[issue26163] FAIL: test_hash_effectiveness (test.test_set.TestFrozenSet)

Change by Roundup Robot : -- pull_requests: +5082 ___ Python tracker ___

[issue26163] FAIL: test_hash_effectiveness (test.test_set.TestFrozenSet)

Change by Raymond Hettinger : -- pull_requests: +5081 ___ Python tracker ___

[issue26163] FAIL: test_hash_effectiveness (test.test_set.TestFrozenSet)

Raymond Hettinger added the comment: New changeset e7dbd0658304f11daf103d56662656efecad006d by Raymond Hettinger (Miss Islington (bot)) in branch '3.6': bpo-26163: Frozenset hash improvement (GH-5194) (#5198)

[issue26163] FAIL: test_hash_effectiveness (test.test_set.TestFrozenSet)

Change by Raymond Hettinger : -- resolution: -> fixed stage: patch review -> resolved status: open -> closed ___ Python tracker

[issue26163] FAIL: test_hash_effectiveness (test.test_set.TestFrozenSet)

Raymond Hettinger added the comment: New changeset b44c5169f64178d2ff2914187b315549e7ab0cb6 by Raymond Hettinger in branch 'master': bpo-26163: Frozenset hash improvement (#5194) https://github.com/python/cpython/commit/b44c5169f64178d2ff2914187b315549e7ab0cb6

[issue26163] FAIL: test_hash_effectiveness (test.test_set.TestFrozenSet)

Change by Roundup Robot : -- pull_requests: +5052 ___ Python tracker ___

[issue26163] FAIL: test_hash_effectiveness (test.test_set.TestFrozenSet)

Change by Raymond Hettinger : -- pull_requests: +5048 ___ Python tracker ___

[issue26163] FAIL: test_hash_effectiveness (test.test_set.TestFrozenSet)

Raymond Hettinger added the comment: I'm getting a nice improvement in dispersion statistics by shuffling in higher bits right at the end: /* Disperse patterns arising in nested frozensets */ + hash ^= (hash >> 11) ^ (~hash >> 25); hash = hash *

[issue26163] FAIL: test_hash_effectiveness (test.test_set.TestFrozenSet)

Change by Raymond Hettinger : -- Removed message: https://bugs.python.org/msg310007 ___ Python tracker ___

[issue26163] FAIL: test_hash_effectiveness (test.test_set.TestFrozenSet)

Raymond Hettinger added the comment: Messages (3) msg309956 - (view) Author: Johnny Dude (JohnnyD) Date: 2018-01-15 01:08 When using a tuple that include a string the results are not consistent when invoking a new interpreter or process. For example

[issue26163] FAIL: test_hash_effectiveness (test.test_set.TestFrozenSet)

Serhiy Storchaka added the comment: As far as I understand, the problem is that the value obtained by XORing hashes of elements has nonuniform distribution. Further transformations, either linear or nonlinear, as well as adding length, don't change the fact that

[issue26163] FAIL: test_hash_effectiveness (test.test_set.TestFrozenSet)

Change by Raymond Hettinger : -- pull_requests: -5041 ___ Python tracker ___

[issue26163] FAIL: test_hash_effectiveness (test.test_set.TestFrozenSet)

Change by Raymond Hettinger : -- pull_requests: +5041 stage: -> patch review ___ Python tracker ___

[issue26163] FAIL: test_hash_effectiveness (test.test_set.TestFrozenSet)

Raymond Hettinger added the comment: Bringing this tracker item back to life. I've had more time to look at the this and the underlying problem was the high bits were not being shuffled back to impact lower bits. By fixing this, all of the tests produce a

[issue26163] FAIL: test_hash_effectiveness (test.test_set.TestFrozenSet)

Changes by Raymond Hettinger : -- resolution: -> wont fix status: open -> closed ___ Python tracker ___

[issue26163] FAIL: test_hash_effectiveness (test.test_set.TestFrozenSet)

Roundup Robot added the comment: New changeset e0f0211d314d by Raymond Hettinger in branch '3.6': Issue #26163: Disable periodically failing test which was overly demanding of the frozenset hash function effectiveness https://hg.python.org/cpython/rev/e0f0211d314d -- nosy: +python-dev

[issue26163] FAIL: test_hash_effectiveness (test.test_set.TestFrozenSet)

Raymond Hettinger added the comment: Rather than making futile further attempts to scramble the bits in the final hash to hide the effect of weak upstream hash inputs, I'm inclined to disable this particular test which was already a crazily harsh torture test. I suspect that even the tuple

[issue26163] FAIL: test_hash_effectiveness (test.test_set.TestFrozenSet)

Eric Appelt added the comment: Here are my test results - I'll re-describe the issue and test so people don't have to read through all the previous text and for completeness. - The basic test case ("letter test"): Consider the set of the first seven

[issue26163] FAIL: test_hash_effectiveness (test.test_set.TestFrozenSet)

