Re: Changes to perl-ldap schema code

[Graham B. <gb...@po...>]

On Mon, Feb 05, 2001 at 10:21:30AM -0600, Clif Harden wrote:
> > 
> > Clif Harden <ch...@po...> wrote:
> > > + =item schema ( OPTIONS )
> > > + 
> > > + Request that a schema search be performed.  This can be used to read 
> > > + schema information.
> > > + 
> > > + The result is an object of class L<Net::LDAP::Schema|Net::LDAP::Schema>.
> > > + Read this documentation for further information about methods that
> > > + can be preformed with this object.
> > > + 
> > > + =over 4
> > > + 
> > > + =item dn
> > > + 
> > > + If a DN is supplied, it will become the base object entry from 
> > > + which the search for schema information will be conducted.  If
> > > + no DN is supplied the base object entry will be determined from
> > > + the rootDSE entry.
> > 
> > The changes look OK. I'd like to see it returning cached Schema objects,
> > because these are quite expensive to keep retrieving and parsing. My second
> > change of last week implemented a cache, it was kinda hacky but generally
> > the right sort of approach.

I don't think it is the place of Net::LDAP to do the caching. As you say these
can be quite large and for Net::LDAP tp cache them could just be a waste
on memory. It should eb up to the user application to cache the object
returned.

> > I'd suggest one change to the documentation above, to the effect that the
> > DN is the name of a subschema *subentry* containing the schema. ie replace
> > the word entry with subentry. I think that clarifies what it is doing.
> > 
> 
> I will leave this up to Graham to decide.

Sounds fine to me.

> > We still need a mechanism for retrieving subschema given the DN of a plain
> > entry (, which is by far and away the *best* way of getting schema. (Maybe
> > we can add this to Net::LDAP::Entry?)
> > 
> 
> I will leave this function for someone else to contribute.

Patches welcome.

Graham.

> 
> I have been using the objectClass attribute for this function since you
> can not be sure that a subschemaSubentry attribute will exist for any
> given entry.  Since I use the caching method described above it works
> great for me.  It may or may not work for anyone else.
> 
> 
> > Cheers,
> > 
> > Chris
> > 
> 
>