Re: Empty IA5String

[John McMeeking <jmcmeek@us.ibm.com>]

"Kurt D. Zeilenga" <Kurt@OpenLDAP.org> wrote on 11/09/2004 08:37:21 AM:

> At 08:47 AM 11/9/2004, John McMeeking wrote:
> >Otherwise, from the perspective of not breaking existing implementations
> >(or applications),
>
> Which applications would we be breaking?  Is there actually
> interoperability between implementor of IA5 string?

I'm assuming there could be a client application that is storing
zero-length strings (IA5 String, that is) in a directory on the basis of
the current RFC's definition of zero or more characters.  If we changed the
definition of IA5 String to one or more characters, and the server vendor
changed their server implementation to match, a previously working (and
valid) application could start failing.  It could also present problems for
the server in dealing with existing entries containing what just became
illegal data.

>
> I argue that allowing empty IA5 strings might actually break
> more applications.
>

I agree with that.  I've seen more than a few applications that would not
work if they got a zero-length value - string or otherwise.  You could
argue that is a problem with the application not taking into consideration
that an attribute can have a zero-length value.  But you could also argue
that something (a syntax) that breaks applications in this way is poorly
designed.


John  McMeeking