[racket] How does the web server find servlets?

I wanted to know how the Racket web server figures out whether a URL refers
to a servlet or not.  In particular, I was curious about how the
"servlet-root" path in the configuration table was used (its default is
".").  The closest thing I could find to documentation was this bullet point
in the docs for "web-server@":

    Execute servlets in the mapping URLs to the given servlet root directory
under htdocs.

I've read this sentence many times and I still can't parse it...  Is there a
word or two missing or something?

Anyway, after digging through the source code (particularly
web-server-unit.rkt), here's my understanding:  If
[host-root]/[file-root]/[servlet-root]/[url-path] refers to an existing file
whose extension is ".ss", ".scm", ".rkt", or ".rktd", then the file is
loaded and served as a servlet.  (It's actually a bit more complicated,
since it handles extra path elements after the servlet name.)  For example,
if host-root is ".", file-root is "htdocs", servlet-root is "servlets", and
url-path is "foo.rkt", then if "./htdocs/servlets/foo.rkt" exists, it is
served as a servlet, otherwise if "./htdocs/foo.rkt" exists, it is served as
a file.

Is this right?  Can this be documented more clearly somewhere?  (Or is it
already?)

Anyway, is there a way to customize this (e.g. to be able to map directory
URLs to servlets) without having to recreate the entire chain of dispatchers
from web-server@ by hand?

--Doug
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.racket-lang.org/users/archive/attachments/20110901/962797df/attachment.html>