[ietf-dkim] LWSP in base64-encoded public key TXT RR

[Dave Crocker]

Eric Allman wrote:
  > For the time being I agree with Paul's suggestion: live with it. This is
> the sort of thing that can be solved in DKIMbis, which I'm sure will be 
> required before DKIM gets to full standard.


I'll suggest something even stronger:

Strictly speaking, this is not a bug, so "fixing" it is not the issue.

Why is is not a bug?  Well, while we might now wish we had specified things a 
bit differently, the specification in its current form works and it does not 
impose undue burdens on implementors or operators.  So the nature of what is 
there, that we might wish were different, falls into the realm of preference 
or taste.

Hence, the way to "fix" the incompatibility being found in the field is to 
change the implementation, not the specification.

As Paul notes, although documenting this issue might be useful, it is not 
"erratum". (And, Eric, that is quite different from either of the alternatives 
you had, in responding to Paul.)  The documentation should not declare what 
was intended, but merely highlight something that exists and that we have some 
basis for knowing is worth flagging because it appears to be easy to 
misinterpret.

Stephen's suggested opening text for the note about this strikes me as exactly 
the right perspective.  Here is what is, rather than here is what we wish, or 
here is a bug in the spec, or any such thing.

A small change from Stephen's text that I would suggest is to not refer to 
"since the spec was completed or approved" but merely to state a clarification 
of what is in the spec.

So, wherever this gets documented, it should say something like:

      "NOTE:  The specification permits folded white space (FWS) in the DKIM 
DNS TXT record.  FWS requires that a <CR><LF> be followed by at least one 
linear white space (LWSP) character.  Implementors and administrators are 
cautioned to be careful to ensure that the TXT records produced conform to the 
specification."


Yes, we might later choose to "enhance" the specification, to allow the case 
that has appeared in the field, but then, we might not.

d/
-- 

   Dave Crocker
   Brandenburg InternetWorking
   bbiw.net