Eric Appelt added the comment: I made a copy/paste error on the second to last paragraph of the previous comment, it should be: Rehashing with the configured algorithm to shuffle the XORed result of the entry and size hashes to finish computing the frozenset hash resulted in an ideal

[issue26163] FAIL: test_hash_effectiveness (test.test_set.TestFrozenSet)

Eric Appelt added the comment: If I understand the reporting properly all tests so far have used SipHash24: Python 3.7.0a0 (default:5b33829badcc+, Oct 30 2016, 17:29:47) [GCC 4.2.1 Compatible Apple LLVM 7.3.0 (clang-703.0.31)] on darwin Type "help", "copyright", "credits" or "license" for more

[issue26163] FAIL: test_hash_effectiveness (test.test_set.TestFrozenSet)

Changes by Christian Heimes : -- nosy: +christian.heimes ___ Python tracker ___ ___

[issue26163] FAIL: test_hash_effectiveness (test.test_set.TestFrozenSet)

Serhiy Storchaka added the comment: Eric, did you tested with FNV or SipHash24 hashing algorithm? Using standard Python hashing algorithm adds hash randomization for frozensets. This is worth at least be mentioned in What's New document. -- ___

[issue26163] FAIL: test_hash_effectiveness (test.test_set.TestFrozenSet)

Changes by Raymond Hettinger : -- assignee: -> rhettinger ___ Python tracker ___

[issue26163] FAIL: test_hash_effectiveness (test.test_set.TestFrozenSet)

Eric Appelt added the comment: Ugh... so I think I made a mental error (i.e. confused myself) in the last comment. The result looked a bit "to good to be true" and I think that there is at least one error in my thinking about the problem. I tried testing with the width set to 2 and then 1 as

[issue26163] FAIL: test_hash_effectiveness (test.test_set.TestFrozenSet)

Eric Appelt added the comment: I spent some time looking at the Objects/setobject.c code where the hashing scheme for frozensets, which essentially works by XOR-ing together the hashes from all its entries. Significant care is taken to shuffle bits at various stages, and the algorithm seems

[issue26163] FAIL: test_hash_effectiveness (test.test_set.TestFrozenSet)

Eric Appelt added the comment: I also looked at hashes of strings themselves rather than frozensets to check the hashing of strings directly. For example, n=3: ['', 'a', 'b', 'c', 'ab', 'ac', 'bc', 'abc'] rather than: [frozenset(), frozenset({'a'}), frozenset({'b'}), frozenset({'c'}),

[issue26163] FAIL: test_hash_effectiveness (test.test_set.TestFrozenSet)

Eric Appelt added the comment: I dug into this failure a little bit and verified that it is specifically the "letter_range" portion of the test that sporadically fails. The hash of any frozenset constructed from floats, ints, or the empty frozenset, as well as frozensets recursively

[issue26163] FAIL: test_hash_effectiveness (test.test_set.TestFrozenSet)

Tim Peters added the comment: I think Raymond will have to chime in. I assume this is due to the `letter_range` portion of the test suffering hash randomization dealing it a bad hand - but the underlying string hash is "supposed to be" strong regardless of seed. The

[issue26163] FAIL: test_hash_effectiveness (test.test_set.TestFrozenSet)

Martin Panter added the comment: I just got this failure again today. I think I have seen it once or twice before. Is the failure actually indicating a problem with Python, or is the test just too strict? It seems that it may be like a test ensuring that a random.randint(1, 100) is never

[issue26163] FAIL: test_hash_effectiveness (test.test_set.TestFrozenSet)

Serhiy Storchaka added the comment: I can stably reproduce this. PYTHONHASHSEED=36 ./python -m test.regrtest -vm test_hash_effectiveness test_set -- nosy: +serhiy.storchaka ___ Python tracker

[issue26163] FAIL: test_hash_effectiveness (test.test_set.TestFrozenSet)

Berker Peksag added the comment: I just saw the same failure on s390x RHEL 3.x: http://buildbot.python.org/all/builders/s390x%20RHEL%203.x/builds/1004/steps/test/logs/stdio == FAIL: test_hash_effectiveness

[issue26163] FAIL: test_hash_effectiveness (test.test_set.TestFrozenSet)

STINNER Victor added the comment: as i expected, the bug disappeared. I'm not interested to investigate a random failure which only occurred once. I close the issue. -- resolution: -> out of date status: open -> closed ___ Python tracker

[issue26163] FAIL: test_hash_effectiveness (test.test_set.TestFrozenSet)

New submission from STINNER Victor: The buildbot "x86 Ubuntu Shared 3.x" build 12691 failed, the previous builds succeeded. Changes: * f97da0952a2ebe08f2e5999c7473c776c59f3a16 (issue #25982) * 775b74e0e103f816382a0fc009b6ac51ea956750 (issue #26107)