- From: <Patrick.Stickler@nokia.com>
- Date: Sat, 6 Oct 2001 01:23:17 +0300
- To: sean@mysterylights.com, www-rdf-logic@w3c.org
> That seems fair enough. Don't forget that there's already a > URI scheme for > literals, "data:". Instead of str:U0043, why not just use:- > > data:text/plain;UTF-8,a Hmmm... well for "structured" data typed literals, insofar they are defined as MIME content types, yes. But the data: URI scheme does not IMO really provide for "fundamental" data types such as integer, float, character, etc. It's useful, and I'm not saying I don't like it, etc. but it doesn't really address the full (or even primary) scope of literal data types. Perhaps it could, if MIME content types existed for those data types, but those would IMO be very odd MIME content types, and that's likely not at all what the authors of the data: URI scheme had in mind specifically... > I also think it exposes a certain amount of nitpicky > pedanticism behind the > current rdfs:Literal threads... Maybe to some extent, yes -- but sometimes pushing in what is considered an important direction can result in a little too much pushing... (though not necessarily ;-) It all boils down to (IMO) to what degree you need strong data typing. There are lots of applications where any old string will do -- or perhaps those strings are produced by some other non-RDF system that employs strong data typing so you don't realize that the actual RDF encoded data is "unconstrained" even if it resembles highly constrained data. RDF defined in terms of "human consumption" will seldom care a nit for the data type of literals. eCommerce applications, and/or e.g. frameworks for managing billions or trillions of digital objects where each object has financial and/or legal significance, will place alot higher importance on the type of literals used, if literals are even allowed to be used at all. One of the highest items on my own personal wish list for RDF based applications has for a long time been some standardized solution to strong data typing. Not just some solution provided by a particular tool or always something that is home grown, but something that I can rely on when I send out lots of data into the SW and can expect that most, if not all, SW agents will "do the right thing" and know exactly what I mean -- and which maximizes the utility and investment put into software components that rely on knowing just what kind of literal is being dealt with at a given point in time. This is really the motivation for proposals like URI encoded typed data literals. I'm just searching for some mechanism (that is standardized) which, within the context of an RDF application, allows me to know what I am dealing with precisely (or at a greater level of precision than just some literal "string"). Granted, any standards endeavor is evolutionary, and one cannot expect everything one wants as quickly as one needs it, but the (apparently notable) nerve that is struck with regards to the role or existance of literals and how they relate to the encoding of (useful) knowledge may be indicative of something that really needs to be addressed in the immediate future, if not in the present work being done by the RDF core working group. OK, enough rambling for tonight (or, rather for today, for you folks in the new world).... crawling back into my hole... Patrick -- Patrick Stickler Phone: +358 3 356 0209 Senior Research Scientist Mobile: +358 50 483 9453 Nokia Research Center Fax: +358 7180 35409 Visiokatu 1, 33720 Tampere, Finland Email: patrick.stickler@nokia.com
Received on Friday, 5 October 2001 18:45:24 UTC