> It seems that this approach would suffer the same problem as the one I
> outlined in "1) Actually separate client data by table". I would have
> to modify the logic of my web app...My web app currently handles all of
> the data, regardless of company, so it would have to aggregate the data
> from the different schemas when pulling data, and be smart enough to
> write back to the proper schemas when writing data.
>
on the public schema you can have a table with the company and schema
and then select the default search path to the company's schema
and the rest of your app will be untouched
leonel