On Thu, Mar 26, 2015 at 2:20 PM, Doug Montgomery <[email protected]> wrote:
> That problem seems to be O(1).  As you noted, I did it once, when I created
> the address.   My mail provider may choose to support all kinds of inbound
> variants (UID@domain) that I don't even know about.
>
> I transmit, document, and exchange exactly 1 version of my address.   I
> would like to publish a key for [email protected].   This seems to scale
> quite well.

You do.

Others use aliases for all sorts of reasons.  E.g., joe+dane (some
joe's address for posting to this list, say).

> I consider it undesirable to publish keys for whatever variants my provider
> chooses to support for their own reasons.  Tomorrow they might decide (for
> their own reasons) to equate other transformations of the string.

If you were the sort who likes to use a different sender address
(typically an alias of your primary address) for each list, then you
would consider it desirable.

> Google never has to figure out the problem you propose.  I set my from:
> address.  I would like to publish a key for that.  Users trying to validate
> my signed email will look up the from address I used.  If their user agent
> sees [email protected] and chooses to lookup [email protected]
> .... well, I am more than happy to have that validation fail.

Google does too have to figure out how to canonicalize your aliases:
because they chose to apply a fuzzy matching rule of their own design.
Google did that because they could.

Nico
--

_______________________________________________
dane mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/dane

Reply via email to