Thank you for the response.  More comments inline

> -----Original Message-----
> From: Jonathan Rosenberg [mailto:[EMAIL PROTECTED]]
> Sent: Thursday, June 06, 2002 2:19 AM
> To: [EMAIL PROTECTED]
> Cc: Sipimp (E-mail)
> Subject: Re: [Sip-implementors] DATE field and REGISTER EXPIRES field
>
>
> inline.
>
> Ken Jordan wrote:
> > Since the DATE field provided in the register response is
> useful to set
> > the
> > UAC's time (per 10.2.5)...
> >
> > If a DATE header field is included in the clients REGISTER
> request, does
> > the
> > registrar take into
> > account the transmission time between the UAC sending the
> registration
> > and
> > the registrar receiving
> > it (expires - (current time - date header time))?
>
> You could try, but generally no. I see no value in that.
>
> >  That is, if the
> > transmission time is 30 seconds and
> > the expires field was 300 seconds,
>
> Thats a pretty screwed up system. In any reasonably well designed
> nework, the registration interval should far exceed the transmission
> latency by several orders of magnitude.
>

Registration times could be fairly short (on the order of a few minutes).
The protocol leaves it up to the registrar to impose a minimum value for the
expires field.  For some applications, a smaller expires time may be
acceptable.  For example, an agent returns to their desk at 4:57pm and
registers their presence as being the next 3 minutes.

Agreed that it is an undesirable network.  Unfortunately customers do not
always have optimum networks.  I'm trying to understand the timeout values
allowed in the protocol when compared to shorter registration intervals.  I
understand that it should work if the network were working properly.  I'm
considering the other side when customers do not always have perfect
networks.  Consider what happens as transmission times approach the timeout
values (yet the message still makes it through).

What are suggested/typical registration timeout values?

As transmission times approach the timeout values, can the intended expires
value (from the clients perspective) be off by as much as that timeout
value?

I'm not looking to change the protocol for worse case scenarios.  However I
do want to understand the intended behavior of registrars given these
conditions.  I would like to know that while this situation is undesirable
the protocol is still able to handle it.

I appreciate your help.

>
> A couple of questions related to
> > the
> > UACs handling of the REGISTER
> > response.
> >
> > 1.  Is the expires value in the response the seconds
> remaining from the
> > original registration til now?
> >      Or is it simply the original expires value?  E.g. REGISTER at
> > 10:00:00
> > with expires=3600.
> >      Fetching the bindings 2400 seconds later would yield
> 1200 or 3600
> > for
> > the expires value?
>
> 1200.
>
> >
> > 2.  Should the UAC use the DATE field (the registrars time) when
> > evaluating
> > the actual time left in the
> >      expires field (similar to what was asked for the
> registrar above)?
>
> No.
>
> -Jonathan R.
>
>
> --
> Jonathan D. Rosenberg, Ph.D.            72 Eagle Rock Avenue
> Chief Scientist                         First Floor
> dynamicsoft                             East Hanover, NJ 07936
> [EMAIL PROTECTED]                 FAX: (973) 952-5050
> http://www.jdrosen.net                  PH:  (973) 952-5000
> http://www.dynamicsoft.com
>

_______________________________________________
Sip-implementors mailing list
[EMAIL PROTECTED]
http://lists.cs.columbia.edu/mailman/listinfo/sip-implementors

Reply via email to