[kronolith] performance on large systems

[John Madden]

> - Make this configurable, at least in the driver code, so that people  
> that still have databases that don't support sub-queries (shouldn't be  
> many) can turn the old behavior on.

If I have the time to circle back on this, I'll give it a go.

> - You can't JOIN with horde_groups, because the groups could come from  
> any group driver, so you have to keep the old way of getting group  
> information.

Same as above, although maybe a bit easier.  If the groups are in LDAP 
though, for example, would any of this be relevant?  You can't JOIN 
against that anyway.

In our case, we use SQL for everything, so this is at least a starting 
point for whoever takes the next step.  I don't mean for the patch to 
replace the sharing driver itself, but the concept here is scalability, 
which I think it certainly demonstrates for anyone more familiar with 
Horde internals.

John




-- 
John Madden
Sr UNIX Systems Engineer
Ivy Tech Community College of Indiana
jmadden at ivytech.edu