unfortunately i need a very quick solution for that problem.

is there a possibility to order the permissions in the security layer?
i dont't know exactly if that would help but i think it could fit my
needs if first the positive and then the negative permission are
evaluated...


Am Don, 2003-08-14 um 10.59 schrieb Oliver Zeigermann:
> Hi Michael,
> 
> did not want to be quibbling ;)
> 
> Anyway, I understand what you describe there seems to be a critical 
> problem for anyone using JDBC stores. What can be done?
> 
> Why does it work with other stores, when no order can be defined in 
> SecurityStore? Are the ACEs in the right order when they are added?
> 
> Would it be a solution if we had this interface:
> 
> public interface SecurityStore extends Service {
> 
>      List getPermissions(Uri uri); // list of NodePermission
>      void addPermission(Uri uri, NodePermission permission); // append
>      void addPermission(Uri uri, NodePermission permission, int index);
>      void removePermisson(Uri uri, int index);
>      void removePermisson(Uri uri); // remove all
> }
> 
> or even simpler:
> 
> public interface SecurityStore extends Service {
> 
>      List getPermissions(Uri uri);
>      void setPermissions(Uri uri, List permissions);
> }
> 
> Oliver
> 
> Michael Smith wrote:
> 
> > Oliver Zeigermann wrote:
> > 
> >> Well, your advice has not been ignored completely as it seems. In
> >>
> >> http://marc.theaimsgroup.com/?l=slide-dev&m=103637094431479&w=2
> >>
> >> Eckehard Hermann pointed that there now is a switch to use the old acl 
> >> semantics. Maybe this can help. Even though this might imply 
> >> reorganizing all existing ACL entries (I am no expert at this, I just 
> >> interpret what is in the thread given above).
> >>
> >> Oliver
> > 
> > 
> > Right. And, as the last message in that thread discusses, this just 
> > leaves the default case broken, and the non-default case is stuck using 
> > the old (and drastically less useful) ACL semantics.
> > 
> > In the case where this distinction is neccesary (i.e. when the behaviour 
> > of the new and old SecurityImpl differs), the old (and now optional) ACL 
> > semantics are not capable of specifying the same access controls as the 
> > new one.
> > 
> > This is why I suggested that the SecurityStore interface needed 
> > modifying (and the jdbc stores along with it), rather than just 
> > reverting to the old/less useful SecurityImpl.
> > 
> > So yes - you're technically correct in that my original message wasn't 
> > completely ignored - but the 'solution' ignored my major problem with 
> > the patch, and left slide broken by default (at least with any of the 
> > jdbc stores).
> > 
> > Michael
> > 
> > 
> > 
> > 
> > ---------------------------------------------------------------------
> > To unsubscribe, e-mail: [EMAIL PROTECTED]
> > For additional commands, e-mail: [EMAIL PROTECTED]
> > 
> > 
> > .
> > 
> 
> 
> 
> 
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: [EMAIL PROTECTED]
> For additional commands, e-mail: [EMAIL PROTECTED]
> 
> 
> 
-- 


Marc Sommer                             I::Dev
+49 721 91374-364                       Schlund + Partner AG
PGP Key-ID: 0743ED19                    http://www.schlund.de

Attachment: signature.asc
Description: Dies ist ein digital signierter Nachrichtenteil

Reply via email to