Best Practices
==============

a URI is not a file, and the process of dereferencing uris into source
resources shouldn't make assumptions about origin server filesystems
  
some common assumptions to watch out for:

* the filesystem is writable
* directory names are valid python package names

* MVC: controller servlet manages its own subspace
* search engines and query strings
* WebDAV and Web Folders
* preserve semantics of HTTP
* GET and HEAD must be "safe"


Resources 
---------

* http://conveyor.com/RESTwiki/moin.cgi
* http://jaguar.sourceforge.net/webware/urldecode.html
* http://jaguar.sourceforge.net/webware/catalina/
* http://www.ietf.org/rfc/rfc2396.txt
* http://www.namesys.com/

Software
--------

UriMapping_ -- "next generation" URI decoding for Webware

-- TerrelShumway_ - 26 Apr 2002

Why not just have some kind of simple Python-based resource-finding
plug-in function/method which parses paths (URIs or whatever) as they
come in? I can see some advantages in having a function or object
method which takes the raw string and scans it for keywords, for
example. Perhaps everyone considered this in the discussions on the
mailing list which I obviously wasn't following closely enough.

-- PaulBoddie_ - 29 Apr 2002

That is what UriMapping_ is about: the goal is to refactor
Application.serverSidePathForRequest into a separate class with
pluggable behaviors (PPR:StrategyPattern).  URLDecode is one way to do
it.  UriMapping explores many different strategy implementations, none
of which should be considered "cannonical" (current behavior would be
the "default" strategy).

-- TerrelShumway_ - 29 Apr 2002