On Jun 09, Giuseppe Cowo Corbelli <[EMAIL PROTECTED]> wrote: > IMHO _all_ indexes on big tables should not be created in the > beginning, but only when DB population is done.
You are right, and losing referential integrity is not a big issue: it can also make the whole script more robust (data is temporary stored in dictionary, and at insert-time they can be in the wrong order). Moreover we take input from a set of files that are assumed to be already coherent... and after all we're not running a nuclear plant! :-) In other news: I've set my out-of-the-box MySQL to use InnoDB tables. Very cool: inserting the data take only 15/20% longer, while the creation of the indexes... well, takes forever from what I know! evil-<g> After 6 hours running and after it has eat the whole hard disk I've had to shoot it, for his own good. :-) I'm more and more impressed by the amazing world of databases. ;-) -- Davide Alberani <[EMAIL PROTECTED]> [PGP KeyID: 0x465BFD47] http://erlug.linux.it/~da/ _______________________________________________ Imdbpy-devel mailing list Imdbpy-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/imdbpy-devel
