Re: #1132: Allowing IP addresses as path-identiites - Poll result

["Charles Lindsey" <chl@xxxxxxxxxxxxxxxx>]

In <43264503.248B@xxxxxxxxxxxxxxxxx> Frank Ellermann <nobody@xxxxxxxxxxxxxxxxx> writes:

>Charles Lindsey wrote:

>>    path-identity = dot-atom-text / bareword

>Hi, you can't say <dot-atom-text> here, it contains some
>characters we don't like (delimiters for 1036 incl. "!"):

Yes, you are correct. I was following the over-liberal RFC 2822, rather
than the down-to-earth RFC 2821. And even RFC 2821 allows things that RFC
1035 doesn't, such as subdomains starting with a digit.														


>Let's stick to some kind of <ldh-string>.

>Yes.  For a "bare" version of <ldh-string> you'd get...

>     ldh-fqdn = ALPHA *( [ "-" / "." ] (ALPHA / DIGIT) )

OK, I have now s/dot-atom-text/fqdn/ everywhere, and I have

   fqdn = 1*( label "." ) label 
   label = ALPHA *( [ "-" / "." ] (ALPHA / DIGIT) ) 


>With that version of <label> there's still an unnecessary
>ambiguity with <IPv4address>, it can be solved by <toplabel>:

No, there is no need for that. Every <label> has to begin with ALPHA (and
that was in RFC 1035), so if you see anything with just digits and "."s
in it, it is either an <IPv4address> or it is a syntax error.



>>    tail-entry = dot-atom-text / bareword

>No <dot-atom-text>, just say tail-entry = path-identity

No, because I want to be able to say, in USEPRO, that if you see any
<path-identity> that is the name of one of your peers, then you don't
relay it to that peer. But that only works if there is no <path-identity>
inside a <tail-entry>.

>Dito, source-identity = path-identity / IP-address

>Actually I don't see the point of bareword here, is that for
>POSTED and MISMATCH in this variant ?  You could enumerate it:

No, you still need <bareword> as a possibility for <source-identity>,
because the article may have arrived by UUCP, or it may have arrived from
an uninteresting IP address leased to a dialup, and what you want to quote
is the "true source of the article" which you only knew because the
sending site authenticted itself with SASL, using some <bareword> as its
identity.


>>    IP-address = IPv4address / IPv6address ;  see [RFC 3986]

>Okay.  The latter has always a colon, so that's unambiguous

I think we have to accept that this grammar is always going to be
ambiguous (but mostly it doesn't matter). Harald's grammar is even more
ambiguous.

>> [2nd version of syntax, making <source-identity>s readily
>> distinguishable from <path-identity>s.]
>[...]


>The third case is wrong, you want an "!" at the end, always.
>The second case is also wrong, missing "!" before the source.

No, I want a "!" at the end never, except when I am trying to construct
the "!!" case. And it is actually wrong in both versions of my syntax. Try
this:

[1st version]
   path-entry = path-identity *( "!" path-identity ) [ "!" "POSTED" ]
                  [ "!" [ source-identity [ "!" "MISMATCH" ] ] ]
[2nd version]
   path-entry = path-identity *( "!" path-identity )[ "!" "POSTED" ]
                   [ "!" [ path-source ] ]

>>  Two "!"s in succession are equivelent to
>> "!site.example.MATCH!"

or a missing <path-source> (which gives rise to "!!") means it MATCHed.

Note also that my syntax severely limited where [FWS] was allowed
(essentially between things added by one site). If you want to relax that,
then I shall have to take care it does not arise inside a "!!".

-- 
Charles H. Lindsey ---------At Home, doing my own thing------------------------
Tel: +44 161 436 6131 Fax: +44 161 436 6133   Web: http://www.cs.man.ac.uk/~chl
Email: chl@xxxxxxxxxxxxxxxx      Snail: 5 Clerewood Ave, CHEADLE, SK8 3JU, U.K.
PGP: 2C15F1A9      Fingerprint: 73 6D C2 51 93 A0 01 E7 65 E8 64 7E 14 A4 AB A5