[17383] in Athena Bugs
Re: references to /mit/infoagents in lynx shell scripts
daemon@ATHENA.MIT.EDU (t. belton)
Mon Nov 22 12:26:43 1999
Date: Mon, 22 Nov 1999 12:26:36 -0500 (EST)
From: "t. belton" <tbelton@MIT.EDU>
To: Greg Hudson <ghudson@mit.edu>
cc: Matthew Munsey <mmunsey@mit.edu>, bug-infoagents@mit.edu
In-Reply-To: <199911220340.WAA04503@small-gods.mit.edu>
Message-ID: <Pine.GSO.3.96L.991122122150.6501F-100000@biohazard-cafe.mit.edu>
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
This was almost the same answer I was going to write - glad I read ahead
before replying. We assume/require that if you're running a binary which
lives in infoagents, infoagents is attached.
A minimal attach is probably a good idea where the scripts permit. I'll
look into that.
-Todd
On Sun, 21 Nov 1999, Greg Hudson wrote:
> Bruce Lewis no longer works for MIT, incidentally. I'm answering as
> the Athena release engineer; Todd Belton will make the actual decision
> as to what to do with the infoagents locker.
>
> > When executing binaries in many Athena lockers, I find it preferable
> > to simply type ~lockername/bin/binaryname rather than going to the
> > trouble of adding or attaching lockers.
>
> This method of invoking programs from lockers is not supported.
> Locker software is not guaranteed to work when the locker is not
> attached.
>
> > the conflict of binaries in the Athena locker path with commonly
> > used local programs.
>
> Locker software should still work when lockers are attached but not in
> the path. So if this is the problem, you can simply attach the locker
> and then run the program from `athdir /mit/lockername` (technically
> correct) or ~lockername/bin (will work most of the time).
>
> > The obvious solution in my mind is the use of the full path in the
> > shell script reference to the lynx binary
>
> That would be poor; it would negate the desirable property that a
> locker can be copied somewhere other than its AFS path and still work.
>
> > or alternatively a minimal attach of the infoagents locker (as is
> > done with sipb, graphics, and gnu
>
> This would be reasonable if Todd wants to do it.
>
