Yes, "silently" is only related to the PANA session termination.

Here is suggested change on Section 5.8.

Change:

"
   If an error request or answer message is sent unprotected (i.e.,
   without using an AUTH AVP) then the message MUST be processed such
   that the receiver does not change its PANA protocol state.
"

To:

"
   If an error request or answer message is sent protected (i.e., with
   using an AUTH AVP), then the PANA session MUST be silently
   terminated after completion of the error message exchange.

   If an error request or answer message is sent unprotected, then the
   message MUST be processed such that the receiver does not change
   its PANA protocol state.
"

Regards,
Yoshihiro Ohba


On Sat, Apr 14, 2007 at 03:36:50PM +0200, MORAND Lionel RD-CORE-ISS wrote:
>  
> 
> > -----Message d'origine-----
> > De : Yoshihiro Ohba [mailto:[EMAIL PROTECTED] 
> > Envoyé : samedi 14 avril 2007 06:28
> > À : [EMAIL PROTECTED]
> > Objet : [Pana] Error message processing
> > 
> > I discussed with Rafa and found some issue on error message 
> > processing.
> > 
> > The current specification specifies the rule on error message 
> > processing, but it is not clear what is the right protocol 
> > behavior after processing an error message.  The erroneous 
> > request/answer transaction will be terminated after a 
> > protected error message exchange.  But what is the next 
> > behavior?  Repeating the erroneous request/answer transaction 
> > until session times out is obviously not a good approach.  I 
> > think that a simple behavior is to silently terminate the 
> > PANA session immediately after any protected error message exchange.
> 
> Agreed. I assume that the "silently" is only related to the PANA session 
> termination, because the PANA peer will have to anyway take into account the 
> error notification ;)
> 
> > 
> > Note that if an error message is non-protected, the following 
> > rule described in Section 5.8 still applies:
> > 
> >    If an error request or answer message is sent unprotected (i.e.,
> >    without using an AUTH AVP) then the message MUST be processed such
> >    that the receiver does not change its PANA protocol state.
> > 
> > Regards,
> > Yoshihiro Ohba
> > 
> > 
> > _______________________________________________
> > Pana mailing list
> > [EMAIL PROTECTED]
> > https://www1.ietf.org/mailman/listinfo/pana
> > 
> 
> 

_______________________________________________
Pana mailing list
[EMAIL PROTECTED]
https://www1.ietf.org/mailman/listinfo/pana

Reply via email to