W3C home > Mailing lists > Public > uri@w3.org > October 2003

Re: Announcement: The "info" URI Scheme

From: Sandro Hawke <sandro@w3.org>
Date: Thu, 02 Oct 2003 13:55:25 -0400
Message-Id: <200310021755.h92HtPlH031149@roke.hawke.org>
To: "Eamonn Neylon" <eneylon@manifestsolutions.com>
Cc: "Hammond, Tony (ELSLON)" <T.Hammond@elsevier.com>, "'Eric Hellman'" <eric@openly.com>, "Patrick Stickler" <patrick.stickler@nokia.com>, uri@w3.org


> >>> To sum up, with regard to the info scheme, less is more.
> >>
> >>Thanks, you've just argued wonderfully why the tag: URI scheme
> >>is so desirable.  With info:, there's still a sort of central
> >>authority; you're still trusting the info-registry in some sense,
> >>even if it's not as great as trusting a domain name owner + the
> >>dns.  With tags, you don't have to trust anyone.  Much better.
> 
> By intention tag and info attend to two different problems: "tag:" allows
> anyone to mint identifiers, "info:" allows identifiers that already exist
> under existing namespaces to be referred to using a URI. The notion that
> anyone can create identifiers under the "info:" scheme is wrong. 

(begin pro-tag argument) 

Agreed, but if we had tags, anyone could make the equivalent of info:
or info:isbn: or whatever.  That is, I could hereby define
   tag:sandro@hawke.org,2003:isbn: 
as the prefix for ISBN URIs like info.  And it would work just as
well, wouldn't it?   

Tag has a weakness releative to info: in that it doesnt naturally push
people to use the same URI for the same ISBN, so some outreach or
marketing would be needed to say "Please folks, everywhere, you should
all use the prefix "tag:sandro@hawke.org,2003:isbn:" when you want a
URI to name something which has an ISBN.  And maybe I'd register
isbn2.org to make it seem more neutral and be a little shorter, like
     tag:isbn2.org,2004:0-13-103805-2
even though that wouldn't affect the semantics.

This works for ISBNs as well as staff id numbers at every little
employer in the world, or my numbering system for identifying all my
pets -- why should some of these go to NISO and others be shut out?
Just use tags.


> >>And dereference is very useful.
> 
> Provided someone somewhere is willing, and allowed to, to provide a
> dereference service then, yes, dereference is useful. However, "info:" is
> concerned with getting namespaces relevant to the community of use made
> available as valid URI identifiers. Given that if http URIs exist for these
> things then we would use them, the need for "info:" stems from the
> non-existence of dereferencable representations for these things on the
> Internet.

(pro-http argument)

So it seems easier to you to get the info: scheme approved and
deployed than to deploy a web server?  If people were instead
encouraged to use

   http://niso.org/2003/isbn/0-13-103805-2

then NISO could provide whatever information it thought was helpful to
people who ended up doing an HTTP GET on it.   If it wanted to
mainting a proper ISBN database, it could provide information about
the book at that address; otherwise it could just point to some ISBN
resources and let people figure it out themselves.

So yes, some thought and organizational committment would be required,
but isn't that true anyway?  What other problems are there with this
HTTP approach?

> >>So I haven't been working very hard on evangelizing tags and
> >>getting it done as an RFC.  I'm still willing to be a
> >>co-author because I think it may be useful to some people,
> >>especially folks who would otherwise end up stuck in DOI or
> >>INFO land.
> 
> Really not sure what this means. "info:" is a scheme and not a
> place.

In my slang, "stuck in INFO land" means having invested a lot in using
info: URIs.   (which I'm suggesting isn't a good idea)

> Please don't
> confuse [doi: and info: ] schemes just because myself and Tony have worked on both of
> these activities.

I see some differences, but I think my argument applies to both of
them, so I applied it to both.  Just use http:.  If that doesn't work
because you don't trust the DNS, then use tag:.  What do you want from
info: (or doi:) that you can't get from http: and tag: ?

> >>So if you are into DOI/INFO for the money, ego, or control,
> >>go right ahead with it.  (I guess I'll also grant that info:
> >>saves you a few characters in your URIs.  Tags were a bit
> >>shorter until one of our changes solely intended to appease
> >>the IESG.)  If you just want to separate yourself from the
> >>tyrrany of dereference, why don't you help tag: along instead?
> 
> Actually Tony and I did contact Tim Kindberg a while back to talk about how
> we might benefit from each others experiences. But would you want to be
> tainted with our names ;)

Heh.  :-) I think the ideas matter more than the people behind them,
for things like this.  It's only when the technical arguments stop
making sense that I start to care about the the personal motivations
in the background.

      -- sandro
Received on Thursday, 2 October 2003 13:58:04 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 21:25:06 UTC