I'm comparing some database system, amongst others Berkeley DB. So I wrote an application that fills random data into the different database systems. To be able to compare the database systems it is important that the random data is equal at every pass --- therefore I initialize srand() with a fixed seed.

This works e.g. with SQLite. The random data is equal every time I run the application. But with Berkeley DB the random data is different anytime. How could that be?? Does Berkeley DB initialize srand() with another seed?

* gerol:

Write your own random number generator, for instance (completely
untested, but the magic number 134775813 shouldn't be too bad):

uint32_t my_seed = 1;

uint32_t my_random(void)
{
my_seed = 134775813 * my_seed + 1;
return my_seed;
}

Then you need not care if anybody else calls rand().

I do not know explicitly if the BDB uses rand() (though the source code
is there, you could probably search) but I do have a test framework
that depends on somethign similar, and have had no issue with it.

Since my test framework is multi-threaded, and rand() is not required
to be threadsafe (according to the ANSI/ISO standards), I do have it
resource-locked, however.