[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Simple] Comments on draft-ietf-simple-message-sessions-07





Ben Campbell wrote:

On Jul 22, 2004, at 9:38 AM, Paul Kyzivat wrote:

A few more comments, starting with the examples:

11.1 - the numbering of the messages doesn't agree with what is in the diagram. Also, message 5 is missing a blank line between Content-Type and the body.

11.3 - I find the sysadmin example highly unlikely. Why would MSRP (rather than page mode MESSAGE) be used for this? A session would first have to be established. This would only make sense if there was already a session established, or there was expectation that many such messages will be sent using the same session.


OK, would a better example be to have a conference owner say "this conference is about to end"?

Yes, that is a much better one! (Wish I had thought of it.)

12. Extensibility

I believe the decision was that we didn't need version numbers because sessions are established using another protocol (e.g. SIP) which can handle version negotiation. But we don't have any provision for versions there either. The only options I can see that are open for revisions are:

- use a different scheme name (not msrp) in the urls. (ugh)

That is sort of what I was assuming...

Ugh! That would be unpleasant. Do the keepers of scheme names want us using them for version control?


- invent a new a-line to carry versioning information that can be
  negotiated.

Another possibility, if we plan now, is to put a little more extensibility into the msrp url, such as:

      MSRP_urls = msrp-scheme "://" server ["/"
        resource] ";" transport *( ";" msrp-option )
      server = [userinfo "@"] hostport
      msrp-scheme = "msrp" / "msrps"
      resource = 1*unreserved
      transport = "tcp" / token
      msrp-option = gen-param

This would of course require a change to the comparison rules for urls, and could make comparison somewhat more complex.

I think we just have a clear idea of a workable strategy for moving to a revised protocol.

I don't have a strong opinion here. Others?

I'm pretty confident that this protocol will need to be revised.

I also suspect we are going to need extra url parameters for something. In fact, it seems that Rohan has already had in mind the use of parameters to multiplex a single msrp url assigned by a relay for multiple sessions at the client.

If we need url parameters anyway then using one to convey the protocol version is little extra pain.

	Paul


_______________________________________________ Simple mailing list Simple at ietf.org https://www1.ietf.org/mailman/listinfo/simple