Re: [apps-discuss] I-D Action: draft-ietf-appsawg-acct-uri-05.txt

["Markus Lanthaler" <[e-mail]>]

On Tuesday, July 02, 2013 7:46 PM, Barry Leiba wrote:
> Yes, the idea is interesting, being able to say, "When I say
> acct:[e-mail], I'm talking about the entity who had that
> account back in 2007.  If the one who has it now is the same entity,
> great.  If not, tell me, and don't do anything else."  It strikes me,
> though, that that's a function of the protocol you're giving the URI
> to, not one of the URI itself.

I think that's really the main question: what does the acct URI identify? In
my opinion it identifies a user (account). There may be different users that
use the same "userpart" over time but that doesn't mean they are the same
users.


> As Paul points out, taking this to the
> mailto: URI (which would be very useful indeed, to be able to say, "If
> this isn't the same user as the one from 2007, don't deliver my email
> message!") would mean that something would have to understand what to
> do with "mailto:[e-mail]?date=20070410".  At the moment, at
> least, nothing does, and there's certainly no way in SMTP to convey
> that.
> 
> A smart email client might be taught to turn that into
> "acct:[e-mail]?date=20070410" and send off a webfinger (or
> web-foot, or whatever) query, and then only do the message submission
> if it got the go-ahead from that.  But in that case, the date-related
> query belongs as part of the web-foot protocol, not as part of the
> URI.

Of course you can work around such limitations by making the protocols
smarter but as history has shown, this is rarely done and leads (at least
sometimes) to massive problems.

Why do you think it belongs in the protocol?


> If you just want to use the URI as a convenient place to store the
> validity date, I don't buy that.

No, I think it goes a bit beyond that.