>I'm glad there is a simple URL option for the entry point. The chance
>of a beginner searching the FAQ using the two-line URL above is, I
>imagine quite remote... at this stage in Web development, there is
>still a need for URLs to be human-readable and human-understandable.
If people want to enter queries manually, there's a form for that, whose
URL is very simple: <URL:http://www.verity.com/vlibsearch.html>.
Your comments were very helpful. As a result, I found out that most of the
parameters are optional. Ah, learning by doing. Turns out that
<URL:http://www.verity.com:8089/?query=www-FAQ-list> works just fine.
How's that for simple?
>Incidentally, I note that the server on port 8089 does not support HEAD -
>it does a GET (and, in this example, returns a document somewhat past
>it's sell by date ;-)
We're getting out of the HTTPd business, at least temporarily, in favor of
CGI, precisely so that we don't have to worry about anything other than
search. We were concerned about CGI performance, since we have customers
who want to handle hundreds of thousands of users, even millions, but it
turns out that there were fairly easy ways to deal with performance.
The reason for the strange behavior you saw is that our search server
doesn't do the basic HTTPd function -- returning documents from the file
system. It's totally search-oriented. And going away in favor of CGI.
In our next implementation, we might tightly integrate an HTTPd and our
search engine, but we're saving that decision for later. In any case, we
don't want to reinvent HTTPd. We'll surely OEM or buy something.
Nick