Linux Desktop Without X11

A writes "Rocklyte systems have announced the first version of their Athene Operating System. It is a desktop and embedded operating system built on the Linux kernel, but without the "aging X11". Instead, it uses the SciTech SNAP graphics system with which it is possible to completely re-theme the desktop to look like the famous AmigaOS GUI or another famous UI. For backwards compatibility, an X11 server is also available in the system. The system can run completely off the CD, without needing to be installed on the harddrive."

Comments

something i always wondered about

[b17bmbr]

why is linux so beholden to X? yeah, i love it's network transparent features ( i use in my class every day), but, look at what apple did with essentially a kernel and subsystem. they could port aqua to linux, since it already compiled under gcc anyways. that might be a huge commercial ticket for linux. certainly there is nothing that says you can't run a proprietary windowing system on top of the kernel, is there? app compatibility would be a huge issue, but like apple's X11, it could run rootless, and almost be unnoticable (except for the widgets).

Re:something i always wondered about

[maynard]

I'm sure they could take advantage of the RENDER extension, but I don't really see the advantage of a whole Postscript interpreter under there.

One of the big advantages to DPS and DPDF is the device independence in rendering text and other objects. That is, it's truly WYSIWYG - what's rendered on the screen is exactly what will be rendered to print (or any other device). For DPS you also have the option of writing applet procedures which run on the display server, similar to the old Sun NEWS system. So, for example, one could write a terminal emulator in postscript and have it run in the display server, thus reducing network load by cutting out most of the transmission of event processing between client and server during a remote display session. This solves the biggest complaint about network transparent X sessions, that being it's a network hog and has terrible latency.

X is fine for what it does - especially given the price - but it saddens me to no end to see the DGS render extension die because no one seems to care, while at the same time everyone bitches about slow old X. GnuStep with Display Ghostscript would certainly have been a better solution than completely rewriting a new display server and the rewriting the windowing environment all over yet again.

So now someone is selling cool new desktop that will never cross the threshold of users necessary to replace X, while others keep dumping more intellectual energy in bogus free X desktops that 'kindof' work. How many times have we done this? How many widget toolkits does X really need? Athena, Motif, TK, QT, GTK... on and on and on. None of them work well together, everyone needs applications that cross toolkit boundaries, and users are left completely in the dark on how to do the simplest thing like cut and paste non-text between applications. Wasn't Simpson Garfinkle bitching about just this in the UNIX Haters Handbook ten years ago?!?!? And everyone laughed because it was true while nothing changed. Feh.

We're long past the point where the history of X, and ridiculous backward compatibility, is impeding growth toward something new and better. Gnome and/or KDE ain't the solution. Nor is GnuStep rendering through xlib. Feh, what a mess.

JMO,
--Maynard

Yes, they are important.

[ethnocidal]

You seem to connect themes with simple bitmap changes, and the like. I agree that simply offering this functionality is less than important.

However, with 'true' theming, the internal function of the GUI (and OS) is loosely tied with the graphical layout and function of the GUI. What does this means? It means that a single system, properly configured, can handle many different interface styles. You could simultaneously offer transition interfaces to users from different GUI camps - Windows, MacOS, NeXT, etc.

This is an immensely important feature for this reason. While many see theming as eyecandy, properly implemented it can serve a very useful purpose; fit the GUI to the user, not the user to the GUI. It should also allow new interface styles to be prototyped - what better way to develop usability than to look at what people with the skillset to change the interface think works best?

XFree86 good, not bad

[0x0d0a]

Most people that dislike X don't understand it.

My favorite complaint is that it's bloated or eats too much memory. It's bogus -- X uses relatively little memory itself, but pixmaps are stored in X instead of in apps. So Linux GUI apps tend to use less memory than they would with a Windows-like environment, but X's memory usage go up.

I actually sat down and modified some code to query X how much memory is being used by each program in pixmap memory. This is memory that would have to be used under Windows. Little things -- gkrellm, a simple dock program that I have running, caches about 2.7MB of pixmaps in X all by itself. This doesn't show up as gkrellm memory usage in top, but it *is* being consumed by gkrellm.

X11 allows network transparency, 3d support, hardware scaling of video, support for more font formats than Windows does, zooming in and out. When combined with a window manager, the X11 architecture is incredibly powerful and flexible.

I wish people would stop complaining about and learn to use X's features.

Re:XFree86 good, not bad

[0x0d0a]

from what I can tell all it does is serve fonts and networktransparancy.

