You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Describe the bug
The decimal code path of num_get::do_get records the initial 768 significant digits. If there are any remaining non-zero digits, it adds 1 to the last recorded digit (with carry). This is incorrect when the result is just below the rounding boundary.
statementreply
changed the title
<xlocnum>: Incorrect rounding when parsing long floating point numbers just below exact values under some non-default rounding modes
<xlocnum>: Incorrect rounding when parsing long floating point numbers just below midpoints
Feb 1, 2023
Describe the bug
The decimal code path of
num_get::do_get
records the initial 768 significant digits. If there are any remaining non-zero digits, it adds 1 to the last recorded digit (with carry). This is incorrect when the result is just below the rounding boundary.Command-line test case 1
Command-line test case 2
STL version
The text was updated successfully, but these errors were encountered: