From mboxrd@z Thu Jan 1 00:00:00 1970 Date: Wed, 27 Dec 2000 15:37:13 -0800 From: Richard Henderson To: Alexandre Oliva Cc: sailer@ife.ee.ethz.ch, Kaoru Fukui , linuxppc-dev@lists.linuxppc.org, gcc@gcc.gnu.org Subject: Re: kerne-2.4.0-test13pre3 with gcc-2.97 Message-ID: <20001227153713.A30728@redhat.com> References: <20001223145258.Postino-022361@smtp01.highway.ne.jp> <3A4613FB.384BAD24@ife.ee.ethz.ch> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii In-Reply-To: ; from Alexandre Oliva on Sun, Dec 24, 2000 at 04:08:08PM -0200 Sender: owner-linuxppc-dev@lists.linuxppc.org List-Id: On Sun, Dec 24, 2000 at 04:08:08PM -0200, Alexandre Oliva wrote: > > Which is not backward compatible, unfortunately. > > Indeed. I wish we had maintained the old syntax for > backward-compatibility, even if with a deprecation warning. We *have* maintained the old syntax. But there are some quirky differences from the old behaviour wrt initialization and sizeof that need to be resolved. To wit: * The linux net/ code insists on initializing a zero-length array member with an empty initializer. We should silently drop the initialization in this case. * Several places in the linux kernel expect sizeof of a zero-length array to be zero. Typically this happens in some edge case like long stuff[N_WORDS-1] where a flexible array member isn't really wanted, but N_WORDS happens to be 1 for some configuration. Fixing this compatibility stuff is on my to-do list. r~ ** Sent via the linuxppc-dev mail list. See http://lists.linuxppc.org/