Re: Fixing /etc for native X.Org

[Alan Barrett <apb%cequrux.com@localhost>]

On Wed, 19 Nov 2008, Quentin Garnier wrote:
> > > * For "postinstall from $SOURCE", we'd need to do some smarts
> > >   about invoking make(1) and determine the build settings
> > >   of MKX11 and MKXORG.   Doesn't etcupdate(8) have some useful
> > >   and robust code for that?
> > 
> > That code is not that robust.  We have to make sure all the variables
> > are passed down to the make which is invoked.  And in the typical run
> > of build.sh install=/, you're told to run postinstall in a certain way;
> > if you had -V arguments to build.sh, they need to be passed down there
> > too.

Well, sure.  But that's the user's problem.  If the user insists on
running postinstall or etcupdate in non-recommended ways (using a source
tree instead of using etc.tgz and xetc.tgz set files), then they are
responsible for passing all the variables that affect the build.  All we
have to do is document it.

> In addition to that, I've spent a bit of time on it today, and
> etcupdate(8)'s mechanism is _not_ the way to go.  It's far from robust,
> and notably won't deal with a read-only source tree.

We should fix that.  Can you explain in more detail?  I would have
thought that the user could pass appropriate MAKEOBJDIRPREFIX variables
and such to allow it to work in a read-only source tree.

> > There's a thrid option:  using /usr/X11 everywhere and providing a
> > symlink to either X11R7 or X11R6 depending on the settings.  That would
> > remove the need for most of my original patch.
> 
> Any other opinion on that?  That would avoid generating the rc.d scripts
> and solve the problem Once and For All, just as we should drop a big ice
> block into the ocean to solve global warming.

Yes please.

--apb (Alan Barrett)