On 07/22/2016 01:14 PM, Martin Sebor wrote:
> r231665 committed in the 6.0 cycle tightened up the checking of
> flexible array members in C++ with the goal of rejecting code
> that could lead to hard to find bugs, while at the same time
> accepting (with a warning) benign extensions also accepted in
> C mode.
>
> c++/71912 is a complaint that G++ rejects one such extension:
> defining structs with flexible array members in a union.  When
> reviewing the code that causes it to be rejected I realized
> that it was, in fact, meant to be accepted and is rejected due
> to a bug.
>
> The attached patch fixes that bug and implements more robust
> checking for this class of problems.  As in C mode, it accepts
> "benign" instances of this idiom (with a pedantic warning) while
> still rejecting the erroneous cases.
>
> There are outstanding problems with flexible array members in
> C++, such as c++/68489 - arrays of flexible array members are
> silently accepted.  In a follow-on patch I will address at
> least a part of the problem.  I submit this one separately
> since it's considered a regression and as such might be
> a candidate for backporting to the 6.x branch.
>
> Is this patch okay for trunk?  For 6.x?

I had looked at the wrong set of test results before posting
the first patch and missed a bunch of test suite failures due
to a bug that let the find_flexarray function descend into
static data members when looking for flexible array members,
causing infinite recursion.

The attached patch fixes that, adds a new test to exercise
the previously untested static data members, and annotates
an invalid regression test with a gd-error directive.

Martin