>>The setting for synchronous is basically what level of safety net do you want if it dies in the middle of something. Setting it to off shouldn't cause any corruption if things go well, it should only come into play if you saw errors or didn't close things down correctly etc.
You're right, my Python code was ended manually sometime for it's taking too much time to finish the INSERT/UPDATE operation, or the image data I get is wrong. >>The unique index you declared is redundant by the way, declaring those three fields as the primary key makes a unique index already to keep track of that. Did you intend to make that on the retry table? The redundant map_index is an attempt to improve insert speed, as I learned drop index before insert operation is a better way to go. But now with my data growing so huge, drop/rebuild index also takes quite a long time, and I never choice to drop then create this index anymore, just leave the index there. Does it still effect my operation and I should just drop it? And if I want to speed up insert operation further more, what measures should I consider? I'v already set synchronous and journal_mode off, use transaction and prepared statement, but when insert data there's still 3 times' speed difference between my code and the *attach-insert* method. -- Sent from: http://sqlite.1065341.n5.nabble.com/ _______________________________________________ sqlite-users mailing list sqlite-users@mailinglists.sqlite.org http://mailinglists.sqlite.org/cgi-bin/mailman/listinfo/sqlite-users
