This is an ancient thread, but since you bring it up, Clark, I'm not
sure I understand how you want to see it work.
Can you give some examples of how you would like the path to look, and
On 04 Aug 2001 17:20:11 -0400, Clark C . Evans wrote:
> On 6/7/2000 Chuck Esterbook wrote:
> |At 02:07 PM 6/7/00 +0000, Jay Love wrote:
> | > If you want to add this, my thought for the algorithm is:
> | > At the end of Application.serverSidePathForRequest() (after
> | > finding the > context), we add a "if not os.path.exists(ssPath):"
> | > Then, do the extension testing.
> | > Then, pull off the rightmost item and do the extension test,
> | > until we find the Servlet or get to ``. If we find the servlet,
> | > we add the extra request path info to the HTTPRequest._fields
> | > dict, with a key of "extraPathInfo", or something like that.
> | Sounds fine to me. For increased efficiency, you could do it the lef
> | to right instead. This might tie into our Context/Application
> | discussions, so that an entry in our "directory mapping"
> | dictionary would specify that everything that started with
> | /Bands went to a particular servlet. I think this is how Java
> | does it in fact.
> I'd like to vote for doing it differently than Java, and
> right to left as Jay suggested. Most of the time I have
> N servlets with some shared context, and this context I'd
> like to encode with the extraPathInfo. Doing it left to
> right does not give you much of an advantage over just
> using a query string...
> I don't see how one is more efficient than the other,
> simply grab the last segment, and that is the servlet
> Webware-discuss mailing list