Alex van den Bogaerdt wrote:
Section 2.3.5 defines domain to be one or more
dot-separated components (note "one or more")
then refers to section >>>5<<< (not 4) and [22]
"5" is the "MX how to". The individual commands
and their syntax are in "4", including this "one
dot" trick to catch non-FQDNs.
It's announcded in 2.4 "General Syntax Principles":
| The syntax for each command is shown with the discussion of
| that command. Common elements and parameters are shown in
| section 4.1.2.
[[22] = 1035]
<domain> ::= <subdomain> | " "
<subdomain> ::= <label> | <subdomain> "." <label>
<label> ::= <letter> [ [ <ldh-str> ] <let-dig> ]
The "start with a letter" bit was apparently dropped
for or ignored by some host labels (?) But kept for
TLDs, Wayne can still use it in his "toplabel" syntax.
rfc2821 seems to contradict itself by formalizing
a syntax it didn't describe, pointing to a formal
syntax it _does_ describe.
Yes, that's why I asked the author about it, because
I thought it might be a typo for the errata. But it
was intentional to catch simple host names (no FQDN).
I can't place "you are welcome to repost as needed"
A permission to post what he said in his mail. This
issue was discussed in USEFOR among others. What he
proposed was to update it by allowing a trailing dot
for cases like 'host AI' (host TV does not more exist).
if that is about the rfc, it seems to me a bug is
acknowledged.
More like an "intentional bug" aka "feature". We do
similar things here, like saying that everybody is
free to use a proper HELO FQDN with a simple policy
"v=spf1 a -all" for this FQDN.
There's nothing random about helo names.
I've no problem to follow you here, but better say
"EHLO names", because 2821 explicitly refers to 821
for the case of HELO, and that syntax is a bit more
ugly (far from "random" of course). Bye, Frank