|
|
ext Eric Burger wrote:
While the analogy works for text messages, it does not for whisper.
One does not send a "one-shot" packet of voice. One establishes a
voice channel, has a conversation, and disconnects.
Fair enough. I originally suggested that private messages be treated as
strictly an instant messaging specific feature. This was before Seoul, i
think, and back then some people felt it needed to be a more general
feature than that.
That said,
This is why the analogy doesn't hold, and all we're doing by
distinguishing between sidebars and whispers is introducing
complexity with no benefit.
If you had a way to mark your voice packets as "private" (and indicate
to whom they are to be sent) in the same way that, say, a PRIVSEND
method in MSRP would mark the instant message private, *and* the mixer
knew how to handle such voice packets, then private mesages would be
exactly the same in voice as in IM.
But that *is* complex, I agree. So maybe private messages simply make no
sense in voice. Still, you can't draw a conclusion from this that rules
out private messages for all other media.
I can propose some text that points out the limitations of certain media
to enable private messages in the framework draft. We can also drop the
name "whispering" and replace it with "private messaging".
Cheers,
Aki
-----Original Message-----
From: Aki Niemi [mailto:aki.niemi@xxxxxxxxx]
Sent: Thursday, November 11, 2004 10:21 AM
To: ext Eric Burger
Cc: XCON-IETF
Subject: Re: [XCON] barnes-xcon-framework-00 comments
Inline.
ext Eric Burger wrote:
I don't agree. Inline:
-----Original Message-----
From: Aki Niemi [mailto:aki.niemi@xxxxxxxxx]
Sent: Monday, November 08, 2004 5:22 PM
To: ext Cullen Jennings
Cc: Chris Boulton; XCON-IETF
Subject: Re: [XCON] barnes-xcon-framework-00 comments
Inline.
ext Cullen Jennings wrote:
The whisper stuff in this context seems new - seems like
the case where A
and B want to have a private conversation while listening
to the main
conversation should be handled as a sidebar with just two
participants. The
other context of whisper where a announcement server
injects a messages in
whisper mode makes sense to me.
Surely, a sidebar can be used to set up a "private"
conversation. But
imagine such a private sidebar of n participants - there are
scenarios
where any two out of those n would like to send private
notes between
each other. If this requires a one-2-one sidebar established,
this very
quickly means a full mesh of sidebars, which is not very convenient.
What is the difference between a full mesh of sidebars and
a full mesh of
whispers? I would offer that it is identical.
Well, if a whisper entailed setting up a lasting context between the
whisperer and the target yes, but this is not what I'm suggesting.
If you submit to the idea that a sidebar has a "context", and
whispers
are context-less because they are sent *within* the current context,
then they are not at all identical.
Think of this in terms of instant messaging. I have two types of
messages I can send to the conference (chat room): public
ones that get
delivered to all current participants, and private ones that get
delivered to only one or some of those participants. Doing
this in MSRP
is trivial, and does not require sidebars, calls or mucking
with mixer
settings.
This is the main difference between a whisper and a sidebar.
The former
is a private message sent within the existing context
(conference or a
sidebar), whereas the latter is a new context altogether.
So whisper:sidebar::MESSAGE:MSRP ?
No. To use your SAT syntax, whisper:sidebar::PRIVMSG:INVITE. See IRC.
Let's not go there! If you want a private conversation,
why not just "call"
the person? If they have a Conference Aware UA, they most
likely have a
multiple appearance UA that can do local whisper.
So your suggestion is that I need to set up a full mesh of "calls" so
that I can send private messages?
Cheers,
Aki
_______________________________________________
XCON mailing list
XCON@xxxxxxxx
https://www1.ietf.org/mailman/listinfo/xcon
|
|