- From: Kingsley Idehen <kidehen@openlinksw.com>
- Date: Mon, 20 Jul 2009 10:55:19 +0100
- To: Dan Brickley <danbri@danbri.org>
- CC: Danny Ayers <danny.ayers@gmail.com>, public-lod@w3.org
Dan Brickley wrote: > On 20/7/09 11:01, Danny Ayers wrote: >> "Second Life objects to become HTTP-aware" : >> >> http://www.massively.com/2009/07/08/second-life-objects-to-become-http-aware/ >> >> >> cool, right? well not exactly, it uses shortlived-by-design URIs: >> >> http://wiki.secondlife.com/wiki/LSL_http_server > > Well, we can't have it both ways. > > Either we want everything of interest to have HTTP URIs. > > Or we want all HTTP URIs to de-reference usefully forever. > > But we won't easily get eternally-useful http URIs for everything > useful that has ever been plugged into the 'net. Yes, ultimately we really have to accept that utility exists in all forms of data access: 1. By Reference (i.e., HTTP URIs) 2. By Value (i.e. Typed and Untyped Literal attribute values) Nothing wrong with using Full Text patterns (or other methods) to locate values, and then from those values locate the current HTTP URIs for optimized graph traversal (e.g. finding related stuff). There are no silver bullets, just lots of options for HTTP based data access and data meshing :-) > > Anyone building systems that assume otherwise is building something > rather fragile. There are a *lot* of data objects in secondlife... > Yes, ditto all the other lifes since "everything is data" when all is said and done :-) Kingsley > cheers, > > Dan > > -- Regards, Kingsley Idehen Weblog: http://www.openlinksw.com/blog/~kidehen President & CEO OpenLink Software Web: http://www.openlinksw.com
Received on Monday, 20 July 2009 09:56:05 UTC