Thanks to visit codestin.com
Credit goes to github.com

Skip to content

bpo-40282: Allow random.getrandbits(0) #19539

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom
Merged
merged 7 commits into from
Apr 17, 2020
Merged

bpo-40282: Allow random.getrandbits(0) #19539

merged 7 commits into from
Apr 17, 2020

Conversation

pitrou
Copy link
Member

@pitrou pitrou commented Apr 15, 2020 •
edited by bedevere-bot
Loading

@pitrou pitrou requested a review from rhettinger as a code owner April 15, 2020 08:24
@pitrou pitrou requested a review from mdickinson April 15, 2020 08:24
pitrou
pitrou commented Apr 15, 2020
View reviewed changes
Lib/test/test_random.py
cum |= v
cpl_cum |= all_bits ^ v
self.assertEqual(cum, all_bits)
self.assertEqual(cpl_cum, all_bits)
Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Note I've enhanced this test a bit to check that all bits take both values 0 and 1 accross 100 draws.

serhiy-storchaka
serhiy-storchaka reviewed Apr 15, 2020
View reviewed changes
serhiy-storchaka
serhiy-storchaka reviewed Apr 15, 2020
View reviewed changes
Lib/random.py
@@ -261,6 +261,8 @@ def randint(self, a, b):
def _randbelow_with_getrandbits(self, n):
"Return a random int in the range [0,n). Raises ValueError if n==0."

if n == 0:
raise ValueError("Boundary cannot be zero")
getrandbits = self.getrandbits
k = n.bit_length() # don't use (n-1) here because n can be 1
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

We can use (n-1) now.

Suggested change
k = n.bit_length() # don't use (n-1) here because n can be 1
k = (n-1).bit_length()

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I think we want to ensure that the values produced for a given seed don't change, so we probably can't change this anymore.

Lib/random.py Outdated
Comment on lines 264 to 265
if n == 0:
raise ValueError("Boundary cannot be zero")
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Maybe use assert?

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This is the same error as in the other implementation (the one without getrandbits).

Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Seems this check causes a performance regression.

Copy link
Contributor

@aeros aeros Apr 15, 2020 •
edited
Loading

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Is there any situation where a user could specifically encounter the above ValueError from the public methods? If not, I think an assertion would make more sense here; seeing as it's a private member that's only being used from within the random module. IMO, the same applies to _randbelow_without_getrandbits().

Also, how significant is the performance regression @serhiy-storchaka?

serhiy-storchaka
serhiy-storchaka reviewed Apr 15, 2020
View reviewed changes
@rhettinger
Copy link
Contributor

This makes shuffle() about 6% slower:

python -m timeit -r11 -s 'from random import shuffle' -s 's=list(range(1000))' 'shuffle(s)'

vstinner
vstinner reviewed Apr 15, 2020
View reviewed changes
Copy link
Member

@vstinner vstinner left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

random.Random has a surprising API. The base class is Mersenne Twister which implements getrandbits(). Inheritance is non-trivial. Does your change work if a subclass implements _randbelow()? What if it implements random()? See Random.__init_subclass__().

@serhiy-storchaka
Copy link
Member

_randbelow() is not designed for overriding. If there is a problem with this, it can be made private (__randbelow()).

If the subclass implements random() but not getrandbits(), its methods will use random() instead of getrandbits().

rhettinger
rhettinger reviewed Apr 15, 2020
View reviewed changes
@pitrou pitrou force-pushed the bpo40282-getrandbits0 branch from 9ccb7bc to 3b86eb8 Compare April 17, 2020 16:26
@pitrou
Copy link
Member Author

pitrou commented Apr 17, 2020

I rebased and applied two trivial changes from reviewing. Will merge if CI is green. Thanks everyone!

@pitrou pitrou force-pushed the bpo40282-getrandbits0 branch from 3b86eb8 to c4673f0 Compare April 17, 2020 17:11
@pitrou pitrou merged commit 75a3378 into python:master Apr 17, 2020
@pitrou pitrou deleted the bpo40282-getrandbits0 branch April 17, 2020 17:32
@vstinner
Copy link
Member

vstinner commented Apr 20, 2020 •
edited by bedevere-bot
Loading

random.Random has a surprising API. The base class is Mersenne Twister which implements getrandbits(). Inheritance is non-trivial. Does your change work if a subclass implements _randbelow()? What if it implements random()? See Random.init_subclass().

I created bpo-40346: "Redesign random.Random class inheritance".

CuriousLearner added a commit to CuriousLearner/cpython that referenced this pull request May 29, 2020
@CuriousLearner
Merge branch 'master' into bpo-26543
e4d193d 
* master: (1985 commits)
  bpo-40179: Fix translation of #elif in Argument Clinic (pythonGH-19364)
  bpo-35967: Skip test with `uname -p` on Android (pythonGH-19577)
  bpo-40257: Improve help for the typing module (pythonGH-19546)
  Fix two typos in multiprocessing (pythonGH-19571)
  bpo-40286: Use random.randbytes() in tests (pythonGH-19575)
  bpo-40286: Makes simpler the relation between randbytes() and getrandbits() (pythonGH-19574)
  bpo-39894: Route calls from pathlib.Path.samefile() to os.stat() via the path accessor (pythonGH-18836)
  bpo-39897: Remove needless `Path(self.parent)` call, which makes `is_mount()` misbehave in `Path` subclasses. (pythonGH-18839)
  bpo-40282: Allow random.getrandbits(0) (pythonGH-19539)
  bpo-40302: UTF-32 encoder SWAB4() macro use a|b rather than a+b (pythonGH-19572)
  bpo-40302: Replace PY_INT64_T with int64_t (pythonGH-19573)
  bpo-40286: Add randbytes() method to random.Random (pythonGH-19527)
  bpo-39901: Move `pathlib.Path.owner()` and `group()` implementations into the path accessor. (pythonGH-18844)
  bpo-40300: Allow empty logging.Formatter.default_msec_format. (pythonGH-19551)
  bpo-40302: Add pycore_byteswap.h header file (pythonGH-19552)
  bpo-40287: Fix SpooledTemporaryFile.seek() return value (pythonGH-19540)
  Minor modernization and readability improvement to the tokenizer example (pythonGH-19558)
  bpo-40294: Fix _asyncio when module is loaded/unloaded multiple times (pythonGH-19542)
  Fix parameter names in assertIn() docs (pythonGH-18829)
  bpo-39793: use the same domain on make_msgid tests (python#18698)
  ...
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers

@vstinner vstinner vstinner left review comments

@rhettinger rhettinger rhettinger approved these changes

@serhiy-storchaka serhiy-storchaka serhiy-storchaka left review comments

@aeros aeros aeros left review comments

@mdickinson mdickinson Awaiting requested review from mdickinson

Assignees
No one assigned
Labels
None yet
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.
8 participants
@pitrou @rhettinger @serhiy-storchaka @vstinner @mdickinson @aeros @the-knights-who-say-ni @bedevere-bot