Re: [Extra] Is this a plausible IMAP extension ?

[Ned Freed <[e-mail]>]

> There is this thing called BIMI that is being debated elsewhere in the
> IETF.  Leaving aside for the moment the issue of whether it's a good idea
> in the first place, it invents an IMAP feature that seems dodgy to me.

That's not all that's dodgy here.

> When an MTA that supports BIMI delivers a message into the mailstore, it
> adds a header that tells MUAs where to find a logo to show next to the
> message.  (Think of it as x-face for corporations.)  Since bad people
> could phish victims with their own header with a misleading image, BIMI
> invents a new IMAP flag that only the delivery MTA can set on messages
> where it has added a virtuous header.  An MUA can test it to decide
> whether to show the logo.  Other IMAP or POP clients can't set the flag,
> but it presumably stays with the message if it's moved from one folder to
> another.

First, this sounds to me like a layering violation (addition of message content
after delivery) trying to sneak it's way in by pretending to be some sort of
trace information. And since we have the ability to annotate messages, the
first question that needs to be asked is why this isn't being done in the
obvious way by simply annotating the message?

My guess is the answer to that is something along the lines of, "Mumble mumble
not all servers/clients support mumble mumble headers are cheap to fetch
mumble", but really, we're in the process of launching a major new protocol in
the client access space, and surely if there ever was a time to get the
facilities people need in place, it's now.

Second, the issue of who gets to set this information exists independent
of how the information is stored or how access control to that information
is implemented - although I do note that annotations have a rich set of
access control features ready to hand that could be brought to bear here.

And that model seems a bit bizarre to me. Why is the MTA necessarily the agent
with the definitive take on the right logo to attach? Does the MTA never ever
ever make mistakes in selecting a logo? Is the MTA invulnerable to attack? Does
the logo never change after the fact? 

These questions, and many others, need to be answered before we even start
to talk about how to implement this.

				Ned