[20] in pc-kerberos
[Forwarded: Re: Kerberos on PC/TCP]
daemon@ATHENA.MIT.EDU (Shawn Mamros)
Fri Jun 3 13:38:21 1994
Date: Fri, 3 Jun 94 13:36:32 EDT
To: pc-kerberos@MIT.EDU
From: mamros@ftp.com (Shawn Mamros)
[Forwarded on behalf of Eliel, who sent it to just me by mistake...]
Received: from ftp.com by mailserv-C.ftp.com (5.0/SMI-SVR4)
id AA16061; Thu, 2 Jun 94 18:24:08 EDT
Received: from ftp.com by ftp.com ; Thu, 2 Jun 1994 18:20:18 -0400
Received: from panix.com by ftp.com ; Thu, 2 Jun 1994 18:20:18 -0400
Received: by panix.com id AA04328
(5.65c/IDA-1.4.4 for mamros@ftp.com); Thu, 2 Jun 1994 18:20:16 -0400
From: Eliel Mamousette <eliel@panix.com>
Message-Id: <199406022220.AA04328@panix.com>
Subject: Re: Kerberos on PC/TCP
To: mamros@ftp.com
Date: Thu, 2 Jun 1994 18:20:16 -0400 (EDT)
In-Reply-To: <9406022157.AA15666@mailserv-C.ftp.com> from "Shawn Mamros" at Jun 2, 94 05:57:04 pm
Mime-Version: 1.0
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: 7bit
Content-Length: 2031
>
> Well, if it's interoperability you want... :-)
>
> Our ticket files use the same format as does MIT's UNIX code, pretty
> much byte-for-byte. So, I don't think there could be too many objections
> there... ;-)
>
> As to location... we use the PC/TCP configuration file (pctcp.ini) to
> determine that. As our documentation states, our ticket files are
> located in the directory designated by the "directory" parameter in
> the [pctcp kerberos] section of pctcp.ini. The name of the file is
> "username.tkt", where "username" is determined by the "user" parameter
> in the [pctcp general] section of pctcp.ini. If for some reason the
> "user" parameter is undefined, we fall back on the filename "ticket.ses".
>
We chose a different approach when doing the DOS port because we knew
that the tickets would be used on machines in public computing clusters.
Having no real way to ensure a /tmp-style clearing of the ticket file,
we went with the KERBMEM scheme of creating a block of memory and
storing the address of this memory in a environment variable. The idea
is that we can at least expect the user to reboot when they leave....
> I can understand how people might think the use of pctcp.ini for ticket
> file location is too "proprietary" for a general solution. If someone
> would like to propose a "standard" for how to locate ticket files, we'll
> gladly adopt it as soon as we can (while still supporting our current
> scheme as a fallback for our customers, of course... :-) Sound reasonable?
>
One scheme might be to store the file name in an environment variable,
a la kerbmem scheme. In fact, if we use the standard of trying to stat
whatever KRB_TKTLOC (I just made that up) resolves to, and if there's no
file, trying to read memory at whatever address it resolves to....
This is right off the top of my head and not well thought out, but I think
y'all get l'idea general?....
- eliel
> -Shawn Mamros
> E-mail to: mamros@ftp.com
>
>
Eliel Mamousette
System Constructs, Inc
Eliel@Panix.COM
