Slashdot Mirror

Remember that the "RAM" allocated to XFree86 when it's running actually includes the entire address space allocated to the video card.

Thus, if you've got 64MB of RAM on your video card, XFree86 appears to take 64MB of RAM according to top.

Some of those things aren't really X's domain. (Printing is really not something that X needs to concern itself with.)

XLib sucks, agreed.

You really need to look on freedesktop.org
Replacement for XLib that uses X11 protocol; and even the same ABI as XLib is available (of course, not as good as native XCB): XCB

In fact; check out all the software on
http://www.freedesktop.org/Software/

Especially their XServer implementation is much closer to the sorts of things you want to do - client side fonts become the norm; XRender is pretty much the way everything's done etc.

Wasn't this what Keith Packard et.al wanted?

Check out this comment from the xwin.org post:

Authored by: fury on Monday, August 18 2003 @ 02:27 AM PDT
By the way (i forgot to mention this) the operators of xwin.org (including Keith Packard) are not involved in Xouvert. xwin.org != Xouvert

On the other hand, the xouvert home page links directly to xwin.org in the page header. Hard to figure...

These are trying to make a difference:
Directfb, Cairo , Fresco and PicoGUI

The discussion about framebuffering was on the XFree86 open discussion mailing list last month.

There are some nice features going into Xouvert, one of them is the auto hardware recognition system which I feel is very cool and should be the way to go.

From Xouvert Wiki:


Ouverter is a new hardware auto-detection system build ontop of Xouvert kernel module. Ouverter
provides hooks to hardware detection and auto configuration. The system would generally auto
configure most common devices; such as mice, keyboards, gaming devices and DMS supported monitors.
Ouverter would also auto detect most modern video cards, using a database of card lookup, this
database would be updated regularly and new card signatures would be added and distributed outside
the usual Xouvert releases (to faciliate new hardware detection). A windows like display change
support would be built in as well.



This is cool and much appericated feature, I wish them luck and wait for their first release.

Do you really want your windowing system to end up like Slashdot?
It already has. At least, the wiki has.

Shoud have added the following to the list: An attempt by corporate influences and the forces behind GNOME development to integrate technologies related to GNOME directly into the X server, providing an advantage to GNOME over all other window managers. In addition, much of the planning of this fork has been done in secret despite the supposed "openness" proclaimed by the team, and only recently has even the name of the fork been shared with the public. An overall damaging and irresponsible influence on Linux, and the open source community as a whole.

Shoud have added the following to the list: An attempt by corporate influences and the forces behind GNOME development to integrate technologies related to GNOME directly into the X server, providing an advantage to GNOME over all other window managers. In addition, much of the planning of this fork has been done in secret despite the supposed "openness" proclaimed by the team, and only recently has even the name of the fork been shared with the public. An overall damaging and irresponsible influence on Linux, and the open source community as a whole.

Err...Keith Packard ditched XFree86 to start his own fork, xwin...

And you're right...it does take longer to get complicated stuff done in free software land than it does when you've got how many guys being paid to do it.

The original

The fork

Thats just the two versions of XFree86, I havn't even bothered to include any commercial X servers (E.g. Metro, Sun etc.)

Seriously, I'm not trolling here. What you seem to be saying is that X is inherently slow because it scales poorly. Or are you blaming GNOME and KDE for the slowness?

I think they really need to get on with it and fork, already.

I think it's interesting that so many sites are copying the basic format of slashdot.org. This is only the most recent example. For others visit xwin.org and osnews.com. Innovation is dead.

This is being discussed here.

I went to xwin.org but could not find any type of list of what they hope to achieve.

It looks like they want to achieve extreme openness. They even put up Wiki, which means they're so open and carefree, they must emulate the goatse.cx guy. ;-)

I think the place to put your requests is WantonDesires. I think you have to register, though IIRC some Wiki sites don't require it, so maybe not. (I haven't used Wiki much.) It may be a good idea to learn how Wiki works first though...in some ways it's easy to use, but there are specifics you should understand. For example, the formatting has different rules than html. The HelpPage is a start.

As to your ideas, here are my thoughts:

1. It may be possible to optimize XFree some, but it sounds like you don't want XFree, but something else. Taking out the socket based communications would not only remove the remote abilities, but it would also reduce security.

   You are probably thinking something along the lines of DirectFB and XDirectFB. Programs use those libs to access the framebuffer device directly. From what I understand, you have to open the permissions on your framebuffer device or just run all your programs as root. Not good if you need decent security (or protection from a buggy program)--this is almost like running Win98. ;-)

2. I think there are arguments for and against enforcing specific standards, but I think most of the problems are either caused by applications not using X properly (it seems too many developers don't understand X, Unix, or even makefiles at all) or are outside the realm of X itself. I also think GNOME and KDE are unnecessary for the most part. They have probably caused a greater rift in standards...

3. Hardware configuration has always been a problem, not only for XFree86, but many other systems. PCI should help solve the problem. The PCI bus gives a vendor string, so determining the hardware you use should be easy. You'll need a massive database, but it should work. Same with USB. I wouldn't be surprised if they already have PCI and USB autodetect functions in xf86cfg, but I don't know because I manually edit my XF86Config file. I'm anal. ;-)

I went to xwin.org but could not find any type of list of what they hope to achieve.

It looks like they want to achieve extreme openness. They even put up Wiki, which means they're so open and carefree, they must emulate the goatse.cx guy. ;-)

I think the place to put your requests is WantonDesires. I think you have to register, though IIRC some Wiki sites don't require it, so maybe not. (I haven't used Wiki much.) It may be a good idea to learn how Wiki works first though...in some ways it's easy to use, but there are specifics you should understand. For example, the formatting has different rules than html. The HelpPage is a start.

As to your ideas, here are my thoughts:

1. It may be possible to optimize XFree some, but it sounds like you don't want XFree, but something else. Taking out the socket based communications would not only remove the remote abilities, but it would also reduce security.

   You are probably thinking something along the lines of DirectFB and XDirectFB. Programs use those libs to access the framebuffer device directly. From what I understand, you have to open the permissions on your framebuffer device or just run all your programs as root. Not good if you need decent security (or protection from a buggy program)--this is almost like running Win98. ;-)

2. I think there are arguments for and against enforcing specific standards, but I think most of the problems are either caused by applications not using X properly (it seems too many developers don't understand X, Unix, or even makefiles at all) or are outside the realm of X itself. I also think GNOME and KDE are unnecessary for the most part. They have probably caused a greater rift in standards...

3. Hardware configuration has always been a problem, not only for XFree86, but many other systems. PCI should help solve the problem. The PCI bus gives a vendor string, so determining the hardware you use should be easy. You'll need a massive database, but it should work. Same with USB. I wouldn't be surprised if they already have PCI and USB autodetect functions in xf86cfg, but I don't know because I manually edit my XF86Config file. I'm anal. ;-)