🚀 go-pugleaf

I've mentioned this on the vdb programing newsgroup, but it is more 
appropriate to discuss this here. No development can occur in VdB until 
one has a relative feeling for the speed of data processing. I want to do 
this development right the first time.

The new VdB 7 changes all the rules.

From my own experimenting I have found SQL to be much faster than just 
using filters, on an individual dbf. On Joins the feeling is the 
opposite. A simple join (outer right) on two databases with about 30,000 
entries, filtered ("where" in SQL) for a specific field (not the joined 
field) takes about 2 minutes. While the outer right join is useful to 
work on live data, would it be faster to do this any other SQL way?

Just for speed comparison, using tradition DBF style rowset.findkey() 
with the index ordered on the "filter" field then doing a lookup from the 
"join key" (not actually joining) onto the second dbf and storing the 
data to an array takes only 5 seconds. (40x's faster than sql)

Is this what one should expect?

Any pointers appreciated.

Thanks.


-- 
Gerald Fay
gjfay@msn.com