LWN: Comments on "The Future of XFree86"

The Future of XFree86

spitzak — Fri, 28 Mar 2003 19:00:43 +0000

They need to write extensions with a library that will use the extension
and EMULATE it when the extension is missing. The emulation can be really
crude (ie draw a colored box the same size as the graphic). As long as
the program works at all and does not crash, people will use the
extension, and the competing versions will be forced to add it.

Xft is an example of an extension where it will use old X fonts (however
badly) when the XRender extension is missing.

Such emulation has been missing from virtually every "extension" for X
ever made. Even the first SHM image extension failed to provide emulation
on old X servers, which is almost inexcusable (since it is trivial)! This
is a serious problem with X and why almost no extensions added since 1989
are in common use (GLX and Xft are the only ones I can think of).

The Future of XFree86 - like gcc v. egcs?

hazelsct — Thu, 27 Mar 2003 20:03:22 +0000

The analogy of gcc vs. egcs may be a good one, as far as assuaging fears of "duplicated effort". Egcs forked in order to bring a more bazaar-like development style to the project for bug fixes, optimization improvements, etc., but never strayed far from the original interface. As gcc continued its infrequent releases (like 2.8.1), egcs folded in the changes. Eventually, the gcc people saw the light, decided to adopt egcs as gcc, and everyone was happy.

But a major difference is that everything egcs did was in the open, here it seems one person worked behind the scenes to start a fork, and also made very large changes to CVS without consulting anybody. He hasn't replied (AFAIK) to the charges of dishonesty (which is apparently backed up by the email record).

The point of all this being, X forks are not necessarily bad, but at least to this outsider, Packard's way doesn't sound at all like the right way to do it.

The Future of XFree86

josh_stern — Thu, 27 Mar 2003 19:22:17 +0000

There is more at stake here than simply competition. The big issue is which
group will create *new* "standard extensions" to X11. Having two popular
groups creating and not supporting each other's extensions would be a
disaster. People are not going to boot a different X server to run different apps,
even if it is only a question of switching virtual terminals.

Bugzilla@XFree86

pointwood — Thu, 27 Mar 2003 09:48:09 +0000

They have also just started a bugzilla bug database: http://www.xfree86.org/#bugzilla

The Future of XFree86 - like gcc v. egcs?

tjasper — Thu, 27 Mar 2003 09:08:24 +0000

OOps, funging should read funding!...

The Future of XFree86 - like gcc v. egcs?

tjasper — Thu, 27 Mar 2003 09:07:06 +0000

Doesn't this sound a little similar to the gcc v. egcs issue that came about a few years ago. Since then, gcc has come on leaps and bounds with it's support for x86 based processors. Hopefully the same will happen to X development.

Also, regarding the lack of commercial support for X development. Maybe Keith is right that the hurdles to get onto the core team are too high and hence the lack of commercial interest. Could it be that it's easier for developers to be influential in other projects, hence commercial funging to push for their own ends?

Just my comments from a grateful user of all this free software....

Trevor Jasper

The Future of XFree86

gnu — Thu, 27 Mar 2003 04:40:23 +0000

Any ideas what Keith decided on his plans? Is he really going to start a new branch off?

The Future of XFree86

cpeterso — Thu, 27 Mar 2003 04:39:07 +0000

Competition is healthy. Soooo many people complain about X (and XFree86 in particular), I think a fork will help jump start X development.