- From: Dan Brickley <danbri@w3.org>
- Date: Wed, 26 Sep 2001 13:27:50 -0400 (EDT)
- To: Peter Crowther <peter.crowther@networkinference.com>
- cc: <www-rdf-interest@w3.org>
On Wed, 26 Sep 2001, Peter Crowther wrote: > [Ever-increasing CC: list snipped] > > From: Dan Brickley [mailto:danbri@w3.org] > > The Bitzi bitprint property looks like an incredibly useful > > one. But the > > thing is to *use* it, not get distracted by premature > > standardisation. If > > it turns out to be as useful as we hope, then doubtless various > > organisations will sing its praises (in prose and in RDF). But > > committee-style endorsement should follow rather than preceed use, > > experimentation and deployment. IMHO much of the RDF design > > is informed by > > this (perhaps unarticulated) attitude, and it is one of the > > technology's characteristic strengths. > > How does one deal with the situation during and after standardisation, where > this incredibly useful property's URI changes? By definition, if it is > incredibly useful, it will have been incredibly used; therefore, any change > to its URI will impact many systems. Unless one can define some form of > bidirectional equivalence, mass prefix changes will cause grief to > developers and users for years afterwards. A fair point. One component here might be a language for making public promises about the quality of service to be expected from certain URIs. For example, I would like to be able to say (in a machine-friendly way) that I promise to manage certain names at xmlns.com in accord with certain policies, that I commit to remembering to pay my DNS fees for all eternity, etc etc. A good/useful/etc property or class name, imho, will be named with care. I'd rather use something like http://purl.org/rss/1.0/channel than, say, a Geocities URL, since the former name is from a set of URIs that an organisation (purl.org / OCLC) have committed to managing with care. (Maybe Geocities have made a similar committment, I don't know, an RDF way of doing so would help me find out...). One aspect of being a good/useful/etc RDF property or class name is not having too much status-of-this-property-or-class metadata burned into its name. As you suggest, needlessly changing URIs is painful. If the world implements to a Schema named: http://example.com/2001/09/some-vocab-proposed-recommendation# ...and then the sponsoring organisation (on basis of enthusiastic feedback from 1000s of implementors) decides the work is fine, it'll only cause distress if they announce the 'finished' version at some new URIs. So one "best practice" criteria we might have for URIs is an expectation that the chosen name for a "we think this may be finished" schema would also suffice for the finished work, ie. http://example.com/not-yet-a-standard.rdf# is a bad name for an RDF vocabulary that you want to urge implementors to sanity test through early adoption. > This doesn't just impact bitprints. I wonder how many otherwise-useful RDF > vocabularies will suffer link-rot if they use URLs rather than URIs*. Error > 404 is both the Web's greatest strength --- allowing distributed authoring > --- and its greatest weakness. Yes. My hope is that the renewed emphasis on URIs that comes with RDF/XML might create a marketplace in which well-managed Web content was rewarded. People who treate URIs carelessly will show up lower in Google etc., be harder to find in directories etc. This is already true to a certain extent, though the message hasn't gotten across too well yet. Dan > > - Peter > > * Both used in this email according to the 'contemporary' view in [1]. > > [1] http://www.w3.org/TR/2001/NOTE-uri-clarification-20010921/ > -- > Peter Crowther, VP Development, Network Inference Limited >
Received on Wednesday, 26 September 2001 13:27:51 UTC