying about if applications are compatable. It almost seems like Wayland is making every mistake X avoided and not doing any of the good things X did. I also dont understand waylands basic principles as far as I understand them. For many years, I thought that the whole point of having a GPU was to offload graphics rasterisation to the GPU rather than do it in CPU. But, from what i hear of wayland, applications give wayland a pixmap, meaning the application has to do all rendering on the CPU? To offload graphics operations to the GPU, a rich 3D vector graphics API needs to be provided for the app to send a larger set of high level graphics operations to the GPU, via the server. The more high level and extensive that API is the more the application can offload to the GPU in rasterization. What i mean by application is both the app and its toolkit it is using. Such an API exists already: OpenGL. And surprise, surprise, X window system already has a protocol that is perfect for sending OpenGL commands to hardware, the GLX protocol which can be implemented on the server end to send high level graphics to video hardware. if you attempt to give apps direct access to video hardware.; I am skeptical of that. The X server is basically a trusted program to have access to video hardware while applications such as web browsers are less trustworthy. X shielded hardware from applications which, like webbrowsers, sometimes have bugs tht allow a remote attacker to work through such an application. By giving applications direct access to hardware we dangerously break through protection rings and allow greater risk for a damaged application to be used to expoit graphics hardware vulnerabilities. Especially true of graphics hardware address space is being page table mapped read write into an applications virtual address space. if you allow wayland applications to be displayed to an X server, it would allay these concerns of mine. It will assure the wayland applications can be used with an X server that has the drivers for old hardware, it will assure I can have my X desktop and multiple X sessions still including by VNC (xvfb+x11vnc) only X servers and so on. I can have my network transparecy. </div><blockquote class="gmail_quote" style="margin:0pt 0pt 0pt 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex"><div class="im"> > If an expose event is not given to an application, how can it know when to > send new data to be displayed in just exposed areas of the app. </div>Because all areas of surfaces are (can be) visible at all times. Expose events simply do not exist, for a multitude of reasons. Cheers, Daniel </blockquote></div>