Brian Aker wrote: > Hi! > > On May 6, 2009, at 10:16 AM, Paul McCullagh wrote: > >> So I don't think that we can compromise in speed at all. The only >> possibility is to document the code, and keep it modular, to make it >> as easy to maintain as possible. > > > Another way to look at the problem. > > Why is there a bitmap in the Table object instead of a bool in the field > object? The field object is already being looped through in almost all > of the contexts. The bitmap blows through the interface (and heck... > notice the pointer swap bits that are done for filesort). > > This is one of the cases where I think the interface should be done > differently to solve the problem. Even if bitset could be done faster... > was it the right solution?
I have to second that. The way data is structured significantly affects performance. Generally, two pieces of information should only be separated if they make sense (from an implementation perspective) as separate entities. In this case I don't think they do, but there are some corners I don't know to well, such as the optimizer, which uses the bitsets as well. Just my few cents, Mats Kindahl > > Cheers, > -Brian > > > > _______________________________________________ > Mailing list: https://launchpad.net/~drizzle-discuss > Post to : [email protected] > Unsubscribe : https://launchpad.net/~drizzle-discuss > More help : https://help.launchpad.net/ListHelp -- Mats Kindahl Senior Software Engineer Database Technology Group Sun Microsystems _______________________________________________ Mailing list: https://launchpad.net/~drizzle-discuss Post to : [email protected] Unsubscribe : https://launchpad.net/~drizzle-discuss More help : https://help.launchpad.net/ListHelp
