Re: [Editorial Errata Reported] RFC5537 (1980)

[Alfred Hönes <ah@xxxxxxxxx>]

> RFC Errata System <rfc-editor@xxxxxxxxxxxxxx> writes:
>
>> Notes
>> -----
>> Rationale:
>>   Contrary to its companion document, RFC 5536, this RFC mixes precise
>>   IETF terminology for protocol elements and colloquial abuse of it in
>>   various places.  For clarity and consistency, it should also
>>   inequivocally make use of the standard terminology; the fields
>>   of the "header" that a protocol layer or sub-layer adds to its
>>   payload are "header fields", not "headers" in itself.
>
> Argh, I thought I caught all of these.  I'm sorry about that.  The
> corrections (a), (b), (c), (d), (h), and (k) are correct.
>
>>   Similarly, denoting as "header field" a "header field value" is
>>   confusing -- items (e), (f), (g), (i), and (j) above.
>
> I don't understand the rationale for this change.  I believe the current
> text for those items is already correct and this change should not be
> necessary.
>
> --
> Russ Allbery (rra@xxxxxxxxxxxx)             <http://www.eyrie.org/~eagle/>

Russ,
thanks for your very quick responses.

The Control syntax seems to be the single issue that needs elaboration.
(I did not want to add too much stuff to the Errata note.)

Quoting from RFC 5536, Section 3.2.3, on top of page 18:

   control         =  "Control:" SP *WSP control-command *WSP CRLF

   control-command =  verb *( 1*WSP argument )
   verb            =  token
   argument        =  1*( %x21-7E )

These are the syntax rules for the entire header field, giving
the header field name, "Control", the separating colon ":",
and the general syntax of the header field value, essentially
<control-command> plus enclosing optional whitespace.

RFC 5537 imports the <control-command> rule (cf. Section 1.4,
on page 4) and states in the second paragraph of Section 5
(please note the "of" in the first line here!):

   The <control-command> of a Control header field comprises a <verb>,
   which indicates the action to be taken, and one or more <argument>
   values, which supply the details.  For some control messages, the
   body of the article is also significant.  Each recognized <verb> (the
   control message type) is described in a separate section below. [...]

Hence, for instance for (e), the ABNF in Section 5.2.1 of RFC 5537,

        control-command     =/ Newgroup-command
        Newgroup-command    = "newgroup" Newgroup-arguments
        Newgroup-arguments  = 1*WSP newsgroup-name
                              [ 1*WSP newgroup-flag ]
        newgroup-flag       = "moderated"

augments the syntax rule for the Control header field *value*.

The syntax of the entire Control header field is still given
by the <control> ABNF production from RFC 5536 (quoted above).
However, taking the RFC prose text literally,

             ...   The syntax of its Control header field is:
        control-command     =/ Newgroup-command
        ...
would pretend to indicate that the "Control:" part were dropped here!

That's why the Errata note adjusts the text to say:
                                                         vvvvvvv
|            ...   The syntax of its Control header field value is:
        control-command     =/ Newgroup-command
        ...

Does that suffice for clarification?

The iterated instances in 5.2.2, 5.2.3, 5.3, and 5.5 are analogous.

Kind regards,
  Alfred Hönes.

-- 

+------------------------+--------------------------------------------+
| TR-Sys Alfred Hoenes   |  Alfred Hoenes   Dipl.-Math., Dipl.-Phys.  |
| Gerlinger Strasse 12   |  Phone: (+49)7156/9635-0, Fax: -18         |
| D-71254  Ditzingen     |  E-Mail:  ah@xxxxxxxxx                     |
+------------------------+--------------------------------------------+