This page describes what needs to be done to preserve the web services currently offered by the SIPB web server (www.mit.edu).

Connections to www.mit.edu on port 8008 are used for an interface to the discuss service. This could, for example, be handled by forwarding tcp traffic on port 8008 to a server that SIPB operates. For example, a tcp connection to www.mit.edu on port 8008 might be forwarded to port 80 on the host www-discuss.mit.edu. (www-discuss.mit.edu is an example name -- the real name may be different.)

Connections to www.mit.edu on port 80 or port 8001 are for a standard HTTP server (e.g., Apache). The document space is identical regardless of whether port 80 or port 8001 is used. Presumably some URIs would be handled directly on a machine named www.mit.edu, and some URIs would be passed along to a server that SIPB operates (e.g., via Redirect or RewriteRule).

Part of the document space is the AFS filesystem, including directories writeable by ordinary users. Ordinary users need to be able to determine the Content-Type header line that will be sent by the web server along with the user's file. The user needs to be able to specify this using a .htaccess file, as described at http://www.mit.edu/faq/mimetypes.html.

It needs to be possible to reference web pages under /afs using the tilde syntax. A URL of the form http://www.mit.edu/~foo/bar means that a Hesiod filsys lookup is performed on "foo", and if that filesystem exists and has a top-level www directory, the URL is interpreted by letting that www directory be the replacement for ~foo. Otherwise, a 404 error is produced. Also, in addition to recognizing the character '~' with these semantics, the sequences "%7e" and "%7E" must also be recognized identically.

There are some cases in which URLs differing only by a '~' refer to different pages. For example, http://www.mit.edu/iap is different from http://www.mit.edu/~iap, and http://www.mit.edu/gif is different from http://www.mit.edu/~gif.

These are the single HTML pages whose URLs need to continue working since they are externally referenced (in links from other sites, users' bookmark files, published books, magazines, CDROMs, etc.)

These are the scripts and directories whose URLs need to continue working since they are externally referenced (in links from other sites, users' bookmark files, published books, magazines, CDROMs, etc.) In general, the top-level URL is needed (e.g., http://www.mit.edu/foo) as well as the top-level URL followed by a ? and CGI arguments (e.g., http://www.mit.edu/foo?a=b) and the top-level URL followed by a slash and CGI arguments or filenames (e.g., http://www.mit.edu/foo/bar).