Today, Jerry Callen gleaned this insight:
> > Whatever someone comes up with, it should incorporate at least these
> > features:
> >
> > * it should actually work :)
>
> Now *that's* a cool feature!
Well, as we were saying, it seems like AFS and coda don't quite from what
I've heard... For the stuff that I'm doing (providing NFS for a whole
company) it's rather important that it does...
> > * should have per-user authentication mechanisms
> >
> > * should have host-based authentication for non-sensitive data
> >
> > * should have optional, but recommended encryption capabilities
> >
> > * should be optimized for speed, naturally.
>
> ACLs would be nice, too, as long as we're smokin' dope here. And some
> local caching ideas from Coda.
Not sure I agree with either of those... I can see where ACLs would be
attractive, but I've never had a problem using standard Unix ACLs to
accomplish what I needed. I won't be offended by their inclusion provided
it's optional, but I probably won't make use of them either. Had them on
HP-UX... hated them. Of course, that was because (at least on 9.x) they
had a bug that broke my shell scripts (the -d option to the shell's test
command failed to identify directories which had extended ACLs).
As for local caching, I don't know how Coda does it but I can imagine
running into problems with out-of-date data... That concerns me.
--
PGP/GPG Public key at http://cerberus.ne.mediaone.net/~derek/pubkey.txt
------------------------------------------------------
Derek D. Martin | Unix/Linux Geek
[EMAIL PROTECTED] | [EMAIL PROTECTED]
------------------------------------------------------
**********************************************************
To unsubscribe from this list, send mail to
[EMAIL PROTECTED] with the following text in the
*body* (*not* the subject line) of the letter:
unsubscribe gnhlug
**********************************************************
