[41] in athena10
Re: ATHENA_SESSION_TMPDIR
daemon@ATHENA.MIT.EDU (Greg Hudson)
Sat Jan 12 01:21:53 2008
From: Greg Hudson <ghudson@MIT.EDU>
To: Tim Abbott <tabbott@mit.edu>
Cc: athena10@mit.edu
In-Reply-To: <Pine.LNX.4.64L.0801120050590.26368@mega-man.mit.edu>
Content-Type: text/plain
Date: Sat, 12 Jan 2008 01:21:44 -0500
Message-Id: <1200118904.6088.7.camel@error-messages.mit.edu>
Mime-Version: 1.0
Content-Transfer-Encoding: 7bit
The original purpose of ATHENA_SESSION_TMPDIR was to create a place to
put ORBit state so that it would not be shared between distinct login
sessions on the same machine. r18905 in the Athena 10 repository
contains the relevant local modification to third/ORBit2. I believe
this is still the only purpose of that variable in Athena.
We don't want components reused between login sessions because different
login sessions will have different AFS and Kerberos credentials; for
example, trying to reuse an Evolution Bonobo component from a stale
login session will typically result in confusing error messages and no
obvious workaround.
I discuss plans for this particular hack in
http://web.mit.edu/release/www/athena10/#session where I propose to
preserve the change. That's not a final decision, though.
If we do decide to modify ORBit in this manner (or get it modified
upstream in time for Hardy, though that seems unlikely since we haven't
been pursuing the matter) then your PAM module approach seems like a
good one.
