Re: [apps-discuss] The acct: scheme question

["Paul E. Jones" <[e-mail]>]

John,

 

That's correct, but not a function of the WebFinger draft, but one of RFC
6415.  The URI template accepts URIs, not bits and pieces of URIs.

 

We had discussed long ago to use only "[e-mail]", for example,
but the group decided to use URIs and "acct" was the preferred URI scheme to
refer to user accounts.

 

What I've been doing was trying to document the agreements various folks had
reached on WebFinger.  Don't shoot the messenger.  That said, I don't see a
good reason to backtrack at this point.  The "acct" URI scheme is out in the
wild, its use has a limited scope and specific purpose, etc.

 

If we were to separate them, we would have the question thrust upon us of
"what URI scheme do I use to refer to paulej's Twitter account?"  It's not
mailto.  It should not be http.  I do agree with the group who reached the
consensus before that "acct" is a reasonable way forward.  Nobody was in
love with "acct", but nothing else worked better.

 

Paul

 

From: [e-mail] [mailto:[e-mail]]
On Behalf Of John Bradley
Sent: Tuesday, June 26, 2012 11:12 AM
To: Peter Saint-Andre
Cc: [e-mail]; Murray S. Kucherawy
Subject: Re: [apps-discuss] The acct: scheme question

 

The "resource" parameter is currently a fully qualified URI, and that is
normalized to acct:

 

The template paramater {uri} also precludes relative URI as near as I can
tell.

 

Perhaps Paul can correct me,  but I think that the request.

 

GET /.well-known/host-meta.json?resource=[e-mail] HTTP/1.1
Host: bar.com

 

Is not allowed by the spec, or be interoperable.    The goal of SWD was to
make the above (slightly different syntax same idea) work.

 

There are a lot of places in the spec where the acct: uri and normalizing
things to it are baked in.

 

There are likely also issues with host-meta as that is where the template is
defined.

 

Paul's likely reaction will be that separating them is not trivial, and he
may be correct in that.

 

On the other hand it is probably the right thing to do, even if it touches a
bunch of things.

 

John B.

 

On 2012-06-26, at 10:40 AM, Peter Saint-Andre wrote:





On 6/26/12 8:37 AM, John Bradley wrote:




The current spec requires normalization of bare identifiers i.e. [e-mail]
to acct:[e-mail].

That would also need to change if we separate the specs.


In what way would it need to change?

Peter

-- 
Peter Saint-Andre
https://stpeter.im/