Making more builtins fold-able, and handling side-effects in args

[Kaveh R. Ghazi]

 > From: kenner@vlsi1.ultra.nyu.edu (Richard Kenner) 
 > 
 > The type of a COMPOUND_EXPR matters since it might be the operand of a
 > PLUS_EXPR.  The type of the first operand should be the same as the
 > type of the COMPOUND_EXPR (so an optimizer could replace it with the
 > first operand if it could determine that the second operand does not
 > actually have any side effects). I agree the type of the second
 > operand does not matter.

Agreed, except I think a reading of the comments for COMPOUND_EXPR in
tree.def says the reverse WRT which arg is ignored and which is
retained during folding.

		--Kaveh
--
Kaveh R. Ghazi			Engagement Manager / Project Services
ghazi@caip.rutgers.edu		Qwest Internet Solutions