[Date Prev][Date Next]
Re: commit: ldap/servers/slapd backend.c controls.c
At 12:42 PM 1/26/2005, Pierangelo Masarati wrote:
>Kurt D. Zeilenga wrote:
>>I think the kludge is still needed, at least for frontend controls.
>I agree, but for instance, while testing manageDSAit and chaining interoperability in selected cases I hit the "iscritical" issue as of ITS#3519 (I admit it was because of a bug somewhere else, and I couldn't find any critical case about it - I guess that's why we never noticed the point :). I suggest we eliminate the SLAP_CONTROL_FRONTEND from those controls that now can work without, or set up a mechanism like a SLAP_CONTROL_MAY_NEED_FRONTEND flag that triggers a notice if the control wouldn't pass the check without the FRONTEND flag, or something like that... I'm not even sure that my fix (that is: list controls provided by global overlays in all databases) is the right approach.
The best way I can think of handling this is just to have
backend_check_controls, if the control is not serviced by the
backend (and/or overlay), to check to see if the control is
serviced by the frontend. This can be done by just checking
the list (bervals would be nice) or though use of some bit array.