Re: GNAT Optimization of Constant Expressions

["Randy Brukardt" <randy@rrsoftware.com>]

"Florian Weimer" <fw@deneb.enyo.de> wrote in message
news:87k5v631kp.fsf@mid.deneb.enyo.de...
...
> It's more likely that the run-time library lacks a few Pure_Function
> pragmas, so the compiler does not know that it's a constant
> expression.  Ideally, GNAT would translate the function calls to GCC
> built-in calls, so that they can be expanded at compile time.  This
> optimization probably kicks in for C and Fortran, whose front ends are
> more closely aligned with the rest of the compiler.

That doesn't match the OP's original description of the problem. He said
that declaring constants in front of the expression made it run fast, even
if they were *not used* in the expression in the loop. That could only
happen if the compiler did a common subexpression elimination in that case,
and a common subexpression elimination is *only* valid on a function call if
it is Pure. (Other functions could have significant side effects that
eliminating the extra calls would make the program incorrect, for example if
the function is a random number generator.)

I think it is more likely that it didn't recognize the functions as being
worth hoisting (there is only one instance of each in the original program)
or it didn't do so because of other concerns. In thinking about it now, the
former is probably more likely.

                         Randy.