From: Steve Dorner
Sent: Wednesday, October 20, 2004 4:00 PM
To: Eli
Cc: Pete Resnick; Matthew Dudziak
Subject: Re: Fwd: RE: Fwd: Bad Mail Headers

At 12:31 PM -0700 10/20/04, Matt Dudziak wrote:
> GO GET the guy who wrote RFC 2822 to READ THIS EMAIL and see what he says.

I have. He says you’re wrong.

I’m not surprised, given that I’ve been involved in drafting things like MIME and 2822, and have had people like Pete Resnick (RFC 2822 editor), Dave Crocker (RFC 822 author) and John Vittal (RFC 733 author) using my software for much of the time.

Doesn’t mean we won’t try to pander to people who misunderstand things this way, but it does mean that, indeed, you may as well stop quoting RFC’s at us. Thank you for your effort, however.

The syntax issue aside, this part is patent nonsense:

> 1) the part after the @ is not HER IP address

There is absolutely no requirement that the LHS of a message-id be an identifier of the client performing the message submission. There are in fact situations where this is impossible, which is one good reason it’s not a requirement.

Now for those who don’t know, at the time of this email, Steve Dorner was the Vice President of Technology at Qualcomm Inc. That means he should be smart – he should be smart and he should be smart about his image to the public. As it turns out, Steve isn’t smart (because he’s *blatantly* wrong – which I will prove below), and he was quite rude – though I will admit I was getting fed up with Qualcomm telling me that I was wrong and their product wasn’t at fault.

Oh, I suppose I should mention this all came from me being a Mail Systems Administrator and trying to deal with a client using Eudora which was not following RFC specifications. The client was very upset that her product wasn’t able to send out email using our systems which she was paying for, and I tried to explain to her that the product had a bug and in turn tried to be nice and inform Qualcomm about the bug so they could fix it since it was pretty major.

Here’s a separate related email I had with other people to ensure that I was reading the RFCs correctly and that I in fact was correct, and Qualcomm was wrong – especially that arrogant Steve Dorner.

From: Scott Perry
Sent: Wednesday, October 20, 2004 6:40 PM
To: Eli
Subject: Eudora and their Message-ID’s with domain-literals not enclosed in “[ ]”

Eudora and their Message-ID’s with domain-literals not enclosed in “[ ]”

This is determined by RFC822 section 4.1, which describes the syntax of E-mail headers. It specifies:

“Message-ID” “:” msg-id

which means that the content of the Message-ID: header is required to be in the format of “msg-id”. The “msg-id” syntax is defined as:

msg-id = “<” addr-spec “>” ; Unique message id

which means that it needs to have “<” followed by the “addr-spec” format followed by “>”. The “addr-spec” syntax is defined as:

addr-spec = local-part “@” domain ; global address

So we now know that the part to the right of the “@” in a Message-ID: header must be in the format of “domain”. “domain”‘s syntax is:

domain = sub-domain *(“.” sub-domain)
sub-domain = domain-ref / domain-literal

where “domain-ref” ends up being the *name* of the computer. Section 6.2.3 permits IPs to be used in place of domains, *if* they use the domain literal format. The domain-literal format is defined as:

domain-literal = “[" *(dtext / quoted-pair) "]”

meaning that there is no question that the brackets are required in the Message-ID: if an IP is used instead of a domain name.

It’s a long, convoluted, and complex path getting to the answer, but it would be very difficult to argue that the brackets are not required when an IP address is used.

I’m not sure what position Scott holds, but he works for Declude Inc. which provides a mail filter for email. He outlined the RFCs perfectly, which was exactly how I had read them as well – so we were on the same channel and clearly proving Qualcomm wrong. However, Qualcomm wanted nothing to do with this problem – especially fixing it it appeared. I was overly appalled that they were so hostile towards me, though I suppose that’s what can happen when your company employs most of the people who write the RFCs for email! Just too bad they’re unable to implement their own work. Pathetic.

Anyways, there’s some fun for today – enjoy it while it’s still here. Always a chance that I’ll get some letter from Qualcomm lawyers telling me to remove this post or something – but in my defense, there were no “tags” on the emails saying it was private and personal, so tough luck!