On 2/16/2013 2:55 PM, Jeff-Relf.Me wrote:
> Again, the "From:" line should be a website, not an email address.
Not everyone who uses NNTP or email has a website. Very nearly everyone
who does use it has an email address.
> Putting a working email address there is stupid, it'll get spammed.
When I last put a working email address in my From header, it was
spammed at the intolerable rate of about 1 message per day. When I had
exactly 0 working spam filters.
> A non⋅working address is even WORSE, it's crazy spam.
Or you can make the domain-name end in .invalid, which is impossible to
route on the internet.
> The "Message-ID:" line should NOT look like an email address;
> ad it should be human readable, like mine, not just unique.
Human-readability is overrated. No one reads a raw message/rfc822 unless
they're trying to debug a problem with a mail/news client.
> The header should be raw UTF⋅8, not US⋅ASCII + encoded words.
Actually, having futzed with RFC 2047 a lot over the last week [1], I
actually welcome the day that everyone supports RFC 6532. The advantage
of seeing RFC 2047-encoded text is that it indicates that someone is
intelligent enough to give you proper charset data. Without it, you have
to guess what charset people are using, since it's not always UTF-8. As
other messages in this thread very clearly show.
[1] That is why I included the emoji in my From header. I'm curious to
know how reliable non-BMP characters are in the real world, having
discovered that the system I'm working with is too happy to split up a
UTF-16 into two surrogate characters in a nominally UTF-8 stream.
--
Beware of bugs in the above code; I have only proved it correct, not
tried it. -- Donald E. Knuth