Tobias Weingartner wrote:
#: > Also, kludging up the directory structure makes no sense to me.  Why
#: > do you want to hack up something beautifull?  Why masacre (sp?) a
#: > thing of simplicity, flexibility and function?  For speed?  Gimme a
#: > break, a quick check of a machine running NetBSD-1.0a/i386, shows
#: > that there are 676 sym-links, of which the X11R6 installation is
#: > using 618, which leaves ~60 for the rest of the system.
#: 
#: A quick check of a machine running NetBSD-1.0/i386 shows that there
#: are 26927 symlinks.  Speed is definitely wanted for symlinks :-).

Okay, but what would the funny dirent business buy us?

#: 
#: Okay, I'll admit that the setup is probably a bit exotic :-) The
#: machine is running a locally developed SW package management system
#: called `store'.  Store keeps each package in a separate directory,
#: either locally on the client or on a store server.  Each client then
#: has a (sym)linktree with links pointing from
#: /store/{bin,lib,share,...} on the client machine to the appropriate
#: package files, either to a local copy or to automounted file systems
#: on the servers.  (Some witty soul decided that `symlink hell' was a
#: more appropriate name ....)  Despite the number of symlinks it works
#: quite well, simultanously handling a variety of processor
#: architecture/OS combinations as well as different versions of the
#: packages.  For those out there with a thoroughly perverted mindset
#: http://www.pvv.unit.no/~arnej/store/storedoc.html should provide more
#: info.
#: 
#: Anyway, this posting is meant as a data point for symlink usage (and
#: maybe a small plug for store :-).  I expect the data point to be
#: somewhere in the high end of the scale :-)

I'd say that this is the exception, not the norm.

POSIX needs to admit that they made a mistake on this one.  But whom
could we petition on this matter?

					--*greywolf;

#: 
#: 					-jarle
#: -- 
#: "Punning is the worst vice. There is no vice versa."
#: