On 11/26/14 15:33, Jakub Jelinek wrote: > On Wed, Nov 26, 2014 at 09:41:16AM +0800, Chen Gang wrote: >> On 11/26/14 8:31, Joseph Myers wrote: >>> On Wed, 26 Nov 2014, Chen Gang wrote: >>> >>>> + gcc_assert (wi::fits_to_tree_p (value, char_type_node) >>>> + || wi::fits_to_tree_p (value, short_integer_type_node) >>>> + || wi::fits_to_tree_p (value, integer_type_node) >>>> + || wi::fits_to_tree_p (value, long_integer_type_node) >>>> + || wi::fits_to_tree_p (value, long_long_integer_type_node)); >>> >>> It doesn't make sense to check for char or short, since you can't write a >>> constant of one of those types. And it doesn't make sense to check for >>> int or long when checking for long long, as the ranges of int and long are >>> subsets of that of long long. So just check long long here. >>> >>>> + buf = (char *) alloca (strlen (macro) + vlen + extra); >>>> + >>>> + sprintf (buf, "%s=%s"HOST_WIDE_INT_PRINT_DEC"%s%s", > > Oh, and please use spaces around HOST_WIDE_INT_PRINT_DEC etc., > with C++11 user defined literals it is always better to have whitespace > in there. >
OK, thanks, I shall notice about it when send patch v4. Thanks. -- Chen Gang Open, share, and attitude like air, water, and life which God blessed