It would be gread if you could access the random seed inside the tests.

In this way, one could generate pseudo random numbers out of this seed and then 
test the functions with random inputs.

The problem now is, that when we test something with a random seed like 
"time(NULL)", and we have an error, it is really hard to see, what the actual 
input arguments to the function where.

However, if this is connected to the googletest random seed, one can reproduce 
it exactly like the order of the gtest_shuffle, and has really deterministic 
and repeatable results based on the (random) seed...