[Date Prev][Date Next]
Re: schema questions
Frank Swasey wrote:
> The order is important. The first name listed is the "proper name" of
> the attribute and the second and any other names listed are the
Sad but true.
>> 2.) Does it matter whether I write "systelno" or "systelnumber" in my
>> ObjectClass definition? (Regarding schema checking during
>> ldapadd/-modify or something..)
> It won't matter to the OpenLDAP server, but if you're using something
> like GQ, or Net::LDAP you will get yourself in trouble. Neither of
> those client side tools understand/follow aliases.
Hmm, well, IMHO aliasing is a task of the _server_, and so it shouldn't
matter what clients think of it. If they get puzzled by aliases, they
[the clients] are perhaps just badly implemented. (Maybe some search
filter processing done by the client? Won't work with aliases,
obviously.) Or they are used "the wrong way"(tm).
> I've gotten myself into quite some trouble during my development
> phase by referencing the alias for an attribute in my PERL code and
> winding up trying to add a duplicate attribute because my PERL code
> didn't know that "fax" was really "facsimiletelephonenumber" for
Well, that's an error in your code. :o)
I don't think that's a task that can be addressed by Net::LDAP.
> I've also found that GQ 0.6.0 has the same problem
That's not a "problem", it's just intendeed behaviour, I'd say. You
could try to add a second facsimiletelephonenumber instead, and the
server would've returned the same error.
I.e. with OpenLDAP, where schema information isn't stored within the
LDAP tree, GQ has no (...) possibility to check the attribute
definition, to see whether it has an alias, nor whether it is marked
"multiple". It will just "try", and slapd will tell you whether that's
> -- so I have take to the practice of using the real names for
> attributes in my objectclass definitions.
That's a good idea anyways. :o)
I had a letter in the post today. It said 'Gas Bill'.
It sounds a tempting offer.
-- Alan Cox