On Thursday 18 February 2010 01:39:17 Tom Hacohen wrote: > As I said, I've been working on opimd quite a bit lately, without > optimizations (i.e without a couple of sqlite tricks to make everything a > lot faster) contact resolving (I have ~200 numbers) without storing > redundant information in database takes approximately 0.2 seconds (measured > by "feel" nothing scientific). > I managed to make it take even less, using a > special querying method I added (query by type instead of fields), and as I > said, with proper sqlite tricks, I bet I can make it take even less.
That looks nice. A query on a single phone number on my Freerunner takes 0.7 seconds now, so when that's reduced to less than 0.2 seconds I'm sure I'll notice the difference. Any idea when this will be included in shr-testing? The program I used to test the speed is at http://shr.pastebin.com/m20d7690c > Those figures support what I said from the begining, there's no reason to > store redundant information in the database. Wait for more updates about > exact figures and improved performance, but I think this settles it, and I > hope you now agree there's no need for storing the normalized numbers in > the database. I did a quick query on the sqlite database directly (same query as in the test above), and it found an answer in 0.02 seconds CPU time. So indeed, the database design isn't causing the slowdown. -- Sybren A. Stüvel [email protected] http://stuvel.eu/
signature.asc
Description: This is a digitally signed message part.
_______________________________________________ Shr-devel mailing list [email protected] http://lists.shr-project.org/mailman/listinfo/shr-devel
