RE: ACL performance again

["Howard Chu" <hyc@highlandsun.com>]

> -----Original Message-----
> From: Stephan Siano [mailto:stephan.siano@suse.de]

> > Actually there's another case, ACL sets. Too bad there doesn't seem to be
> > an easy way to cache these, because they are value-dependent and can be
> > very complex and expensive to evaluate.
> 
> How do these ACL sets look? I couldn't find anything about it in the code. 
> The only places where the val parameter in acl_mask() is used are the two 
> cases mentioned above. 
> 
> Note that some ACLs depend on some attribute values but need not be 
> consindered as value dependant in this context (e.g. normal by dnattr= ACLs 
> (without self) because even though they depend on the value of an attribute 
> in the entry they are valid for the whole entry and not for single values.

OK, I missed that point before, I think you're right.

  -- Howard Chu
  Chief Architect, Symas Corp.       Director, Highland Sun
  http://www.symas.com               http://highlandsun.com/hyc  
  Symas: Premier OpenSource Development and Support