[jdev] voicechat again

Thomas Charron tcharron at ductape.net
Thu Mar 4 11:37:40 CST 2004

> There is nothing wrong with implementing a very flexible,
> not-all-that-specific protocol. Then again, when looked at from a
> users' perspective, there is, indeed, the "overwhelming need" to
> provide a common technical denominator to ensure that the corresponding
> functionality that a user sees is interoperable between as many Jabber
> clients as possible.

  Using JEP-95 and agreeing on some common 'we suggest' mime type, such as
rtp, does this.

> Therefore, IMHO, we must require a Jabber AV client to support at least
> one specified codec to ensure this interoperability.

  Once again, codec has nothing to do with jabber.  We can suggest
options, and in the end, chances are, no client developers are going to
write thier OWN software to do things like RTP.  They will use some
external library to handle this, which will inevitably support existing
standardized codecs such as G.711.

> As long as client programmers are free to choose whatever codec they
> like, you may end up in the all-too-familiar situation that you cannot
> use a certain feature because of purely techno-political reasons. Which
> is fine for geeks who will find some work-around, but it will "spoil
> the fun" for main stream users.

  So the solution is simple.  Just nail down the developers options by
locking down the protocol..  (Yes, I have my tounge FIRMLY planted in my

> Consequently, I'd suggest that beyond the technical details in the
> protocol, Jabber clients and servers will be required to support at
> least one specified codec, although they'd be free to negotiate a
> different, possibly higher-quality one if both/all clients involved in
> the conference support it.

  Again, I fail too see how a jabber server would be required to support
ANY of this.  WHAT part of the servr would have some required change to
support the negotiation of a stream, beyond the obviouse of perhaps
wanting to offer the ability to offer as a gobetween for a raw socket in
the case of boxes behind NATs?  IF this IS the case, then pull back the
wayback machine and visit JEP-03.

More information about the JDev mailing list