wasm-demo/demo/ermis-f/python_m/cur/0258

96 lines
2.2 KiB
Plaintext

From: atai-no-spam at glink.net.hk (>0 Remove -no-spam to reply)
Date: 29 Apr 1999 11:42:16 GMT
Subject: Shelve vs. Deepcopy
Message-ID: <slrn7igh7r.h49.atai-no-spam@aaa.tai.org>
Content-Length: 2001
X-UID: 258
Hi:
I have a class who contains reference to other classes. Now I would like
to test an instance of this class. The test would alter the instance but I
would like to retain this instance, so at first I make a deepcopy of this
instance and perform the test on the copy. However, it take quite a long
time to finish the test. Then I try to use shelve to do the same thing.
I.e. first store the instance on a dbm. Then the copy is obtain from
loading the dbm. And to my surprise, the latter method is much faster than
the first method. With some profiling, I saw that the number of functions
calls of the first method is at least twice that of the second one. The
following are two test programs to illustrate this. When the first one is
run, using shelve is faster. While in the second one, deepcopy is faster.
I hope I have make some sense in the above sentences.
Regards,
>0
test.py
Cut here -----------------------------------------------
import copy, shelve, time
class test:
def __init__(self, num):
self.a = num
self.b = (2,3,`num`)
self.c = (3,7, `num`)
class test1:
def __init__(self, obj):
self.dict = {(1,3):obj}
for i in range(100):
self.__dict__[(`i`,1)] = i
a = test(7)
b = test1(a)
dbase = shelve.open('tmp')
dbase['store'] = b
start = time.time()
for i in range(1000):
c = dbase['store']
stop = time.time()
print stop - start
start = time.time()
for i in range(1000):
c = copy.deepcopy(b)
stop = time.time()
print stop - start
dbase.close()
test2.py
Cut here-------------------------------------------------
import copy, shelve, time
class test:
def __init__(self):
for i in range(100):
self.__dict__[`i`] = i
a = test()
dbase = shelve.open('tmp')
dbase['store'] = a
start = time.time()
for i in range(1000):
c = dbase['store']
stop = time.time()
print stop - start
start = time.time()
for i in range(1000):
c = copy.deepcopy(a)
stop = time.time()
print stop - start
dbase.close()