Re: [GROW] Discourage asdot+/asdot?

[Job Snijders <[e-mail]>]

On Mon, Aug 06, 2018 at 01:26:41PM -0400, Jeffrey Haas wrote:
> > On Aug 6, 2018, at 8:20 AM, Nick Hilliard <[e-mail]> wrote:
> > 
> > Job Snijders wrote on 06/08/2018 14:22>
> >> RFC 5396 https://tools.ietf.org/html/rfc5396 described the "asdot+"
> >> and "asdot" representation formats for AS numbers.  I'd personally
> >> prefer a single canonical way to represent ASNs (asplain), and
> >> while RFC 5396 proposes the adoption of a decimal value notation
> >> 'asplain', I don't think there is a document actively discouraging
> >> the use of asdot and asdot+ We've come across asdot+ notation in
> >> strange places such as RPSL, and I'm not yet sure how to proceed
> >> https://github.com/irrdnet/irrd4/issues/48
> > 
> > Be liberal in what you accept, and conservative in what you send.

I'm not the biggest fan of that philosophy. Especially because in the
milions of objecst that exist in the combined IRR databases, it appears
only four of them have something with ASDOT in the wrong place.

https://tools.ietf.org/html/draft-uijterwaal-rpsl-4byteas-ext-03 didn't
make it to RFC status.

http://meetings.ripe.net/ripe-55/presentations/jakma-critical-considerations.pdf
is interesting too.

> > I.e. check for asdot on inbound nrtm feeds and convert to asplain,
> > with no option to convert to asdot output.  It's inexplicable why
> > anyone would output or use asdot these days.
> > 
> > IIRC, Afrinic is the only RIR which outputs asdot from their whois
> > server.  Everyone else changed to asplain around 2009.
> 
> And these days, if you want as-dot in your stuff, you should expect to
> throw a knob to have that happen.

Indeed.

Kind regards,

Job