[prev in list] [next in list] [prev in thread] [next in thread] 

List:       openldap-devel
Subject:    Re: Run-time registered controls
From:       Pierangelo Masarati <ando () sys-net ! it>
Date:       2005-01-29 14:31:29
Message-ID: 41FB9E41.4020801 () sys-net ! it
[Download RAW message or body]

Kurt D. Zeilenga wrote:

>I've updated controls.c/backends.c with a kludge which provide
>proper criticality handling.
>  
>
I'm getting more and more convinced that the design of 
backend_check_controls() is obsoleted by current code design.  It 
originated from the need to check whether a certain backend was able to 
handle a specific control when control handling implementations were 
essentially backend-specific; now we're moving towards an implementation 
of control handling that it rather overlay-centric, allowing the same 
implementation to be shared by different backends and thus moving it a 
bit towards the frontend, in the no man's land between the two.  So 
either we make backend_check_controls() be intercepted by overlays, or 
we design a different type of check, e.g. something that is invoked at 
the fronted level and uses the requestDN as a hint to trace what path 
the handling of the request will take.  This would likely require 
multiple select_backend() invocations, unless we group the two tasks, 
and let backend_check_controls(), or even backend_check_restrictions(), 
return the selected and validated database.

p.





    SysNet - via Dossi,8 27100 Pavia Tel: +390382573859 Fax: +390382476497

[prev in list] [next in list] [prev in thread] [next in thread]