https://gcc.gnu.org/bugzilla/show_bug.cgi?id=125695
Bug ID: 125695
Summary: Big_Exp special case for (+/-)2 ** K is broken for -2
and even K
Product: gcc
Version: 16.1.1
Status: UNCONFIRMED
Severity: normal
Priority: P3
Component: ada
Assignee: unassigned at gcc dot gnu.org
Reporter: liam at liampwll dot com
CC: dkm at gcc dot gnu.org
Target Milestone: ---
This comes from
https://www.reddit.com/r/ada/comments/1u17ucp/clueless_about_pow_operator_in_gnat/
To_Big_Integer (-2) ** 2 incorrectly produces a result of 4 due to a missing
parity check. In Big_Exp note the parity check in the third Normalize call
which is missing from the fourth.
To_Big_Integer (-2) ** 1 is also broken since in the code above Normalize is
called without a Neg argument, probably not worth a separate issue since it's a
tiny fix:
-- X ** 1 is X
when 1 =>
return Normalize (X.D);
Both bugs have been present since Big_Integers was added.
