modular -> monolithic

[Dave Airlie]

On Jan 21, 2008 11:11 AM, Stephane Marchesin
<marchesin at icps.u-strasbg.fr> wrote:
> On 1/21/08, David Miller <davem at davemloft.net> wrote:
> >
> > If you don't enforce validation of the build of the drivers there is
> > no quality, and people will check in all kind of shit that breaks
> > things left and right.
> >
> > That, from an engineering and code quality standpoint, sucks.
> >
> > And frankly, what you say isn't effectively how things actually
> > work.  You do have to check out large chunks of the various GIT
> > trees (and it's non-trivial to figure out the dependencies,
> > and the build.sh script is out of date half of the time) in order
> > to do X server development or even work on a single driver.  So
> > the non-modular tree in fact makes it HARDER to contribute to
> > the Xorg code base, because getting the build purely out of
> > GIT to work is so damn hard.
>
> I sure agree that some of the XACE changes were quite intrusive. OTOH,
> I'm not sure I agree with the conclusions you made.
>
> It's been my experience (and that's why I'm a bit reluctant when
> people want to switch back to something monolith-like) that it is a
> lot easier to work with modular than with the monolith. I work on a
> graphics driver, but rarely if ever build the xserver bits, because
> fedora rawhide ships up to date bits to me. This has saved me a huge
> amount of time over the last months. So I rarely have the issue you
> mentioned of getting GIT to work. It's just my experience though...
>
> Maybe the ideal model of X development keeps that nice property that
> driver developers do not need a complex and long building of the whole
> xserver, while enforcing consistent API updates.

You can still set the system up to build drivers against old X server
installed headers, the kernel also allows this if the installed bits
are up to date enough.. ask krh for how...

You only start seeing the problem when your installed bits have old ABIs etc..

> >
> > In kernel land, whilst we don't require people to validate the entire
> > build with everything enabled, we have the people who do the patch
> > applying and code checkins who do those build checks and punt the
> > patch back if it breaks some part of the build.
>
> Yeah, but the kernel is also packed with full-time contributors. X.org
> OTOH uses lots of volunteers, I think the situation is very different,
> and I don't think applying the kernel model could work here.

I pointed out to you earlier why this argument doesn't make sense, the
kernel hasn't always had the number of full-timers it does now and
still used the same model and it also has possibly 100x the number of
drivers we have.. so I can't say this argument makes any sense..

Dave.


> > The current situation completely sucks.
>
> It sucks somehow, but it used to be a lot worse :)
>
> Stephane
>
> _______________________________________________
> xorg mailing list
> xorg at lists.freedesktop.org
> http://lists.freedesktop.org/mailman/listinfo/xorg
>