Re: [PATCH 1/4] libstdc++: Fix use of is_nothrow_assignable_v in

2020-03-03 Thread Jonathan Wakely
On 03/03/20 11:30 -0500, Patrick Palka wrote: We are passing a value type as the first argument to is_nothrow_assignable_v, but the result of that is always false. Since this predicate is a part of the condition that guards the corresponding optimizations for these algorithms, this bug means the

[PATCH 1/4] libstdc++: Fix use of is_nothrow_assignable_v in

2020-03-03 Thread Patrick Palka
We are passing a value type as the first argument to is_nothrow_assignable_v, but the result of that is always false. Since this predicate is a part of the condition that guards the corresponding optimizations for these algorithms, this bug means these optimizations are never used. We should be p