RE: LDAP schema propagation to the backend

["Joao Abrantes" <jabra@mail.telepac.pt>]

	So, from what I understand, the backend design should be objectlass and
attribute independent.
	That means a good backend design should only be aware of attibute syntax
(CIS, CES, etc...) and that´s all... Is this correct?

	Joao Abrantes

-----Original Message-----
From: owner-openldap-devel@OpenLDAP.org
[mailto:owner-openldap-devel@OpenLDAP.org]On Behalf Of Kurt D. Zeilenga
Sent: quinta-feira, 27 de Janeiro de 2000 0:21
To: jabra@mail.telepac.pt
Cc: OpenLDAP Development
Subject: Re: LDAP schema propagation to the backend



At 10:04 AM 1/26/00 -0000, Joao Abrantes wrote:
>	I'm curious about how the schema definitions get propagated from the
>frontend to the backend in OpenLDAP.

The information is managed by the frontend.  The backend must
access the frontend accessor routines or, in some cases, the
structures directly.

>What i want to know is: is there any
>structure or any call to configure (or something similar) that imports the
>schema definitions from the slapd.conf file to the frontend and ultimately
>to the backend?

Well, in 2.0, backends should be fairly schema neutral.  I am working
right now on providing routines in the frontend which will eliminate
the backends need for general schema knowledge.  There are, of course,
certain cases where the backend needs some knowledge.  In such cases,
we'll provide access routines such that backends need to chase
through frontend data structures.

In 1.x (and current code), it's a mess.  Your backend will have
to dig for what it needs.

>	I´m asking this because I´m triyng to define a database
>schematic for the MySQL backend and I don´t know (in the backend
>functions) how to get the schema defined in slapd.conf.

I'm not sure why the backend would define schema.  Other than
a few "core" schema elements (like "top"), the majority
of schema definitions are user defined.  (There are access
routines to lookup up routines defined for backends to lookup
what the user has defined).

If the backend itself needs to define some schema elements (which
should be avoided if at all possible), then the backend will
need to provide the definition and routines implementing the
definition to the schema registry (using add functions).

I am actively working on schema codes... the API for this stuff
will change dramatically over the next frew weeks.

	Kurt