[Fwd: Re: [ietf-dkim] canonicalized null body and dkim]

[Michael Thomas]

stephen.farrell at cs.tcd.ie wrote:
>   
>> BTW: we're *waaaaaaay* past last call here. This text has been here for
>> time immemorial.
>> All I propose we do is clarify the meaning of what is actually in the
>> draft now surrounding a
>> corner case. I think the bar should be a lot higher for actually
>> changing the text to mean
>>     
>
> Well, it appears that we've found an interop. glitch, right?
> So the spec has to be changed to clarify/fix that (since we can).
>
> And since different coders have done different things, someone
> needs to change their code as well, once we decide how to fix
> the spec.
>
> Am I missing something?
>
>   
Yes. In my mind there's a substantial difference in clarifying an edge 
condition
where the current text strongly implies one thing with new normative text
to the contrary. The latter requires a lot more justification, IMO -- 
we're way
past last call. Thus far all I've seen by way of justification is 
aesthetics which
is not very good justification, in my mind. The case for keeping it is 
that it
-- intentionally or not -- does the right thing through some mail 
infrastructure
giving a more robust canonicalization. Why would we want to reduce our
robustness, especially at such a late date?

       Mike