Date: Wed, 25 Jun 2003 13:26:53 -0500 From: Juli Mallett <jmallett@FreeBSD.org> To: "David E. O'Brien" <obrien@FreeBSD.org> Cc: cvs-all@FreeBSD.org Subject: Re: cvs commit: src/include stdlib.h Message-ID: <20030625132653.A11826@FreeBSD.org> In-Reply-To: <200306251811.h5PIBWxs078351@repoman.freebsd.org>; from obrien@FreeBSD.org on Wed, Jun 25, 2003 at 11:11:32AM -0700 References: <200306251811.h5PIBWxs078351@repoman.freebsd.org>
next in thread | previous in thread | raw e-mail | index | archive | help
* "David E. O'Brien" <obrien@FreeBSD.org> [ Date: 2003-06-25 ] [ w.r.t. cvs commit: src/include stdlib.h ] > obrien 2003/06/25 11:11:32 PDT > > FreeBSD src repository > > Modified files: > include stdlib.h > Log: > Don't blindly provide alloca() for all compilers -- it is too implementation > dependent. Instead provide one for GCC & Intel's GCC copy and one for lint. > Anyone using any other translator tool needs to look closely at how that tool > can handle alloca. Please make the lint case the default case. The error is silly. If this is how we're going to be, I'm going to go back to IRIX 6.2. At least there, when I munged the headers once, they didn't keep making me re-munge them. A compiler which needs a builtin one should inline it. GCC does unless you request strict-strict-strict conformance. Personally, I have a C-based alloca(3) that I use, it's slow, but I do not care. I also use a variety of odd C translators, and don't really appreciate this at all. And don't tell me to add definitions for those compilers along with lint. This isn't error worthy, especially not in stdlib.h. Maybe in alloca.h. This breaks compiling even things which don't need alloca. Adding more definitions for the default case is akin to "ported" vs. "portable". Thanx, juli. -- juli mallett. email: jmallett@freebsd.org; efnet: juli;
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20030625132653.A11826>