From Christ.Klinge@web.de Mon Mar 20 10:23:20 2017 From: Christ.Klinge@web.de To: openldap-bugs@openldap.org Subject: Re: (ITS#8619) Enhancement request: Nested group support using dynlist recursion Date: Mon, 20 Mar 2017 10:23:18 +0000 Message-ID: MIME-Version: 1.0 Content-Type: multipart/mixed; boundary="===============6832343951014762344==" --===============6832343951014762344== Content-Type: text/plain; charset="utf-8" Content-Transfer-Encoding: quoted-printable
I have encountered some scepticism regarding the benefits of nesting = within OpenLDAP itself. Some have argued, that applications should resolve ne= sted groups or that nested groups should be created using automation instead.= Here, I'd like to respond to these two objections.
 
1. applications should be responsible resolve nested groups
First, I disagree from a philosophical point of view. The identity manag= ement system and thus the user directory is the central point of knowledge re= garding group membership. For the sake of maintainability, all of the informa= tion as to why any given user is member of any of its groups, should be prese= nt at this central location. Whether a user is a direct member of a group or = whether he is member of a sub-group, may interest applications, but what matt= ers most, is that the user is in fact part of both the sub-group and all of i= ts ancestors.
 
Secondly, some applications simply don't have nested group support. = It is a fairly common feature, but it just isn't part of every piece of s= oftware out there. Implementing nesting in the directory removes the need for= support on the application side entirely. In the (from my curent point of vi= ew unlikely) event that some application demands to resolve nesting itself, a= liasing can be used to deactivate dynlist for the given application.
 
2. automation instead of nesting
Automation comes with two caveats which I would like to address individu= ally:
 
2.a additional software
This may come as a no brainer for most, but I'd like to point out th= at automation requires some form of additional software, be it diy scripts or= an application. This increases complexity both due to operation of this soft= ware and its interaction with the user directory.
 
2.b divergent center of information
Instead of maintaining nesting information within the user directory, th= e software used most likely stores its data outside of the directory. Worst c= ase, it is hardcoded into some scripts. Thus, the information as to which gro= ups are related is likely stored outside of the actual directory itself. This= point may be void if the automation system stores nesting information on the= group objects inside of the user directory.
 
Sincerely,
Christopher Klinge
--===============6832343951014762344==--