Canonical Scandal URLs

Our canonical URL is
http://www.cs.cmu.edu/~scandal/.

Keep reading if you want to find out why there are so many different URLs floating around out there for the Scandal project, or go straight to the moral of the story.

The Sordid History

The Scandal project jumped on the WWW bandwagon early, in June 1993, when it was basically the Mosaic bandwagon. At that time, CMU SCS had one rather underpowered web server that was also running an FTP server. Due to security concerns, the httpd daemon was restricted to a non-root port (8001). Also, no aliases were available to shorten the path to our files in the AFS file systems. So our canonical home page URL looked like
http://www.cs.cmu.edu:8001/afs/cs.cmu.edu/project/scandal/public/mosaic/home.html

We quickly realised that the web is the important thing, not Mosaic, and changed that to:
http://www.cs.cmu.edu:8001/afs/cs.cmu.edu/project/scandal/public/www/home.html

Everyone was still copying URLs around by hand (ick!), and some people created links that used the SCS alias of /afs/cs/ instead of the fully qualified /afs/cs.cmu.edu/, resulting in URLs like
http://www.cs.cmu.edu:8001/afs/cs/project/scandal/public/www/home.html

About this time, other SCS projects started creating home pages, and the departmental web server became more and more overloaded. Finally, we gave in and set up our own web server, which serves the same files out of AFS:
http://web.scandal.cs.cmu.edu/afs/cs.cmu.edu/project/scandal/public/www/home.html

Note that web.scandal.cs.cmu.edu is just an alias that currently resolves to parallel.scandal.cs.cmu.edu. We did it this way to enable us to move our web server to a new machine without breaking anything. Of course, some clients can't handle aliases transparently, so URLs of the form
http://parallel.scandal.cs.cmu.edu/afs/cs.cmu.edu/project/scandal/public/www/home.html started to leak out.

We also created path aliases on our server, resulting in
http://web.scandal.cs.cmu.edu/www/home.html

which we thought was getting close to an optimally-short URL. Then the main SCS server implemented its own alias scheme, enabling us to shorten that URL also, first to
http://www.cs.cmu.edu:8001/Web/Groups/scandal/www/home.html

and then to
http://www.cs.cmu.edu:8001/~scandal/home.html

Next, SCS decided that they should get with the program and use the standard httpd port of 80. So we became
http://www.cs.cmu.edu/~scandal/home.html

or just
http://www.cs.cmu.edu/~scandal/

Finally, a second server was added to (transparently) share the load on www.cs. Naturally, some clients can't handle this either, so they report URLs like
http://www-cgi.cs.cmu.edu/~scandal/

And that's where the story ends (for now). I haven't bothered to add up all the possible combinations of servers and aliases, because the final figure would probably be too depressing, but I've tried to gracefully retire some of the older URLs. Just remember, http://www.cs.cmu.edu/~scandal/.

November 1996 update: www.cs is finally able to handle the load. Everything on http://web.scandal.cs.cmu.edu/www/ now redirects to www.cs, and I've added base hrefs everywhere to match...

The Moral Of The Story

Never be the first to jump on a bandwagon.

Jonathan Hardwick. Updated on 2 March 1996.