Looks great.
Applied in r141433!
On Oct 5, 2011, at 4:25 AM, Jim Goodnow II wrote:
> Hmm, how's this? About all that CXXConstructExpr currently does is invalidate
> arguments. The only difference between CXXConstructExpr and
> CXXTemporaryObjectExpr is some source type information and that
> CXXTemporaryObjectExpr's have none or more than one argument. So, this test
> verifies that the constructor is handled by the engine now and verifies that
> arguments are invalidated.
>
> - jim
>
> Index: test/Analysis/misc-ps-cxx0x.cpp
> ===================================================================
> --- test/Analysis/misc-ps-cxx0x.cpp (revision 141126)
> +++ test/Analysis/misc-ps-cxx0x.cpp (working copy)
> @@ -9,3 +9,27 @@
> *p = 0xDEADBEEF; // expected-warning {{null}}
> }
>
> +// Tests for CXXTemporaryObjectExpr.
> +struct X {
> + X( int *ip, int );
> +};
> +
> +// Test to see if CXXTemporaryObjectExpr is being handled.
> +int tempobj1()
> +{
> + int j;
> + int i;
> + X a = X( &j, 1 );
> +
> + return i; // expected-warning {{Undefined or garbage value returned to
> caller}}
> +}
> +
> +// Test to see if CXXTemporaryObjectExpr invalidates arguments.
> +int tempobj2()
> +{
> + int j;
> + X a = X( &j, 1 );
> +
> + return j; // no-warning
> +}
> +
>
>
> On 10/4/2011 3:21 PM, Ted Kremenek wrote:
>> Test case please. :)
>>
> <CXXTemporaryObjectTest.patch>
_______________________________________________
cfe-commits mailing list
[email protected]
http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits
