> The fix is in line with the performance step of the optimization project
> so no problems here. It only shows that we need a refactoring in this area.
Cant we settle by implementing a cache-method and comment the exists()
as 'TBD in MMBase', preferably in 1.8-9?
> IOW we could get more performance when we eliminate more of these queries.
I suspect some more improvements can be done in the verify() method.
And versions.java is something which can be tuned (making it inactive
resulted in a 90 sec speed gain)
> > Fortunately we use different basenames. But we also use different
> > catalogs (database instances). MMBase should also work if we didn't
> > use different basenames. Marcel is looking into this more, and
> > hopefully he will provide a fix soon :-p
I did some debugging with the metadata.getTables() and found the following:
- supplying a catalog does not change the view.
I can set it to 'null' or whatever I want but the tables are the same.
When I do that the result.getString(1) always is the current database.
But setting it with a factory.getCatalog() is somewhat more readable.
- supplying a schema will change the view
this seems to correspond to the 'user' but then again that might be
how our database is configured. Not supplying ('null') will return all
tables in the current catalog.
_______________________________________________
Developers mailing list
[email protected]
http://lists.mmbase.org/mailman/listinfo/developers
