Comments on draft-ietf-mmusic-sdp-directory-type-00.txt

Colin Perkins <c.perkins@cs.ucl.ac.uk> Tue, 14 March 2000 00:09 UTC

Return-Path: <owner-confctrl>
Received: (from majordom@localhost) by zephyr.isi.edu (8.8.7/8.8.6) id QAA04923 for confctrl-outgoing; Mon, 13 Mar 2000 16:09:17 -0800 (PST)
Received: from gamma.isi.edu (gamma.isi.edu [128.9.144.145]) by zephyr.isi.edu (8.8.7/8.8.6) with ESMTP id QAA04918 for <confctrl@zephyr.isi.edu>; Mon, 13 Mar 2000 16:09:15 -0800 (PST)
Received: from bells.cs.ucl.ac.uk (bells.cs.ucl.ac.uk [128.16.5.31]) by gamma.isi.edu (8.8.7/8.8.6) with SMTP id QAA15611 for <confctrl@isi.edu>; Mon, 13 Mar 2000 16:09:22 -0800 (PST)
Received: from 128.9.160.111 by bells.cs.ucl.ac.uk with Internet SMTP id <g.23084-0@bells.cs.ucl.ac.uk>; Tue, 14 Mar 2000 00:08:23 +0000
Received: from csperkins.demon.co.uk (localhost [127.0.0.1]) by csperkins.demon.co.uk (8.9.3/8.8.7) with ESMTP id VAA08716; Sun, 12 Mar 2000 21:37:37 GMT
Message-Id: <200003122137.VAA08716@csperkins.demon.co.uk>
To: finlayson@live.com
Subject: Comments on draft-ietf-mmusic-sdp-directory-type-00.txt
cc: confctrl@ISI.EDU
Date: Sun, 12 Mar 2000 16:37:37 -0500
From: Colin Perkins <c.perkins@cs.ucl.ac.uk>
Sender: owner-confctrl@zephyr.isi.edu
Precedence: bulk

Ross,

I have a couple of comments on the draft describing the "directory" SDP
media type.

 - When defining RTP payload formats, the combination of the m= and
   a=rtpmap lines in SDP to define a MIME type for the payload format. 
   This may imply that the use of m=directory is referencing a top level 
   MIME directory type, which is presumably not intended?

   I wonder if this link between top-level MIME types and SDP m= lines has
   other effects? For example, SDP defines "data" and "control" as possible
   values for the m= line, which may also not map well onto the MIME space.

 - I'm concerned with the handling of directories which timeout or which
   have exceeded their advertised lifetime. In particular, I am concerned
   by section 7 of the draft, which states that "parties that advertise
   within directories should also participate in advertising the directory
   itself".

   One reason for limiting the lifetime specified in the t= field of the
   directory announcement is that the announcer of that directory knows
   that the address allocated to the directory has a limited lifetime.
   Thus, allowing other participants to announce a directory is useful for
   robustness, but may cause problems if they extend the lifetime of a
   directory without renewing any "lease" on the address used for that
   directory.

   I suggest that text is added to state that sessions announced within a
   directory SHOULD NOT be announced outside the time period specified in
   the description of that directory, and SHOULD NOT be announced after a
   deletion message has been received for the directory. 

Colin