Re: Confirmation to advance: draft-ietf-6man-ipv6only-flag-05

[Mikael Abrahamsson <[e-mail]>]

On Sat, 11 May 2019, Brian E Carpenter wrote:

> It's always been my opinion that the flag is input to a heuristic; but 
> from time to time we've had pushback asking for it to be more 
> definitive.

It's been my take that regardless what we write in the document, it's 
going to be treated as a signal and the host will do whatever the device 
manufacturer thinks is best. The "SHOULD NOT" in term of "IPv4 operations" 
leave enough room for interpretation for that.

However, others have interpreted the document differently, so it's obvious 
that the text needs changing. I think we have rough consensus on the flag 
being a signal (much like the M/O flag) that the host can act on if it so 
desires. So if the text can be changed to actually say so, it seems to me 
this is acceptable to a rough majority and we could then progress the 
document.

I am not super happy about leaving the ambiguity out there (see the 
earlier discussion on different OSes doing different things when the M/O 
flag changes) but I think it's the prudent way because we can't predict 
all the use-cases out there, and the different types of devices. The 
implementors know this best, and the document could point out in the 
considerations section some things that have been brought up in the 
discussions so far.

-- 
Mikael Abrahamsson    email: [e-mail]