> Let's not forget that interpretations for URL's is up to the web server: 
 >  If a more intelligent naming scheme is desired the server's free to 
 > implement it.  What we're talking about [in the human-readable sense] is 
 > the naive default implementation that most servers put in and relies upon a 
 > file system.
 > 
 > I'm doing server-side java stuff professionally and one of the cool things 
 > about servlets is that they can apply some knowledge about how to interpret 
 > URLs to deliver the right pages.  URL's like 
 > http://foo.com/servlet/SmartServlet/SomethingUseful can be totally 
 > context-dependent on what pages are served up.


Good point.  It still doesn't help if you move te page to a different
server, though, or even if you simply reorganize how your pages are
organized on the given server.  You'd still have the unhappy choice
between abandoning stale links from outside servers, or maintining a
bunch of pointers from the old structure to the new.

But perhaps the second option isn't so bad!  If people checked the
links their pages use on, say, a weekly basis (there are programs to
do this), then providers would only need to put redirects up for a
week or two.  If such a scheme were common practice, then the vast
number of cases of broken links, where the pointed-to document does
still exist somewhere, would be eliminated.

Such a simple utility could be coded up in a day or so, given an HTML
parser.  However, the scheme critically relies on outside servers
putting redirects up when pages move, which is far from common
practice on the Web today.  I don't think most web servers even
provide a convenient way to *specify* redirects, even by hand.  If
there were a way for a given web browser, it again wouldn't be too
hard to come up with a program to automatically generate redirect
links; maybe this program would take a week instead of a day, but it's
feasible and not complicated.


A real chicken and egg problem.  If the Web had not exploded quite so
soon, maybe such a scheme would have been worked out and become common 
practice from day 1.  At least redirects *exist*.  This scheme could
actually be hacked on top of the existing Web....



Lex