bpo-44954: Fix wrong result in float.fromhex corner case #27834
Merged
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
mdickinson
added
needs backport to 3.9
needs backport to 3.10
mdickinson
commented
Aug 19, 2021
serhiy-storchaka
approved these changes
Aug 20, 2021
| # Regression test for a corner-case bug reported in b.p.o. 44954 | ||
| self.identical(fromHex('0x.8p-1074'), 0.0) | ||
| self.identical(fromHex('0x.80p-1074'), 0.0) | ||
| self.identical(fromHex('0x.81p-1074'), TINY) |
Member
There was a problem hiding this comment.
I tried to find what is the minimal example for non-zero result, but seems it works for arbitrary number of zeroes between 8 and 1: 0x.800000000000000000000...0000000000000000001p-1074.
Member
Author
There was a problem hiding this comment.
Thanks, that's reassuring, and as it should be: 0x.8p-1074 is the exact threshold for rounding, so if it's larger than 0x.8p-1074 by even the tiniest amount, it should round up.
Contributor
|
Thanks @mdickinson for the PR 🌮🎉.. I'm working now to backport this PR to: 3.9, 3.10. |
|
GH-27854 is a backport of this pull request to the 3.10 branch. |
miss-islington
pushed a commit
to miss-islington/cpython
that referenced
this pull request
Aug 20, 2021
) (cherry picked from commit 60b93d9) Co-authored-by: Mark Dickinson <mdickinson@enthought.com>
|
GH-27855 is a backport of this pull request to the 3.9 branch. |
miss-islington
pushed a commit
to miss-islington/cpython
that referenced
this pull request
Aug 20, 2021
) (cherry picked from commit 60b93d9) Co-authored-by: Mark Dickinson <mdickinson@enthought.com>
mdickinson
added a commit
that referenced
this pull request
Aug 20, 2021
miss-islington
added a commit
that referenced
this pull request
Aug 20, 2021
Open
6 tasks
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
5 participants
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
This PR fixes a corner-case bug where the rounding logic in
float.fromhexresulted in a value that was rounded the wrong way. Assuming IEEE 754 binary64 floating-point, the bug can only occur in the case where:2**-107588is not preceded by any zerosExample bad inputs are
'0x.8p-1074','0x.80p-1074'and0x8p-1078. In these cases, the rounding logic tries to interpret thexcharacter as a hex digit, decides that as a hex digitxis odd, and so ends up rounding the value up to2**-1074, when it should have been rounded down to0.The case of a string that doesn't start with
0xis worse: for an input like.8p-1074or.80p-1074or8p-1078, the rounding logic examines the character before the beginning of the string to decide how to round.This bug affects all Python versions back to 2.7; the fix should be backported to 3.9 and 3.10, but I don't think it's critical enough to go into 3.10.0.rc2 - it can wait for 3.10.1.
https://bugs.python.org/issue44954