RE: the advantages of garbage collection

I wasn't look at it as being a positive if it was done in memory as

you are

:-) Joining result sets, IMO, should be done by the DBMS and not the
language.

Problem there is you make the assumption that the DBMS has access to all
the data. Actually, you make two assumptions, that's the first. The
second is that you want the DBMS doing it from a performance point of
view. Web servers scale out well, DBMS server only scale up well.

What if you're trying to sort the results of a webservice call? Or the
data in an XML table? Or even merging and then sorting the data from a
"join" between data in i, data in SQLServer and data from a webservice
call? In all these cases the DBMS alone doesn't have what it needs to
sort, because it doesn't have all the data. Now yes, you could then
write the "joined" row into a table in qtemp and then ask the dbms to
sort that table, but that seems like lots of extra work over just
letting the language/runtime sort it.

-Walden