RE: Bitzi File Metadata RDF Dump

> You seem to have created a new URI scheme for bitprints:
>    
> bitprint:3KIZIJB64XP3NCXAE4ISQZT3QNCTF7VDNK5UNR8ZPQ5MFASNGVB5M
> ISV7ESUSB2MN5R3IY2
> 
> AFAIK, this URI scheme is unregistered. Why go to all this 
> trouble when these bitprints already have URIs? I think it'd be 
> much better to just use:
> 
> http://bitzi.com/lookup/3KIZIJB64XP3NCXAE4ISQZT3QNCTF7VDNK5UNR
8ZPQ5MFASNGVB5MISV7ESUSB2MN5R3IY2

I have to respectfully disagree with your recommendation here, Aaron.

(and I'll state explicitly and sincerely in advance that I am not 
simply picking a fight ;-)

What happens if bitzi.com changes it's name, for whatever reason,
or decides that the lookup agent resides under some other domain name,
thus invalidating all those URIs and furthermore possibly raising all
kinds of legal issues with regards to use of the name 'bitzi.com'
in millions (trillions?) of URIs, if e.g. ownership of that domain
name were to change and the new owner forbids such use of their
newly aquired trademark?

And Bitzy will in any case have to employ redirection for all
such HTTP based URIs as they won't be able to employ a basic
filesystem based web server space due to magnitude limitations on
the number of children of a directory. Yes, I know that one can
fully implement HTTP URI retrieval behavior without any direct
mapping of the URL to any filesystem, and in essence treat the
HTTP URL as a domain specific URN, but even so, I would still 
recommend against any identifiers containing linguistically 
significant content that could limit the "business portability" 
or persistent validity of that resource (the DOI folks outline 
very well the key business cases for this). And if you have to 
redirect to resolve the URI anyway, why not benefit from a true 
abstract identifier, such as their own bitprint: URN scheme?

And what if other folks want to license the ability to generate
their own bitprint: URIs (for whatever reason) but wish to have 
the content located outside the bitzy.com domain?

And if there is no distinct URI scheme defined for the identifier
type, how will you define the syntax and semantics of the identifier
scheme and enforce it in an efficient and generic fashion? The
only syntax and semantics that one can reasonably enforce for an
HTTP URL is that defined for the HTTP URL scheme. 

This is exactly why URNs are needed. The identity of a bitprint
has nothing to do with its location, right? It's a name. Not a
location. It's identity need not itself define how to retrieve
its realization (and there could in fact be several different
realizations, no?).

Defining the explicit bitprint: URI scheme allows for identifiers
to be independent of location, ownership, etc. and further allows
the application of constraints and semantics explicitly defined
for instances of that URI scheme. That's the whole point of a URI
scheme, no?, to serve as a form of data type for which there is
an explicitly defined representation and intepretation governing
instances of that data type. Right?

Names are names. Locations are locations. The URN/URL distinction
is valid. Let's not encourage folks to blur it any further, eh?

I encourage Bitzi to register and use their bitprint: URN (URI)
scheme and not resort to further misuse (IMMHO) of the HTTP URI 
scheme for abstract resource names.

Of course, this issue is really outside the scope of this particular
RDF discussion forum -- as insofar as RDF is concerned, a URI is a URI
is a URI is a URI...  it's just an opaque identifier to RDF...

Humblest Regards,

Patrick "my name is not where I live" Stickler

;-)

--
Patrick Stickler                      Phone:  +358 3 356 0209
Senior Research Scientist             Mobile: +358 50 483 9453
Software Technology Laboratory        Fax:    +358 7180 35409
Nokia Research Center                 Video:  +358 3 356 0209 / 4227
Visiokatu 1, 33720 Tampere, Finland   Email:  patrick.stickler@nokia.com

Received on Wednesday, 26 September 2001 11:32:06 UTC