[rfc-i] issue: canonical formats

Heather Flanagan (RFC Series Editor) rse at rfc-editor.org
Mon Jun 4 14:06:19 PDT 2012


On 6/1/12 8:26 AM, John R Levine wrote:
>>> The input format is definitive.
>>
>> I think that really only works if the input format is also the canonical
>> output format, so it gets seen, checked, and loudly mirrored.
> 
> We don't seem to have much trouble with errors in xml2rfc drafts.
> Mechanical checks can be much more extensive than for line printer
> format, and there's nothing of importance in one that won't affect what
> you see in the output formats.
> 
>> I do think we need a canonical output to support some of the legal use
>> cases.
> 
> Please can we not play junior lawyer?  My understanding is that the
> legal uses entirely boil down to a lawyer or court saying "is this a
> copy of RFC 9999?" and someone who can speak for the IETF saying yes. 
> That doesn't need a canonical output format, just the ability to say
> whether the contents of a document match the contents of the canonical
> one, and no court I've ever dealt with cared about the difference
> between a copy of a document in different layouts or typefaces. But
> neither of us really knows.
> 

I've checked in with the legal committee for the IETF, to get some
clarity and closure on this particular issue.

Just to be a bit more clear on this one, my intent when I recorded that
requirement was that, based on my experience, the legal process doesn't
care what the format is, exactly, as long as it has been documented
prior to being asked that a) we have a canonical format, and b) we have
or can automate the ability to confirm that a document someone hands us
exactly matches what was originally published. I've followed up on this
with IETF's legal committee to clarify they indicate that in all
circumstances, lawyers will print out copies of documents for review in
court.  Printouts are likely to be predominant until the world runs out
of trees.  Canonical format generally does NOT enter in to the discussion.

So, this turns out to be not as significant issue as I had originally
thought.  John, you are absolutely correct.  Thank you for posting!

-Heather Flanagan





More information about the rfc-interest mailing list