[Standards-JIG] PROPOSAL for Live Chat

Justin Karneges justin-keyword-jabber.093179 at affinix.com
Sat Jan 6 19:41:53 UTC 2007


On Saturday 06 January 2007 2:09 am, Remko Tronçon wrote:
> > Better, I think, for us to define a separate protocol for this. Should
> > be pretty straightforward. Anyone want to write a proposal for it?
>
> Personally, I wouldn't care too much about saving traffic in the case
> where characters are inserted or deleted in the middle of the text. I
> would therefore have 2 modes in this protocol: append mode, where you
> send one character to be appended (or deleted) from the tail of the
> message, and replace mode, where you send the whole message that
> replaces the current one (used when appending doesn't apply). This is
> easy to implement in both sending and receiving clients, should make
> most use cases 'efficient' in traffic, and avoids illegal offsets etc.
> Stream compression will take care of a lot of the redundancy in the
> 'replace' mode. As a matter of fact, the 'append' mode might not even
> save a lot of traffic over stream compression. The downside of
> 'replace' mode is that it might involve a lot of redrawing, but then
> again, I don't expect many people editing their messages in the
> middle.

So far the discussion seems to assume that you are composing a message that 
eventually is finalized.  But is this true?  The ICQ live chat, which prompts 
this discussion, has no end to messages.  There is no way to indicate a 
message is complete ('enter' simply sends a newline character, which is just 
a character like any other), and you can backspace 3 hours of conversation if 
you want.

I'm not saying this is how we should do it, but I figured it should be said.

-Justin



More information about the Standards mailing list