On Wed, 2004-06-02 at 01:53, Daniel K wrote: > Hi, > > The getpid() call on my Redhat9 linux box is causing > some trouble for SQLite. It's reasonably complicated, > but SQLite currently assumes that on linux the > getpid() function returns a different value in > different threads.
Have you tried using LD_ASSUME_KERNEL to get the older linuxthreads behavior? Anyway, it shouldn't cause any harm. SQLite only uses the pid to help work around problems with fcntl-locking (esp. when using NFS) and to seed the random number generator. > Not so for my setup! This is a bug in sqlite. A minor one- one that is only exploited when multiple threads attempt to open the same database. Don't do that. That's a sign of bad engineering anyway. Your writes are going to have the EXACT SAME bottleneck they would have without threading- and if your updates aren't happening often enough for that to bother you, then consider making copies of the sqlite database and having each sqlite thread open from a separate file. --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
