Re: Registration of acct: as a URI scheme has been requested

On 26 June 2012 01:30, Bjoern Hoehrmann <derhoermi@gmx.net> wrote:

> * Graham Klyne wrote:
> >Thus, while URIs from different schemes may be dereferenced to obtain web
> pages,
> >mailboxes, etc., the acct: scheme is specifically intended to return user
> >account description(s) when dereferenced, and to obtain said
> description(s)
> >using the WebFinger protocol.  No other URI scheme does that.
> >
> >While I might agree that it's not the most compelling candidate for a new
> URI
> >scheme, I think it does meet the expectations for such a scheme, and
> there does
> >appear to be a significant community who want the capability it provides.
>
> My impression is largely similiar. I also think people who think this
> scheme should not be registered should post to <uri-review@ietf.org>.
>

Thanks for the point, I'll follow that discussion.

I think the concerns I have can be summarized as follows:

*disclaimer: these are my personal queries and in no way represent the TAG
or W3C *

- Seems unclear to me whether the whole webfinger community is unilaterally
behind the acct: naming scheme
- Increased implementation complexity
- "Simple Web Discovery" has shown that this problem can be solved without
requiring a new URI scheme
- The same problem can be solved using mailto: without requiring a new URI
scheme
- There seems to be strong calls in the community for user@host to be a
parameter, how will this be handled
- The same problem can be solved using HTTP, without requiring a new URI
scheme
- The same problem can be solved using SPARQL, without requiring a new URI
scheme
- The scheme could equally be a URN
- It will take several years to gain adoption of acct: and initially other
schemes have a bigger network effect
- Developers will get confused between acct:user@host mailto:user@host and
user@host
- Linked data systems will have to have another branch in order to do joins
- If the account whose object is a mail address requires a URI scheme, why
not the user: thing: agent: person: etc.
- Does this open the door for more and more schemes to be registered by the
bigcos, diluting the space
- I'm unsure that the mapping from identifer user@host ->
acct:user@hosthas been documented, for example, one of the motivating
use cases brought
up was for twitter, yet the RFC seems unclear how this would work
- The relationship between mailto: address and acct: address appears not to
be returned in the descriptor document

Having said that, I think that the acct: scheme registration could have
some positive side effects.  Namely that it will get people to think about
URIs and identity, and multiple identity linking.  Perhaps the linked data
(using http URIs to name things and w3c recs) will benefit from some
friendly competition in terms of naming, a kind of democracy of ideas.
Perhaps my biggest concern is that once this precedent is set they may be a
future land rush to create apps, or solve technical problems, using a new
uri scheme, thereby diluting that name space.

So, although I find acct: confusing and perhaps reinventing, in the spirit
of tolerance, I'm not fearful of competition.  HTTP URIs will continue to
benefit from the network effect, and any other scheme will have a chance to
state its case.  Any outcome can have positive implications.  It's up to
the IETF to curate and maintain quality in the namespace as it sees fit.  I
look forward to following discussions there.


> --
> Björn Höhrmann · mailto:bjoern@hoehrmann.de · http://bjoern.hoehrmann.de
> Am Badedeich 7 · Telefon: +49(0)160/4415681 · http://www.bjoernsworld.de
> 25899 Dagebüll · PGP Pub. KeyID: 0xA4357E78 · http://www.websitedev.de/
>
>