[4155] in Release_7.7_team
Re: I guess we should modify the "lockers" man page.
daemon@ATHENA.MIT.EDU (William Cattey)
Mon Dec 15 10:37:35 2003
Date: Mon, 15 Dec 2003 10:37:16 -0500
Content-Type: text/plain; charset=US-ASCII; format=flowed
Mime-Version: 1.0 (Apple Message framework v553)
Cc: William Cattey <wdc@mit.edu>, release-team@mit.edu
To: Greg Hudson <ghudson@mit.edu>
From: William Cattey <wdc@MIT.EDU>
In-Reply-To: <1071331564.16233.80.camel@error-messages.mit.edu>
Message-Id: <8FD7DF6F-2F14-11D8-B1C5-000A957B2E52@mit.edu>
Content-Transfer-Encoding: 7bit
The problem I'm trying to solve is to unmake the decision to let
lockers cryptically fail.
We are including the athenalocker script in the OpenAFS release so that
there will be a
/mit/lockername path present. But if we really wanted to be nice to
clueless users, we could additionally in the 'locker' man page point
out that the '/mit/lockername' assumption is an assumption that should
be tested up front.
I realize I'm trying to nibble away at a belly full of 'imperfect
world', and yet...
-wdc
On Saturday, December 13, 2003, at 11:06 AM, Greg Hudson wrote:
> On Fri, 2003-12-12 at 18:41, Bill Cattey wrote:
>> Now that there's some expectation that software in "lockers" might be
>> running on systems without the suite of Athena tools, I think it might
>> be apropos to add to the lockers man page something like the
>> following:
>
> So, the kind of lockers you may have in mind are ones like the matlab
> locker, which have gargantuan front-end scripts wrapped around one or
> two programs.
>
> We also have to keep in mind lockers like the gnu or graphics locker,
> which have many programs installed and do not generally use front-end
> scripts. They also don't generally depend on Athena tools... but they
> do frequently depend on the locker being accessible through
> /mit/lockername and not just the AFS path.
>
> So, I'm not sure whether it's appropriate to put language about
> /etc/athena/version in the lockers man page. We've already made an
> explicit decision to let locker software cryptically fail when people
> don't use it via the Athena locker mechanism, and I can't imagine how
> we
> would change that without screwing something else up.
>
