Re: [forum] Suggestion for XFree86

[Torgeir Veimo <torgeir@pobox.com>]

Let me first say that I really appreciate the efforst of the members of
the XFree86 project. I use your software every day.

On Thu, 2003-03-20 at 19:07, Frank LaMonica wrote:
> David Dawes has asked that anyone with an interest in X or XFree86
> contribute to this discussion, so I am taking this opportunity to air a
> suggestion that I hope the current BOD and slate of XFree86 officers will
> seriously consider.

I'm not shure a voting model would work at all. A voting model works
when there is a budget with money to spend. Where one can hire someone
to perform a task, to develop something. It doesn't work when most of
the work is voluntary. You simply cannot say to someone who works in
their spare time to do this and that, because it was the outcome of a
democratic vote. 

A better model is one where working code speaks by itself, where the one
can judge based on technical merit, is the best model. I think the
problem a lot of people see is that this is not how things works with
how the XFree86 codebase evolves. 


I think a lot of things would improve, if the barrier to experiment with
the current XFree86 codebase was simpler. 

I'm not saying experimenting is impossible as it is, but there are
barriers to entry. First of all, it's about attitude. Don't dismiss new
developers, when some unexperienced developer ask a question on any of
the developer lists. 

Second, think about which areas of the code repository, the code itself,
or the build system which could be improved to make it easier for
someone unfamiliar with the XFree86 cvs tree to make use of parts of the
code. For instance, Imake may have been very good five years ago, but is
not very widely used today. Could things have been improved by using eg.
autoconf? What if someone wants to make use of the X server protocol
code to implement X11 compatibility on their fuzzy logic windows system
that only runs on 68k amigas? Are those things as easy to do as they
could be?

A lot of people are of the opinion that the X protocol itself is a
barrier to improving the "multimedia desktop experience" on Linux.
Instead of dismissing their claim, let them experience it for
themselves. Make it easier for them to try to implement their own
personal alternative to the XFree86 X window server. Let them fail, gain
some experience in why things are done the way they are, and you gain
valuable resources which will later contribute back to the XFree86
codebase.

Can the modularization of the XFree86 code be strengthened? Can eg XAA
drivers be made easier to use outside of the XFree86 X server. If
someone crazy enough wants to make use of these drivers in the kernel,
let them try and fail if they absolutely must. Make it easy for joe
developer to copy anything under directory xyz/Xserver/moduleB into a
separate directory, and do a make to get a dll with functionality he can
use to implement his Q windows server with integrated lisp interpreter
for his welsh window manager. Just make it easier for these to 

What about the XFree86 cvs repository, how big is that? Are there
directions available for those that only wants to get and build their
own XFree86 X server? Do they need to download 300 megabytes of source
in order to discover how to change a property in Imakefile X in
directory Y in order to build only the server?


Think as a new fresh developer. Where should I start if I want to work
on feature X of the X11 window system?

Imagine if you could attract 100 new developers. Even if most of them
work on crazy stuff, the one or two of them that come up something
really new and powerfull, is worth all the whining of those that only
wants to reimplement the X server in C++ or phyton, that wants someone
else to implement feature X for transparent trash cans for them, etc.


I guess a lot of my comments are really not related to the discussion
here, which seems to revolve around really experienced XFree86
developers wanting more influence on development directions. But I do
think my comments are valid. 

-- 
Torgeir Veimo <torgeir@pobox.com>