Subject: Re: conditional symbolic links
To: Ty Sarna <>
From: John Brezak <>
List: current-users
Date: 11/24/1993 09:05:39
> John Brezak wrote:
> > 
> > You also might consider '@host' too. This is from DCE/DFS and is equally useful.
> I don't know about "equally", but I can see where it could be handy.
> I hope to hack on it this weekend, and I'll include @host.
> > BTW- DCE/DFS has BOTH '@sys' and '@host'.
> > 
> > @host resolves to the same thing as gethostname()
> > @sys resolves to the same thing as uname -h (Unless you have AFS; then it is
> > 	"fs sysname")
> I don't have an -h.  I have old binaries, so maybe this is a new
> addition.  Or did you mean -m? I think the @sys identifiers should be
> more distinctive than what -m gives...  Perhaps follow AFS's lead, and
> use $arch+"_"+lower($osname)+removedots($version) [eg, i386_netbsd09,
> amiga_netbsd09a, sparc_netbsd10, etc].  Unfortunately that yields
> somewhat unwieldly names.  Maybe just use "n" instead of "netbsd"?
> Comments? 
Right. (I don't know where I got "uname -h" doesn't exist; Gotta
stop putting DCE into my coffee).

I agree. The @sys should look like "i386_netbsd09" as you propose. Granted
it looks unwieldly but so does "hp700_ux900" - sigh.

> Any comments on implementing these at the global VFS level vs individual
> filesystems? I'm leaning back towards the latter now...
I think individual file systems would be most correct. For instance (if someone did)
AFS or DFS should do what is approrpiate for their handling of '@sys' and '@host'.

 John Brezak                    UUCP:     uunet!apollo.hp!brezak
 Hewlett Packard/Apollo         Internet:
 300 Apollo Drive               Phone:    (508) 436-4915
 Chelmsford, Massachusetts      Fax:      (508) 436-5103