On Tue, Dec 15, 2009 at 12:25:59PM -0500, Jay Pipes wrote: > I'm thinking that you and Stewart would benefit from a summary, > including all the example code me, Paul, and Toru have been working > through, so I will put that together shortly. Basically, what Paul, > Toru and myself have been putting together is a flexible API that > would allow the engine to handle the SQL statement entirely if it > wanted to, to let the kernel execute the SQL statement entirely, or > a work-sharing API that allows the engine to participate at various > times in the execution of the statement by the kernel...
which gets into QFP (Query Fragment Pushdown) which some engines (well.. remote ones really, or ones that store computed agregates or can execute them better) is aiming to do... there's some proof of concept stuff in NDB for this now (although the SQL server part is the tricky horrible bit). there's rather crazy advanced levels you can go to with it (which some may find useful)... but small steps is the winning point :) -- Stewart Smith _______________________________________________ Mailing list: https://launchpad.net/~drizzle-discuss Post to : [email protected] Unsubscribe : https://launchpad.net/~drizzle-discuss More help : https://help.launchpad.net/ListHelp
