Re: Models: Matching Rule Uses

[Hallvard B Furuseth <h.b.furuseth@usit.uio.no>]

Kurt D. Zeilenga writes:
>>  "extensibleMatch filters can also use matching rules from an
>>  attribute type's AttributeType description."
> 
> I don't think that's correct.  Listing a rule in
> an attribute type description doesn't imply that
> rule can be used in an extensibleMatch filter.
> 
>> At least that's what you said in a private mail earlier.  Or if
>> I misundersood, add an explicit note about the opposite instead.
> 
> I could have been confused...
> 
> I can think of a number of possible opposites, but none
> make sense to me.  Please clarify.

Never mind for now.  I think the discussion below will cover it.

>> Also, am I right in assuming that a matching rule use may only
>> list attributes with a syntax which satisfies the restrictions
>> specified in the matching rule defintion in [Syntaxes] 5.2.*?
>> If not, something needs to be said about what these restrictions
>> mean.
> 
> I'm not sure which restrictions you refer to here.  Can you
> be more specific?

Example: [Syntaxes] 5.2.3 caseIgnoreIA5Match says:

   The caseIgnoreIA5Match rule compares an assertion value of the IA5
   String syntax to an attribute value of a syntax (e.g the IA5 String
   syntax) whose corresponding ASN.1 type is IA5String.

So caseIgnoreIA5Match can be used with attributes whose corresponding
ASN.1 type is IA5String, but not e.g. bit string syntax.  I'm hoping
that this also means that a caseIgnoreIA5Match matching rule use can
only list attributes whose corresponding ASN.1 type is IA5String.

>> Finally, it must be specified which attributes a Matching Rule
>> may be used with if it has no corresponding Matching Rule Use.
> 
> Not publishing a rule use (or any schema element)
> only means the server doesn't know that use (or isn't willing
> to let you know it knows that use).
> 
> Maybe 4.4 addition we previously discussed should say:
>   Clients SHOULD NOT assume a published subschema is complete nor
>   assume the server supports all of the schema elements it publishes 
>   nor assume the server does not support any elements it did not publish.

Heh.  It gets complicated, but I guess that's the right text.

-- 
Hallvard