Re: RFA: PATCH to make fold_indirect_ref_1 fold more things

[Jason Merrill <jason@redhat.com>]

On 10/29/2010 11:25 AM, Richard Guenther wrote:
> So - what kind of testcases are you trying to handle?  (the middle-end will
> optimize this stuff keeping TBAA info correct during forwprop)

The RECORD_TYPE folding was for this testcase:

struct C { // literal type
   int m;
   int n;
   constexpr C(int m) : m(m), n(-m) {}
   constexpr bool is_neg() { return m < 0; }
};

constexpr bool check1(const C& c, int C:: *pm) { return c.*pm < 0; } // #1

constexpr bool check2(const C* pc, bool (C::*pm)() const) { return
(pc->*pm)(); } // #2

constexpr C c(-1);

static_assert(!check1(c, &C::n), "Error");
static_assert(check1(c, &C::m), "Error");

static_assert(check2(&c, &C::is_neg), "Error");

For constexpr we need to evaluate suitable constant expressions at 
parsing time so that we can use them as template arguments, array bounds 
and such like.

I guess I can just do this folding in the constexpr expander...

Jason