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

[Michael Thomas]

Dave Crocker wrote:
>
>
> Michael Thomas wrote:
>> Mark Delany wrote:
>>> I don't think the "lateness" has anything to do with it since there 
>>> are bugs in either direction and no one even noticed until recently. 
>>> It can hardly be the case that it is breaking much to apply the fix.
>>
>> Well, we have a deployment and changing this from the current sense 
>> would be
>>  breaking messages that would have otherwise survived. That is what I
>> recently discovered. Do you have evidence to the contrary with your DKIM
>> deployment?
>
>
> Sorry, Mike, but that particular line of argument isn't applicable 
> here.> Hence,
> it was pure academic exercise.
>
> Working group specs are subject to semantic change up to the point of 
> IESG
> approval. Anyone deploying code based on a spec prior to that moment 
> is taking a
> well-advertised risk.
Huh? I'm saying that changing this is *NOT* academic: there are things 
in the
real world which will cause more message signature to fail if we make this
change. You're not in favor of that are you?

I thought this was the entire point of running code: to find out how the 
spec
works in real life. If that's just an academic exercise, there's 
something seriously
wrong.

       Mike