Date: Wed, 27 Oct 1999 16:01:18 +0200 (IST) From: Eli Zaretskii X-Sender: eliz AT is To: Laurynas Biveinis cc: djgpp-workers AT delorie DOT com Subject: Re: gcc-2.95.2 In-Reply-To: <3816FF74.6CCD5AC@softhome.net> Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Reply-To: djgpp-workers AT delorie DOT com X-Mailing-List: djgpp-workers AT delorie DOT com X-Unsubscribes-To: listserv AT delorie DOT com Precedence: bulk On Wed, 27 Oct 1999, Laurynas Biveinis wrote: > IMHO, strict aliasing is A Good Thing and breaking the code once and > forever is better than supporting it for years and years... I agree in principle. But this is the same issue as with the fflush-before-conio-functions problem: you can't educate people by forcing them to adhere to standards, no matter how Good those standards are. In the end, we wind up answering all those FAQs ourselves, and people don't learn anyway... > -fstrict-aliasing > was present in EGCS from 1.1, and people were warned, 'this optimization will > be on by default starting from next EGCS version...'. But it seems that > only few people actually tried it with their code to see if it still works. It is nevertheless significant that the GCC maintainers turned that option off by default, even though they held (and still hold) the same views as you do. Evidently, ``by popular demand'' still counts with them. > I don't think that many people will fix their code now, they will complain > about this optimization in the next GCC version. So this does not > solve issue. I'm merely arguing that DJGPP has no reasons to be more zealous in this respect than the core GCC distribution is. Strict aliasing is not yet part of the adopted C Standard; when it will be, it will be another story.