On 10/20/16 10:05, Markus Trippelsdorf wrote:
>
> Thanks for the commit. But I think the comment is wrong:
>
> + /* We will only warn on unsigned shifts here, because the majority of
> ^^
> This should be »signed«.
>
Oops. Thanks for noticing.
This is what I am going to check in as obvious:
--- ChangeLog (revision 241376)
+++ ChangeLog (working copy)
@@ -1,3 +1,7 @@
+2016-10-20 Bernd Edlinger <bernd.edlin...@hotmail.de>
+
+ * c-common.c (c_common_truthvalue_conversion): Fix the comment.
+
2016-10-20 Jason Merrill <ja...@redhat.com>
* c-cppbuiltin.c (c_cpp_builtins): Update __cpp_concepts value.
Index: c-common.c
===================================================================
--- c-common.c (revision 241376)
+++ c-common.c (working copy)
@@ -3328,7 +3328,7 @@
TREE_OPERAND (expr, 0));
case LSHIFT_EXPR:
- /* We will only warn on unsigned shifts here, because the majority of
+ /* We will only warn on signed shifts here, because the majority of
false positive warnings happen in code where unsigned arithmetic
was used in anticipation of a possible overflow.
Furthermore, if we see an unsigned type here we know that the
result of the shift is not subject to integer promotion rules. */
Bernd
