On Tue, Sep 20, 2016 at 04:24:35PM +0200, Marek Polacek wrote:
> > (to skip over the koenig_p etc. cases) and fallthrough into the argument
> > handling and add another if (function == NULL_TREE) handling after that,
> > which would just build another internal call with the tsubsted arguments.
> I added the assert here, but I spent time implementing this, and it didn't
> really work, because without a testcase it was hard to say whether what I
> had was correct. E.g., how to determine the return type of the internal
> function, etc.
I agree that without the testcase it is hard and that assert is good enough
for now. For return type of the internal function, I'd just tsubst the
TREE_TYPE of the CALL_EXPR though.