RE: reframing the Empty IA5 string discussion

["Ramsay, Ron" <Ron.Ramsay@ca.com>]

But, for example, does caseIgnoreMatch say that "    " and "" are equal? I would have thought that "    " and " " are equal and that neither equals "".

If this were the case, I would expect the next in your sequence of caseIgnoreIA5Match lines would be
     caseIgnoreIA5Match says "   f"   and " " are equal.

Ron

-----Original Message-----
From: owner-ietf-ldapbis@OpenLDAP.org
[mailto:owner-ietf-ldapbis@OpenLDAP.org]On Behalf Of Steven Legg
Sent: Thursday, 11 November 2004 16:21
To: Kurt D. Zeilenga
Cc: ietf-ldapbis@OpenLDAP.org
Subject: Re: reframing the Empty IA5 string discussion



Kurt,

Kurt D. Zeilenga wrote:
> I see that there are few reasonable options here:
> 1) IA5 strings precludes zero-length values
> 2) IA5 strings allows zero-length values
>    2A) a zero-length values only matches a zero-length value
>    2B) a zero-length value matches " " (and all values
>        equivalent to " ").
> 
> I prefer 1 over 2A over 2B (see prior posts for why).

There are two separate questions here. Whether or not the IA5 String syntax
is restricted to preclude zero-length values (I don't care either way) there
is still the potential for zero-length IA5String values to be compared by
caseIgnoreIA5Match through component matching. 2A versus 2B still has to be
addressed. On that score I prefer 2B as it makes more sense when taking the
limit.

     caseIgnoreIA5Match says "   foo" and "foo" are equal
     caseIgnoreIA5Match says "   fo"  and "fo" are equal
     caseIgnoreIA5Match says "   f"   and "f" are equal
                     2B says "   "    and "" are equal
                     2A says they're not - that seems odd to me!

Regards,
Steven

> 
> (Regardless of the choice LDAPprep can and should be designed
> to be idempotent.)
> 
>