On Fri, 15 Jun 2018, Jakub Jelinek wrote:
> Hi!
>
> The match.pd framework uses op0 == op1 || operand_equal_p (op0, op1, 0)
> style comparisons heavily; the reason why we don't optimize away say
> volatile int i;
> ...
> int j = i - i;
> into int j = (i, 0);
> is only because both the C and C++
Hi!
The match.pd framework uses op0 == op1 || operand_equal_p (op0, op1, 0)
style comparisons heavily; the reason why we don't optimize away say
volatile int i;
...
int j = i - i;
into int j = (i, 0);
is only because both the C and C++ FE convert the volatile arguments to
non-volatile ones befor