Re: draft-arkko-roamops-rfc2486bis-02.txt

josh_mendel@infonet.com Fri, 15 October 2004 21:39 UTC

Envelope-to: radiusext-data@psg.com
Delivery-date: Fri, 15 Oct 2004 22:25:58 +0000
To: jari.arkko@piuha.net
Cc: radiusext@ops.ietf.org
Subject: Re: draft-arkko-roamops-rfc2486bis-02.txt
MIME-Version: 1.0
Message-ID: <OF837560DD.AFDA8B64-ON88256F2E.0074B49E-88256F2E.0076F6F5@infonet.com>
From: josh_mendel@infonet.com
Date: Fri, 15 Oct 2004 14:39:20 -0700
Content-Type: multipart/alternative; boundary="=_alternative 0076F6EF88256F2E_="

Hello Jari,

>  Typically what is commonly referred to as the prefix domain, the 
> portion before the '!', is used for the intermediary domains where the 
> suffix domain, portion after the '@', is used as the home domain....
> 
> I think we have had this discussion in the past already,
> and came to the conclusion which is in the current draft...
> 

I have looked through the radext mailings and saw some discussion but 
could not find anything specifically regarding the reason for preferring 
suffix over prefix in determining the routable realm.  I understand it is 
too late to change anything when in last call so just for my edification, 
is it because it was agreed to preserve the designation of the suffix as 
"the realm" from the original RFC?   It is actually standard practice with 
all the large roaming aggregators with which we partner to use the prefix 
domain as priority over the suffix when both exist in the string.  Not 
blaming anyone or group but it is unfortunate that since there was nothing 
in the original RFC with regards to chaining domains that a practice 
developed that is non-compliant.  I understand not wanting to define the 
'/' since it would break the original RFC but unfortunately those roaming 
partners all use it for the prefix delimiter as well.   I guess it is 
fairly common these days to have a standard practice that does not comply 
with the standard and that's life...

Best regards,
Josh