On Mon, May 07, 2001 at 03:03:47PM +0200, Oystein Viggen wrote:
> Quoth Marcus Brinkmann:
> > Right. Same for /usr/X11R6/man and /usr/X11R6/include. It should be easy.
> <X11/whatever.h> is actually a bit of a different reason. There would
> be no problem having .h files with the same names here, just like
> <time.h> and <sys/time.h>.
What I meant is that there shouldn't be any clashing /usr/include/FOO that
is also in /usr/X11R6/include/FOO. Using <X11/*> makes absolutely sense.
> Anyway I think we can conclude that all _should_ be fine and dandy, but
> I won't believe it until I actually see it. (I think I'll have to leave
> work early today...)
I couldn't believe it for /usr -> ., too, but it was really painless (except
dpkg-shlibdeps of course ;) I think there were two conflicts, one was
a link in ae and one was a link in cpio.
> > IMHO, the LSB/FHS people should quiver in their boots because of the
> > whole /usr/X11R6 excemption. We are just correcting some historical bugs
> > here.
> Agreed. Separate subtrees are nice if you use make install and rm -Rf
> as your packaging system, but really just a bit silly when we have dpkg
> and rpm.
Exactly. I use stow in /usr/local, for example (installing to
> > When we have some experience with it, I will try to build some Debian X
> > packages with prefix "" (thanks, Roland) and a few symlinks corrected.
> > Branden is also interested in this, so we might as well do some leg work.
> > It would really be cleaner for us this way.
> Would Branden actually consider something to the effects of a
> /usr/X11R6 -> . symlink?
Let us first work out a patch, and then test it.
`Rhubarb is no Egyptian god.' Debian http://www.debian.org email@example.com
Marcus Brinkmann GNU http://www.gnu.org firstname.lastname@example.org