Re: [Idr] New Version Notification for draft-ietf-idr-large-community-11.txt

[David Freedman <[e-mail]>]

just had an offlist conversation with job, and he has recommended that I elaborate on this point as it may be non obvious to some.

The GA is intended to be an ASN, but IMHO AS_TRANS is not an ASN, it is a signal between peers in an OPEN exchange and a signal between hops that information is missing in some way, it is not an ASN in the sense that it should (or perhaps now even *could*) be used by anybody as a local ASN for a router.

As such I'm aware of code that exists which discriminates against AS_TRANS, there are various libraries which handle data structures like ASNs (which could be used in , say , creating or validating a GA field) which would be very unhappy if fed AS_TRANS as their initial input.

I understand we want to progress the draft , and that and a line has to be drawn between what could be unwise in implementations , vs what could be unwise to be seen on the internet , and that a registry can exist for the latter , but this is something I feel strongly should live on the former side , and if a list of values is going to be specified in the draft , I feel this should be in it (irony aside).

Dave 

> On 3 Dec 2016, at 10:14, David Freedman <[e-mail]> wrote:
> 
> last minute nit: if we are going to recommend that particular reserved ASNs are avoided for the global administrator, can we please include AS_TRANS (RFC6793), there is probably a certain irony in not doing so.
> 
> Dave