X is responsible for actually rasterizing and displaying every pixel that you see on the screen. It renders fonts, yes, and very nice antialiased ones. It handles network code, yes. It (well, it and Mesa) do 3d hardware rendering -- in Windows user terms, all of the video card drivers in Windows combined with DirectX. It does hardware scaling -- if you play a movie, Xv is used to display the thing. It handles combining multiple monitors via Xinerama. It acts as the intermediary in copying and pasting data between apps. X deals with tablets, joysticks, mice, keyboards and handing off data from them to apps. X provides framebuffer access to memory. Unlike Windows, X lets you fine-tune precisely what timings are used on your monitor, if you want to squeeze the last little bit of performance possible out of your monitor.

if you want to do anything usefull you have to add a window manager

Sure. X could have included a window manager, but the folks that write it realize that different folks prefer different types of window managers. Some prefer really simple WMs like twm, metacity, or kwm. Others prefer glitz and don't care about plenty of overhead, and use enlightenment. Others like poking at and customizing their window manager, recoding bits of it while it's running (a la emacs), and use sawfish. The list goes on and on. Most *ix folks tend to feel a bit irritated when being forced to use the Windows environment -- there's no possibility of choice, and relatively little of customization.

a cut & paste manager

Well, you *can* use a multi-clipboard program, (of which there are a collection to choose from) but Windows doesn't provide this functionality natively either. Just as with window managers, this modularity is done deliberately. Distributions can prepackage a multi-clipboard program if they like -- so the end user experience can be "there's one, it's preinstalled, and I don't have to worry about it" -- but you aren't *forced* to use any single one.

a toolkit of somesort (gtk for example)

Again, Windows happens to force people to use a single widget set. I'm not a tremendous fan of chunks of the Windows set (anyone that's done gtk programming and Win32 programming knows that layout in gtk is *much* better than the forced pixel-level layout used in Win32 and the Macintosh Toolbox), but it can't really be changed for backwards compatibility reasons.

X is modular. If a widget set falls behind the times, a new one can be produced. I'm not sure if you've ever seen Athena, but it was one of the earlier widget sets available for X. I suspect that most desktop users would not like the way it operates. With Windows, you'd be stuck lugging around Athena forever. With X, you can simply move to something newer, like gtk.

hell even windows 3.1 does far more then X and can be cut down under a meg and still be 100% usefull, not to mention that adds a multitasking ( a bad one but still) to the OS (dos)

Win 3.1 and X are completely different beasts. They don't do even remotely the same task.

Win 3.1 is marketed differently. X *has* a partial equivalent in Windows, but you cannot obtain it separately from the rest of Windows. However, it's really irrelevant. You'd never use X without a kernel, so the fact that Windows 3.1 does scheduling isn't really useful.

(i.e. drag and drop doesnt work 80% of the time unless all you use is kde apps)

Drag and drop cooperation between gnome and kde is relatively new. Yes, it was added recently, and it takes a while to get in. I used Mac OS in the 7.x days, when drag and drop support was added...and the same thing happened -- actually, it was even worse, if anything.

I'm not saying that X is unilaterally more featureful than Mac OS or Windows. Drag and Drop is a particular weak point that's being added to a lot of apps right now. Overal

Myabe X11 just needs another revision

[headbulb]

X11 isn't as Bad as everything thinks.
The way I see X being slow is that widgets need to be on server-side instead of client-side. Right now the client Draws everything useing X primitives, sending the raw data (pixmaps, whatnot) to the server over the network. Now if the server had the widgets on its side the client would just have to tell the server the type, size, position of the widget, Instead of sending a pixmap.

This would help things such as less bandwidth, less cpu overhead for eash client.

Maybe this could even be implimented in a X-
extentsion

Maybe I am just showing my ignorance here, But an idea is an idea.

I'm happy with XFree

[pivo]

Yeah, this was sort of a revalation to me recently. I just got a new machine at work. Of course it came with Windows (XP Pro) installed on it so I played with it for a few days before blowing it away and putting RedHat 9 on it. I was sort of hesitant to do so because windows in XP seemed to pop up fast and the whole system seemed very responsive (not that I was doing anything heavy duty.) Another plus is that font rendering is actually better than Windows, and about equal with Macs. That's really nice.

Anyway, now that I've got RH installed (w/XFree 4.3.x) I am very happy to say that X seems just as responsive as Windows, even when I am doing something heavy duty, and I'm using KDE as well. This was the first time in about five years I've used any kind of Windows, it was a nice validation of X as far as I am concerned.

XFree, at least without propriatary drivers, might not be great for games, but it makes my development life a lot more joyful than other non-networked windowing environments would, and that includes the kludgy windows terminal services crapola.

Re:Themes schemes

[NanoGator]

"If I could re-theme my hardcoded windows GUI, themes would be the most important thing, ever. "

You can re-theme it. Check out this thread here.

Here's what my desktop looks like. It's customized with my own (in progress) artwork on it. And yes, those are buttons and multiple desktops there. Some of the stuff there is default, and some of it I added on my own.

So yes, you can modify your 'hard-coded' theme. Somebody's already gone through all the work to do it.

I've been waiting for this day...

[neo]

I'm so happy to hear someone finally ditched the X windows. Now maybe we can get some decent applications without needing to code the whole UI experience every time.

This may be the one.