Re: [Pce] [Gen-art] Genart last call review of draft-ietf-pce-hierarchy-extensions-10

["Adrian Farrel" <[e-mail]>]

Hi,

Thanks, Dan, for your response here.

Just to follow up on one point:

>> 2. In section 3.2.1 or section 4.1 if the originator sends PCC or PCE
>> sends an open with P flag =0 can the response open be sent with a
>> P flag =1 and if yes what should be the action of the originator. I 
>> did not see any text about this case.
>
> DK>> According to RFC 5440, the Open message is not a request response
> exchange. Both entities send an Open message, and indeed they may even
> overlap. In our H-PCE extensions the P flag indicates that the sender
wants
> the receiver to act as its parent. In the example we give, it should be
fine
> for one party to not set the P flag in the first Open message seen on the
> wire, and for the second part to set the P flag in the second Open message
> seen on the wire. Therefore, I don't think this needs to be address as an
> implementor should be familiar with the Open message behavior. 

Dan is right that the Open is not a request response exchange, but two Open
messages that flow (one in each direction). The second might be sent in
response to the first, or they might be originated independently and be
matched up.

The P flag applies to the sender's requested relationship with the receiver.
It is possible for:
- neither Open to have the P flag set
- both Opens to have the P flag set
- either one of the Opens to have the P flag set
But, if both Opens have the P flag set then the session set-up must fail
according to 3.1.1.
If one speaker doesn't like the setting of the P flag it receives (most
notably if the receiver does not want to act as a parent) it fails the
session according to 4.1.

Thanks,
Adrian