[ogpx] Fwd: Re: Feedback to draft-hamrick-vwrap-authentication-00.txt

[Meadhbh Hamrick <[e-mail]>]

Sorry. Yes, this was supposed to go to the list.

---------- Forwarded message ----------
From: "Meadhbh Hamrick" <[e-mail]>
Date: Mar 6, 2010 8:18 AM
Subject: Re: [ogpx] Feedback to draft-hamrick-vwrap-authentication-00.txt
To: "Carlo Wood" <[e-mail]>

hey carlo.

i think the intent was that names (first, last and account) were to be
LLSD strings, which are defined in the LLSD / LLIDL draft. at a first
order of approximation, this means they can consist of any unicode
code point (note that this is not precisely true. refer to the type
system draft for more info.)

the rationale behind using a first and last name is to support
existing systems like Second Life and OpenSimulator, both of which
still consume first/last/password in the login sequence. the auth
draft includes an option to log in using an "account" identifier.

i think the idea here is that implementers that currently expect a
first/last/pass that want to use VWRAP to transport the authentication
info would use the agent identifier while implementers that wanted to
use a different ID ( like email address ) would use the account
identifier.

so they're there for two different use cases. agent id is for legacy
systems (and for people who just want to continue using
first/last/pass) and account id is for future implementations that may
want to use a single id.

that these two points weren't obvious probably points to a need to add
some verbiage in the draft discussing the motivation for account vs.
agent identifiers, and to mention (at least in passing) that
identifiers are strings as defined in the type-system draft.

thx for bringing this up, carlo.

-cheers
-meadhbh
--
meadhbh hamrick * it's pronounced "maeve"
@OhMeadhbh * http://meadhbh.org/ * [e-mail]




On Sat, Mar 6, 2010 at 6:26 AM, Carlo Wood <[e-mail]> wrote:
> 2.3.1.  Agent Identifier
>...