This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [patch]: Add support of new __int128 type for targets having 128-bit integer scalar support
On Fri, 16 Apr 2010, Kai Tietz wrote:
> 2010/4/15 Joseph S. Myers <joseph@codesourcery.com>:
> > You don't appear to have addressed my previous review comments in this
> > patch version (for example, the point that __int128 should be handled like
> > int, void, _Bool etc., getting "two or more data types in declaration
> > specifiers" for use with those specifiers instead of having explicit
> > checks for all of them). ?Furthermore, there are changes to existing code
> > (patch hunks that just change whitespace, for example) that appear
> > irrelevant to the addition of __int128.
> >
> > --
> > Joseph S. Myers
> > joseph@codesourcery.com
> >
>
> Hmm, I just re-read my patch I sent and what white-space changes you
> are talking about? I don't see in this patch pure white-space changes.
See the @@ -8725,6 +8738,7 @@ hunk for example, inserting a blank line, or
@@ -9024,7 +9105,7 @@, changing indentation. I refer to
<http://gcc.gnu.org/ml/gcc-patches/2010-04/msg00354/int128doc.diff>.
> Also I modified in this patch that __int128 is handled like void,
> _Bool, int, float, etc. I don't see you point. Could it be that you've
If it's handled like them, you wouldn't need special diagnostics such as
"both %<__int128%> and %<char%> in declaration specifiers" because the
generic diagnostic I mentioned would have caught this earlier. And the
check in the RID_LONG case would go with all the other checks for
particular keywords such as _Bool, rather than before the "long long long"
check in your present patch; likewise for RID_SHORT and probably other
cases. The falling through from cts_int128 to cts_int is also suspicious
since almost none of the code is actually relevant to both cases.
--
Joseph S. Myers
joseph@codesourcery.com