Problem:

Several of us have constructed one kind or another of HTML document that is
the answer to a Frequently Asked Question. For example, we have:

        o http://mole.uvm.edu/~gcd/FAQ
        o http://moose.uvm.edu/cgi-bin/faqs.pl
        o http://mole.uvm.edu/~waw/Printers.html

to name a few repositories.

One problem with this approach is, of course, the lack of centrality of the
information base. Everyone develops their little piece, which is good, but
the pieces do not yet settle into an overall framework. This, I postulate,
is because 1) we have fixated upon unix servers and 2) our unix servers are
rather fixated upon security and privacy.

A second problem to this approach is the difficulty in indexing the
information -- especially when it is distributed over several file systems.

As a *possible* and perhaps temporary solution, I have fixated upon the
following scheme:

I have created a directory on NIRVANA called

        \\Nirvana\Staff\COMMON\DOCS\FAQSHTML

Since this directory lives in the Staff Common directory, just about anyone
on staff and in the Waterman 113 complex has write access to the directory.

I have a commercial httpd server -- O'Reilly's WebSite 1.0g -- installed on
my Dell 560/L Win95 workstation. The server has been instructed to map the
above directory into the URL:

        http://132.198.103.10/CITFAQS/

The WebSite server also comes equiped with a simple but servicable index
and search utility. Access to this utility is via URL:

        http://132.198.103.10/CITFAQS/INDEX.HTM

I have made copies of many of the FAQ files found in the helpline, GCD, and
WAW directories on moose and mole, placed these copies in the NIRVANA
directory, and indexed the files. Thus, searches using the above named
search utility for terms such as "eudora" or "wordperfect" or "listserv"
yield meaningful results.

Restrictions: of course, this is far from perfect. 1) whenever a new
document is added to the directory, the index utility must be rerun. This
is in fact pretty simple, but still a manual process. 2) NIRVANA isn't
smart enough to support long file names via Windows (or is it the other way
around? If it isn't, can we make NIRVANA smarter?) Anyway, files
transplanted from our current UNIX servers may require some reconfiguration
to fit into a DOS-style 8.3 naming framework.

---------

Meanwhile, on another front, I'm working on indexing the rest of the UVM
WWW. More as the situation develops.

---------

Questions and comments are encouraged.



| Wesley Alan Wright <[log in to unmask]>; http://mole.uvm.edu/~waw/    |
| Academic Computing Services      *    *   __0__  *                       |
| Room 238 Waterman Building               / \ | \     *                   |
| University of Vermont          *      *    \77  *                        |
| Burlington, Vermont 05405-0160     *        \\         *                 |
| U.S.A. Voice: (802) 656-1254        *    *    vv      *                  |
|        FAX:   (802) 656-8148 This message copyright 1996 WAW & UVM       |