Sat Aug 7 16:10:00 GMT 2004
On Sat, 07 Aug 2004 15:25:05 +0100, Nathan Sidwell <email@example.com> wrote:
> I rejected the idea of
> #define build_int_2 (x, y) build_int_cst (NULL, x, y)
> as I will be changing the API of build_int_cst, in that the returned
> node might be shared, and wanted to examine every use, and flag to
> others that it is now different.
Makes sense. What do you suggest for creating a new INTEGER_CST with the
same value but different type?
build_int_cst (newtype, TREE_INT_CST_LOW (old), TREE_INT_CST_HIGH (old))
seems a bit verbose.
More information about the Gcc-patches