W3C home > Mailing lists > Public > uri@w3.org > February 2010

registries associated with URI schemes

From: Erik Wilde <dret@berkeley.edu>
Date: Tue, 23 Feb 2010 11:34:13 -0800
Message-ID: <4B842DB5.1020702@berkeley.edu>
To: URI <uri@w3.org>

http://tools.ietf.org/html/draft-ietf-geopriv-geo-uri-04 is the most 
recent version of a proposed URI scheme for geolocation. this scheme 
establishes a new IANA registry for scheme parameters in general, as 
well as a specific sub-registry for coordinate reference systems (one 
pre-defined parameter) used for geolocation values. i have two general 
questions about this:

- in general, is there some guideline/policy on how extensibility and 
registries should be designed for URI schemes? the proposed scheme 
allows to dynamically create new sub-registries for parameters which 
take predefined values. this means that the proposed registry structure 
is actually fairly complex.

- specifically, the registry proposed in this draft contains information 
(the geolocation coordinate reference system) that maybe should be 
shared across technologies. the W3C geolocation API, for example, also 
uses a geolocation coordinate reference system. the proposed HTTP 
location protocol 
also uses a geolocation coordinate reference system. is there some 
guideline/policy that tries to avoid such a duplication of information? 
in this specific case, i think the way to go would be to have an ID for 
describing the geolocation concepts, and then have specific technologies 
(such as the URI scheme and the geolocation API) refer to that ID. this 
worked for languages and country codes, but maybe because there was an 
external entity (ISO) managing those lists? does the IETF have a process 
that helps to implement this design pattern without help from the outside?

generally speaking, registries are a good thing to keep things evolvable 
and extensible, but registry sprawl seems to be a problem. and since 
registries typically have no API, it is almost impossible to write code 
that uses registries in a meaningful way. API-enabled registries 
probably are no problem at all (the DNS is more or less that, i would 
argue), but API-less registries might make it hard to implement support 
for URI schemes using this kind of extensibility.

kind regards,

erik wilde   tel:+1-510-6432253 - fax:+1-510-6425814
        dret@berkeley.edu  -  http://dret.net/netdret
        UC Berkeley - School of Information (ISchool)
Received on Tuesday, 23 February 2010 19:34:43 UTC

This archive was generated by hypermail 2.4.0 : Sunday, 10 October 2021 22:17:53 UTC