Re: [CCAMP] RFC 4207 LMP Erratum

[Alfred Hönes <ah at TR-Sys.de>]

Dieter,

Your message almost has been scrubbed as spam at our provider,
because of the lack of 'normal' content.  ("HTML-only" is one
of the most efficient criteria in spam filtering today!)
However, I have managed to 'de-HTLMize' the message ...   :-)

> Hi Alfred,
> could you please explain why multiple <TRACE> objects for
> a single interface are needed? I fail to understand what the
> application is that requires this.
> Thanks,
> Dieter

First of all, I would like to refer you to the public errata report,
which -- you might have observed -- is almost 4(!) years old:

http://www.rfc-editor.org/errata_search.php?rfc=4207&presentation=records

I apologize for not having the time to fully re-read RFC 4207 now,
due to current overload, but I hope that the reasoning presented in
the Errata Report in 2005 is still valid.  I do not recall all the
details, but I'm quite sure it had not been based on specific
application scenarios, but rather on lack of consistency with other
parts of the RFC.  Please check and let me know if something's
wrong there.  If there indeed are good arguments against the
conclusion in the Errata Note, the quoted other parts of the RFC
might need reconsideration/clarification instead.

In short: either _all_ parts of the RFC should use plural for
TRACE object(s) in a TraceMonitor Message, or _none_.

Below, I have attached the original report once sent to the
authors of RFC 4207, which did not solicit a response those days
and has been entered later into the new RFC Errata database
by busy beavers at the RFC-Ed...


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 at TR-Sys.de                     |
+------------------------+--------------------------------------------+


--------  begin forwarded message  --------

> From: <ah at TR-Sys.de>
> To: jplang at ieee.org, dimitri.papadimitriou at alcatel.be
> Cc: rfc-editor at rfc-editor.org
> Message-Id: <200512201701.SAA26759 at TR-Sys.de>
> Date: Tue, 20 Dec 2005 18:01:34 +0100 (MEZ)
> Subject: RFC 4207 (LMP SDH Test)

Hello,

I've been stuck in an apparent inconsistency in the recently
published RFC 4207 authored by you.

RFC 4207 defines the syntax of the TraceMonitor message in Section
4.1.1, on page 6.
Similarly, the TraceMonitorAck and TraceMonitorNack Messages are
specified in Sections 4.1.2 and 4.1.3, respectively, on page 8.

While the former specifies a single <TRACE> object to appear
in a TraceMonitor message, the latter specs uses wording like
"all of the TRACE Objects in a TraceMonitor message" or
"TRACE object value(s)".

IMHO, it makes much sense to indeed allow multiple TRACE Objects
(with different trace types, but all related to a single Interface)
in a single TraceMonitor Message.

To resolve the inconsistency, I therefore propose to submit an
Errata Note for RFC 4207 aimed at adapting the syntax specified
in section 4.1.1 to the remainder of the specification:

------ cut here ------

RFC 4207, in Section 4.1.1, near the bottom of page 6, specifies:

           [...]  The format of the TraceMonitor message is as
  follows:

   <TraceMonitor Message> ::= <Common Header> <MESSAGE_ID>
                              <LOCAL_INTERFACE_ID> <TRACE>

It should say:

           [...]  The format of the TraceMonitor message is as
  follows:

   <TraceMonitor Message> ::= <Common Header> <MESSAGE_ID>
                              <LOCAL_INTERFACE_ID>
                              <TRACE> ...

------ cut here ------

If you agree with this Errata Note, please forward your consent
to the RFC-Ed.

[Note:
 I purposely have inserted the additional line break to emphasize
 that the repetition is to be applied to <TRACE> only; and I have
 used the more conventional style of repetition indication, '...',
 not the ABNF (RFC 4234) style "1*<TRACE>", to remain consistent
 in style with the basic LMP specification (RFC 4204) and because
 there is no ref. to RFC 4234 in RFC 4207.
 You might wish to change these details however, if you prefer.
]

Best 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 at TR-Sys.de                     |
+------------------------+--------------------------------------------+

--------  end forwarded message  --------