Forum OpenACS Development: Re: Some thoughts with respect to the permissions system

Collapse
Posted by Don Baccus on
A minor point ...

"By doing so, we could eliminate all objects from the
permissions table that don't have directly assigned privileges WITHOUT
having to join against the context hierarchy (I'm not even sure
there is a use for an explicit context hierarchy anymore in this case)."

But you would have to join against the object table to pick up context_id, right?  Aren't you just replacing one join with another?

Now, there are more rows in the context hierarchy table than there are in the objects table but each row only contains two integers, while objects are considerably larger.  The size of  this table hasn't really been an issue thus far.

Yes - indeed, that was part of the idea. Trade joining against a large table for joining against a smaller table.

Also you could actually keep joining against acs_object_context_index,  and transform that table just into a flat (object_id,context_id) mapping.

Anyway - all said and done I don't think this would be a big win after all.

The bitstring privilege approach still feels like there should be something to gain from, but indeed the question is 'how much'.