RE: : Using DIAMETER_MISSING_AVP for optional AVP

Hi,

> Based on this definition, it is only applicable to AVP 
> "required by the command code definition". It is not clear if
> we could reuse the same result code for an optional AVP.

In my opinion here "command code definition" refers to the whole
definition of the command, not just the ABNF. Some parts of the
definition are often presented in "free" text. For example,
Diameter Application command definition might state that
Pet-Poodle AVP must be present in the message in case it
is Friday the 13th. :)


> My assumption was that a "SHOULD" was possible here because you could
> have relyed on the command code ABNF descritpion to find the missing AVP.

So if the command-code ABNF definition defines three required AVPs
and you receive DIAMETER_MISSING_AVP without Failed-AVP how do you know
which one was missing? You need to know the message contents and the
command-code definition (including the rules for "conditional" AVPs) if
you want to know what AVP was missing. It doesn't matter at all whether
the missing AVP is required or optional.


BR,
Mikko


> -----Original Message-----
> From: [email protected] 
> [mailto:[email protected]]On Behalf Of
> ext MORAND Lionel RD-CORE-ISS
> Sent: 02 November, 2004 18:02
> To: Loughney John (Nokia-NRC/Helsinki); [email protected]
> Subject: RE : [AAA-WG]: Using DIAMETER_MISSING_AVP for optional AVP
> 
> 
> > From 3588:
> > 
> >    DIAMETER_MISSING_AVP               5005
> >       The request did not contain an AVP that is required by the
> Command
> >       Code definition.  If this value is sent in the 
> Result-Code AVP,
> a
> >       Failed-AVP AVP SHOULD be included in the message.  The
> Failed-AVP
> >       AVP MUST contain an example of the missing AVP 
> complete with the
> >       Vendor-Id if applicable.  The value field of the missing AVP
> >       should be of correct minimum length and contain zeroes.
> > 
> > My guess is that if a particular implementation or 
> > application was expecting an AVP to be included, this would 
> > be a reasonable way to do it.
> > 
> 
> Based on this definition, it is only applicable to AVP 
> "required by the
> command code definition". It is not clear if we could reuse the same
> result code for an optional AVP.
> If there is a general consensus on this topic, maybe a modification to
> the existing definition should be done.
> 
> Besides, if we could reuse the same result code for required 
> or optional
> AVP, should we state that "a Failed-AVP AVP MUST be included" 
> instead of
> "should be included"?
> My assumption was that a "SHOULD" was possible here because you could
> have relyed on the command code ABNF descritpion to find the missing
> AVP. If the above assupmtion is correct, it is not possible 
> anymore and
> you have to use the Failed-AVP.
> 
> Lionel
>