Re: New text: Attributes with no equality matching rule

["Kurt D. Zeilenga" <Kurt@OpenLDAP.org>]

At 01:09 PM 3/10/2004, Hallvard B Furuseth wrote:
>Kurt D. Zeilenga writes:
>> To address this issue, I suggest the following changes be made
>> to [Models]:
>> 
>> Delete from 2.3 the sentence:
>>   If the attribute type is defined with no equality matching rule,
>>   two values are equivalent if and only if they are identical.
>
>No, keep the definition of equivalence.  Equivalence is not used for
>evaluating AVAs and so on, but to ensure that no values of an
>attribute are equivalent, and that values equivalent to the stored
>value is returned:
>
>- last sentence of the paragraph above your suggested change:
>
>  No two values of an attribute may be equivalent.

I actually added the "no equality rule" sentence above
specifically to address issues of creation of an attribute
with multiple-values, not to address issues of individual
value addition/creation nor AVA evalation.

So, I suggest moving the paragraph break up one sentence
and to keep the above sentence.  That is, create a paragraph
that reads as follows:

  No two values of an attribute may be equivalent.  Two values are
  considered equivalent only if they would match according to the
  equality matching rule of the attribute type.  If the attribute
  type is defined with no equality matching rule, two values are      
  equivalent if and only if they are identical.  


>- [Protocol] 4.10 (Compare Operation):

I have no problem with your [Protocol] suggestions.

>>     - attribute value assertions
>
>..."(such as matching in filters and Compare operations)"

Instead:
  (such as matching in search filters and comparisons)

(I rather avoid use of the term "operations" when discussing
the abstract data model.)


>>       using values of such a type
>>       cannot be performed.
>
>...because at that point, [Models] hasn't said what AVAs actually do.
>
>
>> I note that I didn't say anything about syntax checking of
>> values as this is already covered in section 2.5.4.
>
>Not sure what you mean here, but it should probably be mentioned
>somewhere that LDAP differs from X.501(1993) section 12.4.5a in that it
>can perform syntax checking for attributes without EQUALITY matching
>rules.

I was noting that I did not make add a specific statement about
syntax statement as section 2.5.4 already said:
  Attribute values conform to the defined syntax of the attribute.

If you could offer specific wording to call out the difference,
I'd appreciate it.  Right now, all the wording I've thought of
is a bit too intrusive.


>Also, you might update this statement in 2.5.4 (Attribute Values)
>to mention that the name attributes also need EQUALITY rules:
>
>  Only attributes whose descriptions have no options can be used for
>  naming.

Suggest:
   Only attributes whose types have an equality matching rule
   and whose descriptions have no options can be used for naming.

or it may be better to reword this as two sentences, each of
which excludes a set of attributes.

>Finally, this might be copied to [Protocol] 4.6 (Modify Operation):
>
>>     - individual values of a multi-valued attribute cannot be
>>       added or deleted;
>
>(...if the attribute has no EQUALITY matching rule.)
>
>BTW, that is actually stricter than rfc2251, which only forbids
>_deletion_ of individual values, not addition.

Well, if you consider that rfc2251 said that servers MUST
act in accordance with X.501, one can view RFC 2251 as just
not reiterating all the restrictions that are enforce.
I intend to categorize this change (in the Changes since
appendix) as simply addressing a technical omission.

Kurt