On Sun 20/Jul/2025 09:26:34 +0200 Bron Gondwana wrote:
On Sun, Jul 20, 2025, at 00:58, Dave Crocker wrote:
On 6/18/2025 2:15 PM, Bron Gondwana wrote:
But also, did a lot of thinking about how to support multiple RCPT-TO in a
single SMTP transaction.
Bron, Hi.
Simple question: Why?
Simple answer - I spoke to someone from Microsoft about their Enterprise
architecture and the kind of mail usage patterns they see.
There was extensive discussion about multiple recipients here, on the working
group mailing list, where decisions are made for the working group's woirk.
I am pretty sure my reading of that discussions is accurate, which simplifies
to: the number of cases that currently use multiple recipients is vanishingly
small, and so such support is not essential.
*Adding mechanisms that are intended to support vanishingly small portions of
cases -- and especially where that support only provides efficiency rather than
necessary functionality -- is pretty much always a terrible idea for a global
standard.*
It is not just efficiency. Multi-recipients is a feature deeply rooted in
every mail packages. Some of them offer the option to do single recipient
transactions (but might still need to verify if filtering and queuing follow
the same logic.) Some other package simply doesn't have that option, so
requiring it would significantly delay or impede DKIM2 adoption.
True, but we only have a fraction of the world's users on this mailing list,
which is why I've been actively trying to talk with people who aren't here, but
who do chunks of the world's email.
And still, this message and the one preceding it are multi-recipient.
It adds complexity to everyone's code, needs testing and ongoing support, and
gets exercised infrequently enough to make it likely that it won't actually
work when it is needed. That is, it is expensive and fragile.
And then there is the small matter of working group rough consensus that runs
contrary to the pretty-clear rough consensus that I thought I saw before,
/against/ support for multiple addressees in a DKIM signature.
Various threads concluded that single-recipient is only needed for Bcc:, which
includes most cases of forwarding; that is, when the recipient address is not
otherwise mentioned in the (received) header.
I was one of the driving voices behind wanting to simplify things to only have
one address, but I'm persuaded enough by the feedback I've had about corporate
email usage patterns that I feel it's worth investigating how to better support
them.
Best
Ale
--
_______________________________________________
Ietf-dkim mailing list -- [email protected]
To unsubscribe send an email to [email protected]
