Re: X Strike Force X.Org X11 SVN commit: r135 - trunk/debian
On Sat, May 21, 2005 at 01:03:45AM +1000, Daniel Stone wrote:
> On Fri, May 20, 2005 at 10:30:53AM -0400, David Nusinow wrote:
> > On Thu, May 19, 2005 at 11:59:23PM -0500, X Strike Force SVN Repository Admin wrote:
> > > +* Burn xserver debconfage to the ground, redesign and reimplement, taking into
> > > + account the many lessons of the past 5 years (and hopefully leveraging the new
> > > + "autoconfiguration" code in the XOrg X server).
> >
> > My goal is to push as much of this stuff in to the server as possible,
> > leaving as little for our scripts to do as we can. I know it'll be possible
> > to do things like heuristics for monitor resolutions in the server, the
> > code just needs to be written. During this process, if pieces of the old
> > debconfage can simply be removed as they become unnecessary.
> >
> > Once we have a functional and up to date X.org in Debian I'm going to begin
> > work on this. If we want to re-write the current debconf stuff during this
> > process, that's definitely feasible too.
>
> So, the autoconfig in X.Org is, uhm, interesting. The best results I've
> seen from any system have been from my meticulously hand-tweaked Debconf
> scripts, based from xserver-xfree86's Debconf stuff. The major
> difference with mine, apart from using xresprobe and the like, is that a
> lot of the md5sum insanity has gone.
Yeah, I've kept your re-worked scripts in the X.org packaging so far. I'll
probably start messing with them more as time goes on and bug reports come
in, but it's a relatively low priority for me. I need to take a closer look
at how you re-worked the md5sum stuff, but from what I saw it looks like a
good balance. New users are really confused when the md5sum thing gets in
the way. I may want to provide a debconf question saying "Your file has
been modified by hand. Are you *sure* you want to edit it?" If they answer
yes, I would back up the hand modified file so as not to lose it and allow
the packaging stuff to re-do the config. From the way users behave in
#debian, this seems to be what they want and expect.
> Long-term, it absolutely needs to go. Short-term, it should not be a
> priority over, say, actually getting xorg into the archive this
> century.
Agreed. Priority number one is getting sarge out the door, followed by
getting X.org in to unstable, but not before etch is in the works and we
can start to gleefully break stuff :-)
- David Nusinow
Reply to: