Re: [usefor] Issue with ABNF grammar in RFC 5536

[Michael Bäuerle <[e-mail]>]

Julien ÉLIE wrote:
> Michael Bäuerle wrote:
> > Julien ÉLIE wrote:
> > > 
> > > Paul Kyzivat noticed in the IETF Gen-ART mailing-list
> > >     https://www.ietf.org/mail-archive/web/gen-art/current/msg14921.html
> > > that there is an issue in the ABNF of RFC 5536 (USEFOR).
> > >
> > >      The following news header fields extend those defined in Section 3.6
> > >      of [RFC5322]:
> > >
> > >        fields        =/ *( approved /
> > >                            archive /
> > >                            control /
> > >                            distribution /
> > >                            expires /
> > >                            followup-to /
> > >                            injection-date /
> > >                            injection-info /
> > >                            lines /
> > >                            newsgroups /
> > >                            organization /
> > >                            path /
> > >                            summary /
> > >                            supersedes /
> > >                            user-agent /
> > >                            xref )
> > >
> > > to:
> > >
> > >      The following news header fields are defined by this document:
> > >
> > >        news-fields    =    approved /
> > >                            archive /
> > >                            control /
> > >                            distribution /
> > >                            expires /
> > >                            followup-to /
> > >                            injection-date /
> > >                            injection-info /
> > >                            lines /
> > >                            newsgroups /
> > >                            organization /
> > >                            path /
> > >                            summary /
> > >                            supersedes /
> > >                            user-agent /
> > >                            xref
> > >
> > >      The definition of message in Section 3.5 of [RFC5322] is changed to:
> > >
> > >        message       =   *(fields / obs-fields / news-fields)  
> >                            ^  
> > >                          [CRLF body]  
> > 
> > Wouldn't this asterisk introduce the same "infinite loop of empty
> > strings" situation as in the erratum noted above?  
> 
> As "fields" in RFC 5322 amended with erratum 2950 no longer has an 
> infinite loop of empty strings, *(fields / obs-fields / news-fields) 
> also do not have, don't it?

Agreed. Looking at RFC 5322 again I'm no longer sure that there really
was an infinite loop even without erratum 2950 applied.
For an infinite loop of empty strings something like this is required:

   *(*a)  ; Can result in ∞(0a)

Looking at:

   *(trace
     *optional-field /
     *(resent-date /
       resent-from /
       resent-sender /
       resent-to /
       resent-cc /
       resent-bcc /
       resent-msg-id))

The second and third repetition operators are not nested (alternatives)
but the result can reduce to:

   *(trace)

if both of them use zero repetitions. Note that there is no slash after
trace and trace must contain at least one "Received" header field:

   trace  =   [return]
              1*received

Therefore the content in the outer parenthesis can not be empty.

Have I overlooked/misunderstood something?
Can somebody show an example how to create an infinite loop of empty
strings from the original syntax?

-- 
Michael Bäuerle