--On 29 January 2008 15:56:44 +0000 Mike Cardwell
<exim-users@???> wrote:
>> RFC2822 does say that you can add other headers, but not arbitrarily
>> named - the name can't clash with a registered name.
>
> I can't find where it says that... It does list a bunch of headers that
> are to be used to store a certain type of information such as "Subject,
> From, To" etc, and I can see why you wouldn't use those headers for
> storing arbitrary data. Eg: "From: Thunderbird Client" would be bad.
>
> But I don't see anywhere where it says I can't use "User-Agent: Whatever
> I want" in email headers.
I didn't say that you can't use it. I said it's non-standard. However,
although I'm unable to find the rfc that suggests it, it's quite common to
use X-something for non standard headers.
It's certainly unwise to use something that might later be registered for a
different purpose.
>
>> However, that doesn't mean
>> that every imaginable header is "Standard". Indeed rfc2076 lists several
>> headers as "not internet standard". I'd suggest that a "standard" header
>> is one that's registered with IANA according to rfc3864:
>> <http://tools.ietf.org/html/rfc3864>. I guess, for clarity, I should
>> have used the phrase "registered".
>
> Hmmm. Can you quote a relevant passage from a relevant RFC that says or
> even mildly suggests that you "MUST NOT" or even "SHOULD NOT" use this
> header for this purpose?
I can't. However, we all use lots of spam detecting heuristics that aren't
justified by RFCs.
>> I suspect that Microsoft are adding spam points to messages with
>> non-standard headers, which would explain why some messages are
>> acceptable when they don't contain a user-agent header.
>
> Email systems add all sorts of unusual headers to emails. I'd suggest
> that it's less likely for a spam to contain a "non-standard" header than
> a ham, not the other way round.
>
> Mike
--
Ian Eiloart
IT Services, University of Sussex
x3148
