site_archiver@lists.apple.com Delivered-To: darwin-dev@lists.apple.com Nice sentiment, but none of the alternatives to autoconf+autoconf+(g)libtool have proven themselves to be anything but a massive pain in the rear, tending to be fuelled entirely by the arrogant belief that half the stuff autoconf does is unnecessary, rather than simply unnecessary _on their platform of choice_. More to the point, that doesn't actually -help- any given that it's been a de facto standard for longer than Mac OS X has been publicly released ;) -- Terry _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev@lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.appl... On Aug 31, 2009, at 10:18 AM, Mo McRoberts <lists@ilaven.net> wrote: On Mon, Aug 31, 2009 at 17:54, Terry Lambert<tlambert@apple.com> wrote: Maybe people should quit using configure entirely. Just saying... The problem with configure is that it localizes code to a given platform, rather than abstracting the platform differences from the code. It's a constant struggle to keep platform dependencies from creeping into the code. It's also a pain when it comes to building universal binaries, unless you are willing to Lipo together individual binaries into a single image. Add to that the nasty tendency to check for intefaces using the linker instead of <unistd.h> feature test macros, and you have a recipe for failure, even if you ignore the cross-target byte order and other issues that plague the code that depends on it. Frankly, xmkmf should have won; at least it abstracted architecture dependence based on macros, and let you safely cross-build. This email sent to site_archiver@lists.apple.com