Re: [bess] Benoit Claise's Discuss on draft-ietf-l3vpn-acceptown-community-09: (with DISCUSS)

["David Smith (djsmith)" <[e-mail]>]

Adrian>If we had "ACCEPT_OWN handling SHOULD be controlled by configuration, and if controlled by configuration it MUST default to being disabled"...

I agree this may be most optimal. 

/dave

-----Original Message-----
From: Adrian Farrel [mailto:[e-mail]] 
Sent: Thursday, February 05, 2015 8:24 AM
To: Benoit Claise (bclaise); 'The IESG'
Cc: [e-mail]; [e-mail]; [e-mail]; [e-mail]; [e-mail]
Subject: RE: Benoit Claise's Discuss on draft-ietf-l3vpn-acceptown-community-09: (with DISCUSS)

Bit disappointed that this is a Discuss.

But let's discuss it.

> Ron's point, part of the OPS-DIR review, look valid to me. Can we 
> please discuss it.
> 
> This document is well written and well thought out. It is almost ready 
> for publication with one small issue.
> 
> In Section 2.3, the authors say, " ACCEPT_OWN handling SHOULD be
> controlled by configuration, and SHOULD   default to being disabled. IMO,
> they should say, "ACCEPT_OWN handling MUST be controlled by 
> configuration, and MUST default to being disabled."
> 
> AFAIKS, you would never want to build a router where ACCEPT_OWN 
> behavior is always on and cannot be disabled by configuration. 
> Likewise, you would never want to build a router where ACCEPT_OWN behavior is the default.

I reject specifications that control what one might want to build. We produce specs to define interoperable behavior and to ensure the Internet works. We don't legislate for people producing product in niches or that is entirely unsalable.

However, let's separate the two SHOULDs.

Suppose one wanted to build an implementation where the feature is not controlled by configuration and is always disabled?
In that case you would be banned from doing so if "ACCEPT_OWN handling MUST be controlled by configuration", so I would say that "SHOULD" is correct in the first case.

I suspect the second "SHOULD" is a consequence of a compound sentence.
If we had "ACCEPT_OWN handling SHOULD be controlled by configuration, and if controlled by configuration it MUST default to being disabled" then that might be closer to correct according to what Ron is suggesting.

Thanks,
Adrian

PS. Would have helped if the original review had reached the AD, shepherd, and WG. Maybe also the IETF list.