Hi,

Mark Andrews via RT wrote:
> DNS developers testing ambigious corner cases will do this but no
> one else ever will.

Better to test the heck out of the ambiguous corner cases than not,
IMHO.

I also happen to think that ambiguous corner cases shouldn't cause zone
transfer failures...

> The RFC itself needs to be fixed.  It is self contradictory.
> 
> One of the following needs to be changed.
> 
>       Tag values MAY contain US-ASCII characters 'a' through 'z', 'A'
>       through 'Z', and the numbers 0 through 9.  Tag values SHOULD NOT
>       contain any other characters.  Matching of tag values is case
>       insensitive.
> 
> or 
> 
>    Tag:  Is a non-zero sequence of US-ASCII letters and numbers in lower
>       case.

I actually don't think these two excerpts are contradictory at all. §5.1
("Syntax") is describing the wire format, while §5.1.1 ("Canonical
Presentation Format") is describing the presentation format. The
reasoning in the rejection of erratum 4061 explicitly says that they can
differ.

Now, I disagree with the reasoning in the rejection of erratum 4061 and
it smells like a post-hoc rationalization for a dumb mistake in the spec
to me, and I think RRTYPE specifications should be required to have
1-to-1 convertibility between wire formats and canonical presentation
formats (i.e., without resorting to generic RFC 3597 presentation
format). I don't know of any text anywhere that explicitly requires that
currently, though.

-- 
Robert Edmonds