Hi, > I do not have strings in different character sets. While there may > exist programs that do, most do not. > The way I added to the API allowes your very uncommon case too.
That is only an assumption, isn't it ?! IMHO a general API should not make assumptions about the "rareness" of use cases. > But I have no strings in Latin2. I always you ONE character set inside > my applications. It makes no sense to complicate the world by having > many at the same time. You do not have 2 character sets if you use UTF-8 for the strings that need to be UTF8-encoded for LDAP anyway. It's only one: UTF8 > Yes it will. But if you look at my current use the local character set > is Latin1 and no data in my LDAP directory will contain anything > with characters outside Latin1. > I expect it is the same in most places in the world - you use data > that is handled by the local character set in use. This is an asumption too, isn't it ? > >If you have to take Unicode anyway, you now can choose between > >the various representations. So, why not taking UTF-8 if it is the > >optimal encoding for perl-ldap (since it does not need any mapping) ? > > UTF-8 is not optimal for character handling. Are you sure about that ? Perl 5.6+ uses UTF8-encoding of characters internally for the new "character semantics" (the new semantics of strings that allows a single character to use more than a byte). That is another advantage of my approach with UTF8: If Perl and perl-ldap support it correctly, your application can do the character set conversion with PerlIO on the user interface side, then work with the regular Perl strings in character semantics [with all the amenities that Perl offers like case-ignoring regexes, lc(), uc(), ...] and finally pass that values to perl-ldap (or the other way around) I need to say the "If Perl and perl-ldap support it correctly" parts because I know that Perl 5.6.x has some issues with the Unicode support (and unfortunately I do not have access to a 5.8.x machine to test it). On the perl-ldap side I remember some discussions on the ML that there were some issues with "character semantics". I don't know If they are still there ? Peter -- Peter Marschall eMail: [EMAIL PROTECTED]
