Re: [dbound] [art] [DNSOP] not DNAME, was Related Domains By DNS (RDBD) Draft

[John C Klensin <[e-mail]>]

--On Wednesday, February 27, 2019 20:48 -0500 John R Levine
<[e-mail]> wrote:

> On Wed, 27 Feb 2019, John C Klensin wrote:
>> at least one aspect of it is probably worth mentioning.  It
>> requires understanding "tree" but not any special
>> understanding of IDNs, variants, etc.
> 
> You don't even have to go that for.

Actually, you are identifying cases I was deliberately avoiding,
precisely because fixes have been proposed that might almost
work but that still don't solve, or allow solutions for, the
tree problem.

>  Let's say you do this:
> 
> foo.com. DNAME bar.org.
> 
> Then www.foo.com will be an alias for www.bar.org, but DNAMEs
> only affect names below themselves, so foo.com will remain
> undefined, and you can't put a CNAME and a DNAME on the same
> name.  This makes it useless for what most people imagine they
> want to use it for.

Yes.

> There's the additional issue that an MX with a target of a
> CNAME or DNAME doesn't work reliably,

"doesn't work reliably" may or may not be a synonym for "is
explicitly prohibited by SMTP" but the latter is true is any
event.

>  and the point you made
> that if you've got variants, you can get a very bushy tree
> every time a variant character appears in a label.

Yep.  And that raises several other issues with this proposal,
all of which I was trying to avoid because they are (i) more
complicated and (ii) involve more or less subtle DNS issues.

> This issue has been argued at great length with proposals like
> BNAME and CLONE so let's not redo it here.

Indeed.

    john