Hello,
This patch isn't quite right, consider the following:
struct BaseWithFlexible { int a[]; };
struct NotSoEmptyBase { int b; };
struct Derived : BaseWithFlexible, NotSoEmptyBase { };
In this example, 'Derived' has no fields but the flexible array member
still interferes with 'NotSoEmptyBase'
another test case:
struct BaseWithFlexible { int a[]; };
struct Derived : virtual BaseWithFlexible { };
In this case, the flexible array member would index into the
v-table/vb-table pointer of 'Derived'.
final test case:
struct BaseWithFlexible { int a[]; };
struct Derived : BaseWithFlexible { virtual ~Derived() {} };
For this final case, the flexible array member would index into the
v-table/vf-table of 'Derived'.
--
David Majnemer
On Fri Feb 21 2014 at 10:47:39 PM, Curtis Dunham <[email protected]>
wrote:
Hi all,
Please consider for review this patch.
If a class deriving from an base class with a flexible array has no new
data members of its own, the issue of the data member offset never comes
into play. I have a large, mature codebase that tripped over this recently
added, but slightly overzealous, semantic check.
thanks,
Curtis Dunham
_______________________________________________
cfe-commits mailing list
[email protected]
http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits
_______________________________________________
cfe-commits mailing list
[email protected]
http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits
