Stefan Manegold wrote:
On Sat, Sep 06, 2008 at 11:22:09AM +0000, Martin Kersten wrote:
Update of /cvsroot/monetdb/MonetDB5/src/mal/Tests In directory sc8-pr-cvs16.sourceforge.net:/tmp/cvs-serv8410/mal/Tests
Modified Files: tst026.stable.out tst1505.stable.out tst220.stable.out tst280.stable.out tst903.stable.out Log Message: The function aggr.count() now returns a lng, because we have to support tables with more then 3G tuples. The grouping operations still assume 32 bits to be sufficient.
Since we (as far as I can tell) can only support ~2G tuples on 32-bit systems, what about using "wrd" instead of "lng", i.e., 32-bit on 32-bit systems & 64-bit on 64-bit systems?
In fact, with with Sjoerd recent changes, BAT cardinalities are of type "BUN" which is currently defined as "oid"; hence, BATs are limited ~2G tuples in case of 32-bit OIDs (even/also on 64-bit systems with 32-bit OIDs) --- a VoID column can obvoiusly not handle more in case of 32-bit OIDs ...
Stefan
would indeed be more 'precise' or less of an overshoot. We need to adjust the group count as well.... so might do it in one blow tonight Niels will adjust SQL for this change i have to shop now : (((