On 10 Jan 2019 16:07:27 -0500, "John Levine" said:
The theory is that you submit the whole message to Google and it
probes the recipients before it accepts the message, but now you have
the added issue of how to report back that receipient A can handle it
but recipient B cannot.
That will teach me to reply before caffeine. ;) I was discussing the MUA-MSA
interaction. And you'd think that somebody like me would know that stuff :)
You can receive a RCPT TO, and probe before returning a 250/550. You can cheat
a bit if you turn on PIPELINE, then you can accept up to the DATA, and do
probes in parallel with accepting the next RCPT before feeding back all the 2xx
Recipient OK and 5xx Won't Work replies.
That of course just kicks the can back up the road, as the new problem is "OK,
so knowing that, what does the MUA do *now*?"
This is the point where the software engineer looks at the e-mail from the
manager that shows the ship date, and says "Screw it" and puts in code to
just base64 the damned thing :)
I'd really rather the duct tape be applied to make external-body work
Agreed. The last time I looked at this (admittedly quite some time ago), the
biggest missing parts were an RFC-standard way to denote an expiration time
("this URL good until <datestamp>") and a secure way to pass an indication of
what credentials are needed to access the object.
ietf-smtp mailing list