- From: Paul Prescod <paul@prescod.net>
- Date: Tue, 27 May 2003 13:12:43 -0700
- To: Micah Dubinko <MDubinko@cardiff.com>, www-tag@w3.org
Micah Dubinko wrote: > ... > > I think Norm is pointing to an untested assumption here: that developers are > going around pasting Namespace URIs into browser location bars. > > I don't see that happening--instead, they paste Namespace URIs (and error > messages, and method names, and most anything else) into their favorite > search engine and go from there. Okay, let's test that. Google for "urn:schemas-microsoft-com:vml" Now Google for: http://www.w3.org/2001/XMLSchema Now consider these two cases: Now Google for: urn:uuid:92AB0DCB-907D-11D7-9474-0003939B6294 Now Google for: http://www.prescod.net/mynewns Google won't find anything on the last two because I just made them up...but you can imagine how Google could use your request as a hint that it should spider that site and return the information to you when it can. What can it do with the UUID? And of course if Google fails to find something useful what are you going to fall back to? In one case you have a fallback. In the other, you don't. > An additional advantage of the latter approach is that it can locate things > avaliable even over non-HTTP protocols, which would be terribly confusing at > best to point to with http://... Obviously you would not use an HTTP URL to point to documentation available over FTP. You would use an FTP URL. Or a bittorrent URL (which happens to be an HTTP URL anyhow!) or a Jabber URL or whatever. Paul Prescod
Received on Tuesday, 27 May 2003 16:13:07 UTC