[Date Prev][Date Next]
[Chronological]
[Thread]
[Top]
FW: draft-good-ldap-ldif-04.txt
Has anyone any comments / suggestions about the issue below?
I would have to agree with Gordon that the discussion probably DOES
belong in RFC 2251, but I'm still unsure about the semantics of what
should happen in the case where a zero-length attribute value is
added for an attribute type whose syntax disallows a zero-length value.
Thanks
------------------------------------------------------------------------
----
Mark (Mez) Morrell
Nexor, PO Box 132 Telephone: +44 115 9520582
Nottingham Fax: +44 115 9520519
NG7 2UU e-mail:
mailto:m.morrell@nexor.co.uk
> -----Original Message-----
> From: Gordon Good [mailto:ggood@netscape.com]
> Sent: 21 July 1999 22:09
> To: Mez Morrell
> Subject: Re: draft-good-ldap-ldif-04.txt
>
>
>
>
> Mez Morrell wrote:
>
> > Hi Gordon,
> >
> > Could I just ask for clarification on the intended meaning/use of
> > zero-length attribute values?
> >
> > I guess that for your seeAlso example, the intention is
> that the value is
> > an empty DN value referring to the "root" entry?
> >
> > For other attribute types where it does not make sense to
> have a zero-length
> > attribute value (e.g. commonName), is the implication that
> it is simply left
> > to a particular implementation to decide whether such an
> attribute value is
> > permitted or not? Would there be any guidelines as to what
> SHOULD be done
> > with such values, e.g. ignore all such attribute values and
> continue parsing
> > the file, or reject the whole entry, or would it again be down to a
> > particular implementation?
> >
> > My assumption would be that, as there is the facility to comment out
> > unwanted
> > lines in the LDIF file using the # character, that explicit
> zero-length
> > attribute values are intended by the author of the LDIF
> file to be an actual
> > attribute value?
> >
> > Thanks in anticipation......
>
> Mez,
>
> RFC 2251 allows zero-length attribute values, and the intent
> is that LDIF be
> able to represent LDAP operations with 100% fidelity. Hence,
> that's why LDIF
> allows them.
>
> I think a discussion of when and where zero-length attribute
> values makes sense
> or not probably belongs in RFC 2251 and not in the LDIF
> document. Roland Hedberg
> pointed out that X.500 explicitly disallows zero-length
> attribute values, so
> perhaps there is an impedance mismatch between the X.500 and
> LDAP specifications
> here. I'll discuss this with Mark Wahl, but feel free to
> raise this issue on the
> ldapext mailing list.
>
> -Gordon
>
>