modular -> monolithic

[David Miller]

From: Nicolas Mailhot <nicolas.mailhot at laposte.net>
Date: Tue, 22 Jan 2008 22:32:12 +0100

> What Xorg needs more than a new build system is a ruthless long-term
> release manager who has the authority to:
> 1. refuse patches that make tinderbox turn red and
> 2. kill old code and configurations no one actively maintains (old
> drivers, non hal-input if hal is the future, etc)

I'm very glad someone brought this up.

The problem currently, which Linux didn't have, is that the people
most qualified to do what you suggest and are currently hacking Xorg
won't step up to do it.

And it's not because they are not able, and it's not because
they aren't themselves willing to do it.

The problem is that doing so puts them into a political bind they
don't want to be in.  They fear that rejecting peoples patches and
"taking control" of the tree will be perceived as a power-move by the
company they work for.

This is also, BTW, a major reason why the fork from the XFree86
project didn't happen years earlier than it did.

Linus had the luxury of taking hold of the tree from the beginning,
gaining trust and a sense of complete transparency over the years
before he was employed by anyone.  With that precedence in place even
once he worked for a company he told the world that it would not in
any way influence kernel development, and people had every reason to
believe him.

None of the Xorg developers can do this so easily, there is too much
history and too much (at least perceived) assosciated corporate
interest.

Now if we could get an independant organization like the Linux
Foundation to pay one of the primary Xorg developers a 6-figure salary
to be the Xorg patch integrator and release manager, it could work.