Chula comes with a guid generator class who's implementation is a bit naive.  Considering Python already has (probably much better) functionality builtin, I figured I'd see how fast they are.

Here's the way I tested them:

# Python imports
from random import randrange
from uuid import uuid1, uuid4
import time

# Chula imports
from chula.guid import guid

count = 10000

def timeit(fcn):
    def wrapper():
        start = time.time()
        fcn()
        print('%s %fs' % (fcn.__name__.ljust(20), time.time() - start))

    return wrapper

@timeit
def builtin_randrange():
    for i in xrange(count):
        x = randrange(100000000000000000000)

@timeit
def builtin_uuid1():
    for i in xrange(count):
        x = uuid1().hex

@timeit
def builtin_uuid4():
    for i in xrange(count):
        x = uuid4().hex

@timeit
def chula_guid():
    for i in xrange(count):
        x = guid()

if __name__ == '__main__':
    builtin_randrange()
    builtin_randrange()
    builtin_uuid1()
    builtin_uuid1()
    builtin_uuid4()
    builtin_uuid4()
    chula_guid()
    chula_guid()

Here are the results:

jmcfarlane@monster:~/Desktop$ python test.py 
builtin_randrange    0.053161s
builtin_randrange    0.053091s
builtin_uuid1        0.495005s
builtin_uuid1        0.487182s
builtin_uuid4        0.382016s
builtin_uuid4        0.380982s
chula_guid           1.156092s
chula_guid           1.154981s

To finish things off... this one really surprised me:

jmcfarlane@monster:~$ time for i in $(seq 10000); do uuidgen > /dev/null; done
real	0m22.086s
user	0m0.560s
sys	0m4.750s

So the moral of the story, is that Chula's guid generation isn't particularly fast, but shelling out to uuidgen is somehow even slower.