On Wed, Nov 16, 2011 at 5:31 PM, Emmanuel Lecharny <[email protected]>wrote:
> Hi, > > I think we can safely move the OperationalAttributeIntercepto**r after > the SchemaInterceptor. This is important if we want to remove the > re-entrant chain call for schema modifications, as we would then be able to > continue the modification without having to go through all the interceptors. > > When we try to modify the schema by modifying the subschemaSubentry entry > (ie, cn=schema), we modify the attributes by adding (or removing) values, > like : > > injecting ( 1.3.6.1.4.1.65536.0.4.3.2.2 NAME 'templateObject' DESC 'test > OC' SUP top STRUCTURAL MUST ( templateData $ cn ) X-SCHEMA 'other' ) into > 'objectClasses'. > > What happens then is that we go down the chain up to the > SchemaInterceptor, where we parse the value, and then g through the full > chain with the converted Entry. We can simply continue to process the entry > starting from the SchemaInterceptor instead. > > The only little trick is that we may have more than one modification, but > that's not a big deal. > > I'm going to try this approach. > > OKIE - let us know how it works out. -- Good Luck, -- Alex
