Re: [Idr] I-D Action: draft-ietf-idr-large-community-04.txt

[Job Snijders <[e-mail]>]

On Tue, Oct 25, 2016 at 04:24:38PM -0400, Jeffrey Haas wrote:
> On Tue, Oct 25, 2016 at 07:47:57PM +0000, Jakob Heitz (jheitz) wrote:
> > We added that as a result of Jeff Haas's email attached.
> > 
> > It actually makes sense.
> > However, if we want to keep parity with https://tools.ietf.org/html/rfc7606#section-7.8
> > it's no big deal for me.
> > 
> > Jeff?
> 
> Dave's point on specs is, as usual, right on.  And I think it's generally
> reasonable to default to 7606 behaviors.  (Which would be one or more.)
> 
> My disagreement is more a disagreement with 7606's zero-length behaviors
> with respect to (extended-)communities and the zero length behavior.  In
> that spec, it's a syntax error to receive a length 0 attribute and, as such,
> you should treat-as-withdraw.
> 
> However, that seems to contradict the general intent of the spec which is to
> do error handling behaviors better.  Since 0 is semantically clear (no
> comms), stripping the attribute and continuing seems to be somewhat more
> graceful than treat-as-withdraw.  
> 
> That said, 7606 reflects WG consensus.  It seems wise to keep parity with
> it.

Thank you gentlemen, I've removed the sentence from the document.

Kind regards,

Job