D
dbeavon
Guest
Does anyone know how best to leverage the cost-based query plans in ABL (ie using selectivity statistics)? I know its a big question but OE has cost-based query plans on the SQL92 side of things and it would be wonderful to take advantage of them in a regular ABL program too. The problem with the ABL record loops (ie "FOR EACH") is that not only do you have to know which tables to join, you have to manually specify what order to join them and even what index to use (assuming there are more than a few). We'd like to give the users the ability to use multiple ad-hoc filter criteria on two different tables, whereby sometimes it makes sense to run the outer loop on table 1 and sometimes it makes sense to run the outer loop on table 2. SQL92 should know what to do based on regularly maintained statistics. I'm not finding the answer to this question. While there are plenty of FAQ's on OE sql92. Nothing brings together the SQL92 world with the ABL world. https://community.progress.com/community_groups/openedge_general/w/openedgegeneral/1613.bi4p-sql-92-frequently-asked-questions I remember back in the day you could use inline SQL89 in ABL but I think that stuff is all deprecated by now. My only idea at this point is to build a SQL query by hand and shell out to "sql92" to run it and generate some sort of result set of primary keys from the two tables, which ABL would at last be able to deal with competently. Not pretty....
Continue reading...
Continue reading...