RE: Proposed issue: site metadata hook (slight variation)

> -----Original Message-----
> From: ext Miles Sabin [mailto:miles@milessabin.com]
> Sent: 12 February, 2003 13:46
> To: www-tag@w3.org
> Subject: Re: Proposed issue: site metadata hook (slight variation)
> 
> 
> 
> Patrick.Stickler@nokia.com wrote,
> > I'm a little ignorant in this area, but precisely how can a proxy
> > interfere with an MGET request if not explicitly configured to
> > block it?
> 
> Have you actually tried it with a range of deployed proxies? 
> I'd expect 
> you to get back a 400 Bad Request, or maybe a 501 Not Implemented, or 
> perhaps some other random 4XX or 5XX response.

Interesting. I'll have to try that. If what you expect turns out
to be the case, then that would certainly be a non-trivial obsticle
to MGET.

Though I would expect that would nix use of WebDAV through proxies,
and I haven't heard any major cries of woe to that end, so presumably
this is a non-problem.

> And never mind proxies. Can you tell me how I can make an 
> MGET request 
> using, eg. the standard Java HTTP client (HttpURLConnection), 
> or build 
> a Java Servlet which can respond to an MGET? ... hint: I can't. 

Well, just because the Java code is short-sighted and non-extensible
does not mean that the MGET approach is not the most optimal.

Since HTTP already provides a means for servers to tell clients
about bad requests or unimplemented functionality, it seems pretty
stupid to add additional, and non-future-proof constraints
specific to a particular incarnation of HTTP to the Java API.

Fifty lashes with a wet towel to Sun...  ;-)

> And 
> that's not really a limitation in the Java client or Servlet 
> implementations: _you_ might know that MGET has similar network 
> semantics to GET, but I don't really see how any piece of software 
> which doesn't have explicit support for it can be expected to know 
> that.

Ummm... well, I would expect that software *would* have explicit
support for it, i.e. that it would be a semantic web application,
not a web application.

And servers that don't have explicit support for *any* of the
proposals wouldn't work either. Obviously, software is going
to have to be extended to support the semantic web.

Patrick 

Received on Wednesday, 12 February 2003 07:44:06 UTC