[Date Prev][Date Next] [Chronological] [Thread] [Top]

Re: ldbmcat does not produce RFC2253 DN: lines (ITS#135)

d.begley@nepean.uws.edu.au wrote:

> True.. aren't the syntaxes "fixed" though (ie., I thought I saw an RFC that
> explained each of the possible syntax types for attributes)?

These are the 'standard' syntaxes from RFC2252.

> Or can anyone essentially define their own syntaxes?

Not anyone.  An extension provider (some kind of shared library loaded at
startup) may add new syntaxes, with code to go with it.  Currently, not much
processing is done at the syntax level and the little that is done will be
taken care of by equality, ordering and substring matching (this begs the
original question, since now we must ask how these are defined and who
can define them, additional code again).  But there are requirements there
that might make dynamic syntax addition difficult.

Let me offer an example from RFC2251, section 4.1.6:

   Clients MUST NOT send attribute values in a request which are not
   valid according to the syntax defined for the attributes.

It is difficult to comply with it in a world of extensible syntaxes.  I think
that this requirement should be taken with a grain of salt, anyway, it is
unreasonable to ask that from most clients.

Anyway, since this topic (hashing v3 implementation options) is not appropriate
for our bug-tracking system, I have dropped openldap-its from the list of
recipients and added openldap-devel instead, since that is the list where v3
development is discussed (the stable version is v2).  If you are interested
in v3 development, you are welcome there.