- From: <noah_mendelsohn@us.ibm.com>
- Date: Wed, 3 Nov 2004 19:16:21 -0500
- To: <Patrick.Stickler@nokia.com>
- Cc: chris@w3.org, GK@ninebynine.org, joshuaa@microsoft.com, skw@hp.com, timbl@w3.org, www-tag@w3.org
Patrick Stickler writes: >> The idea of having (preferably recognizable) "micro >> parsing schemes" within http: URIs is interesting >> and something I have to think about a bit. I don't think I proposed that either. IMO the Web Arch document has it about right on URI opacity [1]. URI's may have regular structure, but don't look in there except for structure actually licensed by the specifications and/or owner of the URI, and even then only when necessary. Nonetheless, I think that any reasonable URI allocation approach that covers the values in XML Schema types is likely to have a degree of regularity, e.g.: http://www.w3.org/2004/SchemaSimpleTypes/Integer/12 http://www.w3.org/2004/SchemaSimpleTypes/Integer/13 W3C date space has essentially the same sort of regularity . Consider as a handy example, the URI quoted below at [1]. The URI is hierarchical, per RFC 2396, and if you ask the W3C they might tell you some things about how they're using the hierarchy. For most purposes, it's good practice to treat that URI as opaque. Still, W3C will tell you that this is probably a name first created around 2004, and that it's a name intended to be used for Technical Reports. There are circumstances in which it's appropriate, with care, to rely on such structure, but as the Web Arch document warns you, there are consequences when you do. Same for the integer URI's proposed above. I don't think it's in all cases a mistake to cons up the string http://www.w3.org/2004/SchemaSimpleTypes/Integer/13579 to represent the integer resource 13579, providing W3C has told you that all of the URIs of form http://www.w3.org/2004/SchemaSimpleTypes/Integer/* have been assigned for use in this manner. Still, you should isolate such dependencies to the extent possible. The Web itself, RDF, etc. will and should treat this as an opaque resource identifier. FWIW, one might well argue that URIs of the form: http://www.w3.org/2004/SchemaSimpleTypes?typeName="Integer"+value="12" could be in the mix for consideration. I have no strong opinion. Certainly user agents are regularly aware of the structure of the parameters to such URI query strings. Long ago Dan suggested a target of <=3 contributions per thread per person on a topic [2], unless there's specification text that's being productively refined. To paraphrase James Bond [3], "I've had my three", so I don't expect to follow up on this thread any further. The point was that we might consider URIs for members of the XML Schema Simple types. This thread has safely logged that proposal and your responses. I suggest we let it go until such time as it might prove useful to some other workgroup. Noah [1] http://www.w3.org/TR/2004/WD-webarch-20040816/#uri-opacity [2] http://lists.w3.org/Archives/Public/www-tag/2002Mar/0147 [3] http://www.mi6.co.uk/sections/movies/dn_quotes.php3?t=dn&s=dn -------------------------------------- Noah Mendelsohn IBM Corporation One Rogers Street Cambridge, MA 02142 1-617-693-4036 --------------------------------------
Received on Thursday, 4 November 2004 00:17:29 UTC