Re: Openldap ldbm cache corruption

[Chris Ridd <C.Ridd@isode.com>]

On Tue, 12 Jan 1999 18:13:43 PST, "Kurt D. Zeilenga" wrote:
> At 01:58 AM 1/13/99 +0100, Harald Tveit Alvestrand wrote:
> >it might be nice to downcase rather than upcase - seems weird, but
> >there's a couple of cases (German double S, French (France) accents)
> >where a downcase preserves information that upcase loses, if done
> >according to a native locale.
> 
> We currently do everything in the "C" locale.  I'd prefer using
> uppercase characters because they are (generally) more noticeably
> different from the 'provided' DN.

Upper-case is better, because some matching rules (eg 
caseIgnoreOrderingMatch) match using explicitly uppercase values. 
X.520(93) 6.1.2:

=======
6.1.2 Case Ignore Ordering Match

The Case Ignore Ordering Match rule compares the collation order of a 
presented string an attribute value of type DirectoryString, without 
regard to the case (upper or lower) of the strings.

	caseIgnoreOrderingMatch MATCHING-RULE ::= {
		SYNTAX		DirectoryString {ub-match}
		ID		id-mr-caseIgnoreOrderingMatch }

The rule returns TRUE if the attribute value is "less" or appears 
earlier than the presented value, when the strings are compared using 
the normal collation order for their syntax after lower-case letters in 
both strings have been replaced by their upper-case equivalents.

Where the strings being matched are of different ASN.1 syntax, the 
comparison proceeds as normal so long as the corresponding characters 
are in both character sets. Otherwise matching fails.
=======

The current draft of X.520(97) 6.1.2 has identical text.

Chris