Comment 0 for bug 618311

 [Running ./run_tests from repo]

This error won't happend all the time, actually just 1/20 of the time.

 Why this happens ?

 Because in the code, we have at a certain point this :

for temp in xrange(1, int(random.random() * 20)):
    backend.fake_add_random_task()

 The problem here is that if int(random.random() * 20) is 1. There will be no tasks.

[ Same behaviour as :

i = 0
while i < int(random()*2):
    print "1"

That will only print "1" once in every 2 runs.

This makes that 1/20 times, this test result isn't correct.

Also in the patch, I also made some other improvements in the code of that module. (Using randint(start, limit) instead of int(random.random()*limit)

 Hope the patch looks fine, I'm not very good in publishing my work. (I don't use patches or bazaar very often.)