[165] in Athena User Interface
Integration with non-Zephyr IM systems
daemon@ATHENA.MIT.EDU (Christopher D. Beland)
Tue Jun 13 02:51:04 2000
Message-Id: <200006130651.CAA22514@No-Whammies.mit.edu>
To: aui@MIT.EDU
Date: Tue, 13 Jun 2000 02:50:59 -0400
From: "Christopher D. Beland" <beland@MIT.EDU>
So I've been doing some research, trying to see if there's a way we
can get most of our gzephyr for free by using a generic GUI chat
client that also supported other popular IM protocols, such as IRC,
AIM, and ICQ. My conclusion: we can't. What follows is a general
clue dump and recommendations for what unsupported software we should
make easily available in this area. Doing this would, I think,
satisfy a significant and growing user demand for these services,
which has shown up in a couple of places, with a trivial amount of
effort on our part.
In general, there are four reasons to go with one-protocol clients
rather than combined ones:
- Feature lag. One-protocol clients (especially in Unix land, which
lags behind the Windows world) tend to have support for more useful
features sooner.
- Ease of use. Combined clients tend to be more confusing,
especially for someone who would only want to use one protocol. Both
setup and operation are more complicated, and user interfaces
generally much less mature.
- Security. While most IM protocols have very minimal security to
begin with, it's even less secure to store multiple passwords for
various services in configuration files or on a third-party server.
None of the combined clients I've seen even bothered to keep passwords
from being printed in plain text on the screen.
- Stability. One-protocol clients don't have inter-protocol
glitches, and tend to be more mature and stable.
I recommend providing the following as unsupported software, in an
Internet-related or IM-related Gnome menu:
- GnomeICU - ICQ - Helixcode distribution
- XChat - IRC - Helixcode distribution
- Tik - AIM - sipb locker / tik.sourceforge.net
- GAIM - AIM - Helixcode distribution
The first two combine the best in usability and feature support of
available clients I saw. They are also already part of Gnome.
I recommend Tik over GAIM (Gnome AIM) because it has better feature
support. I've also never had any major difficulties getting and
keeping Tik running, whereas I had to install a new version of GAIM
myself to get it working under Linux. The newest version is even
cooler than the one in the SIPB locker, but requires Tcl/Tk 8.2. It's
possible that GAIM will eventually overtake Tik, especially if we are
stuck with an older version. There's no pressing reason I can see not
to include both of them on the menus since GAIM is already part of the
Helixcode distribution and has the potential to become more popular
and featureful as it gets more attention. (There's also a web Java
version of AIM direct from AOL, which might be useful to note in
documentation, for people who might want to run AIM on unsupported
platforms. It's at http://toc.oscar.aol.com/.)
In my poking around, there were only three projects I found
particularly exciting.
The first was Gabber (http://gabber.sourceforge.net) which is a Jabber
client for Gnome. Jabber (http://www.jabber.org) servers talk to
their clients using a home-grown but open source and extensible
XML-based protocol. The clients implement generic person-to-person,
multi-person, text-only, and multimedia chat features, speaking only
Jabber XML. Translator programs serve as gateways between Jabber
servers and external "transports" which currently include IRC, AIM,
and ICQ. The promise of the system for MIT would be to get a chat
client for free by simply creating a zephyr gateway for the server.
Theoretically, we would also be able to gateway any of the supported
services to any other.
Unfortunately, this seems to be a losing proposition. We would have
to Kerberize not only the server and its translator, but also the
clients. The Jabber client interface, due to the complicated
gatewaying interfaces, is extremely confusing to both set up and use.
Existing clients also do not support anything like zephyr subscription
management, so we'd have to add that in, too. We'd also have to run
our own Jabber server with Keberberos and Zephyr support, in addition
to an IRC gateway to any and all IRC networks we'd want users to be
able to connect to. (The only public Jabber IRC gateway now running
is to a small network mostly used internally.)
The second interesting project was Everybuddy
(http://www.everybuddy.com). This is a client which supports ICQ,
AOL, MSN, and Yahoo, and is implemented in GTK. As such, it might be
useful to steal some message displaying code from. Otherwise, it's an
interesting project to watch for future developments.
The third is TikLeash, which has not yet been released, but which
promises to unify ICQ, AIM, and IRC in a Tcl/Tk client
(http://gtcs.com/tik/).
Anyway, I've gotten some useful ideas for how to make gZephyr a
state-of-the art IM GUI. I'll be floating a design as soon as I get
some more important work done.
-B.
===============================================================
Christopher Beland - http://web.mit.edu/beland/www/contact.html
Got spam? Stop it at the source. http://spamcop.net
===============================================================
