Howard Chu wrote: > Michael Ströder wrote: >> Howard Chu wrote: >>> In RE24 the config schema are only exposed for DEVEL builds. >> >> Why that? Especially why was it changed compared to RE23? > > They were probably exposed in RE23 by mistake. The config schema uses an > OID from the OpenLDAP Experimental arc and the Project's policy is not > to publish experimental OIDs for general use, only for Development.
But then back-config SHOULD be hidden by DEVEL as well. Frankly no-one will ever recognize when you change an OID arc. There would be no issues with LDAP clients at all. > Given that things in cn=config have pretty much stabilized, we probably > ought to have migrated to a non-experimental arc for RE24. Yes, please. Given that back-config is endorsed to be deployed in production schema-aware LDAP clients must have the chance to retrieve required information for processing the data. > Note that the cn=config schema still relies on an unfinished spec > (draft-chu-ldap-xordered) and that needs to be finalized still. Whatever "unfinished" means. From the IETF point-of-view this is simply a proprietary feature not documented anywhere in the IETF process so far. From OpenLDAP project point-of-view it is yet another not fully documented feature. This is not a real issue which should hold back providing the schema to the client. But if back-config is activated in normal builds then the interface definition SHOULD be complete which includes providing the relevant schema information in subschema subentry. Ciao, Michael.
