Oscar Figueiredo <Oscar.Figueiredo(a)di.epfl.ch> writes:
>>>>> "Hrvoje" == Hrvoje Niksic
<hniksic(a)iskon.hr> writes:
>> Perhaps, but I think that the code to do anything ldap-ish would
>> still look very ldap-ish as opposed to abstract-database-ish.
Hrvoje> I agree with Bill. Abstract interfaces are a nice thing up to
Hrvoje> a point, i.e. as long as they serve to us and not the other way
Hrvoje> around.
Hrvoje> If LDAP serves a purpose totally different than a flat
Hrvoje> database, it's quite fine to have a different interface for the
Hrvoje> two.
Just to add my 2 cts, I also like abstraction and uniform interfaces
(after all EUDC is a simple attempt at that) but I agree 200% with Bill
and Hrvoje. The LDAP APIs are still evolving, adhering to them is the
simple way to go. Even if we devised a good interface above the current
API now, nothing guarantees we could easily fit forthcoming evolutions
into it. I, for one, won't go that way but I won't discourage anyone
that would like to implement such an interface.
After spitting out those ideas this morning, I think you could make it fit
fairly easily, but whether anyone thinks the costs are worth it. There IS
code out there that uses the existing LDAP routines (Emacs/W3, EUDC, etc),
but nothing that really uses the db routines (can anyone contradict this?
I couldn't find anything), so...
If we reworked the database routines like I suggested, you could write the
lisp-level database routines and layer them on top of LDAP to experiment
with. Then you could see which one makes more sense. The whole database.c
needs to get reworked anyway, and probably moved to a module. I don't know
if I'll have time to do it or not though. Damn job. :)
-Bill P.