Re: LDAP Comments

[[e-mail]]

> A read operation can be simulated in LDAP using a BASEOBJECT search
> with a filter testing for the existence of the objectClass attribute.
> It should only cause one chaining operation, just like a read.
> 

Right,
but then I would strongly recommend in case the READ and LIST are not
explicitely exhibited at LDAP cleint level, to add a chapter
of recommendations to the client designers and developers
on how to write "interworking" procedures
  eg.
	1. the way to get a single entry by a search base-object
		and as far as possible never a search one-level
	2. to avoid using Quipu specific attributes, except once
	the client has the proof the DSA it is interworking with
	is a QUIPU DSA
	aso... (to be completed in the light of experience)

	
> 
> The QUIPU approach obviously has some good points.  But I agree that
> it imposes unacceptable limitations in the long run.  Something will
> have to be done about it eventually, and I believe you are correct
> that the IC will have to do it.  So, I wouldn't bet you a bottle of
> "faugeres", even if I knew what that was!                 -- Tim

It has indeed many good points, and the challenge is for every implementation
to try to equal the performances of QUIPU while avoiding that kind
of traps..
But, it is not related :-), faugeres is a french red wine from
"Les hauts cantons du Languedoc", which I intend whatever will happen to
LDAP to have you taste and 'criticize' a few weeks from now,
when you will be our guest


-- PAP

> 
> 
> 
>