site_archiver@lists.apple.com Delivered-To: darwin-dev@lists.apple.com On Jul 17, 2006, at 8:29 PM, Damien Sorresso wrote: -Eric _______________________________________________ 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 17 Jul, 2006, at 10:16 PM, Eric Albert wrote: On Jul 17, 2006, at 8:08 PM, Damien Sorresso wrote: No, they're two separate binaries. (Trust me there's a good reason for this.) I hate to say this, but I don't quite trust you on that. :) What's the reason? I've seen an awful lot of cases over the past year where folks thought they needed separate binaries but I've yet to see one which actually did. Maybe you'll be the first.... Because the application I'm writing has to run on all versions of Panther, down to 10.3.0. (Part of its function is to make sure the user is up-to-date on point releases.) To compile a universal binary, you need to use gcc 4.0, and in order to run applications compiled with gcc 4.0, you must be running 10.3.9. How's that work for you? :) Not too well. :) You only need to use gcc 4.0 for the Intel side, but the PowerPC side of your binary can use any Mach-O compiler at all, including gcc 2.95.3 or even CodeWarrior. If you're building both sides with Xcode, you can do this entirely within the Xcode UI: <http://developer.apple.com/documentation/DeveloperTools/Conceptual/ XcodeUserGuide/Contents/Resources/en.lproj/05_07_bs_building_product/ chapter_34_section_6.html> It might be interesting to build the PowerPC side of your binary with gcc 4.0 as a test case to make sure your dlsym call continued to work there. Even so, I'd still suggest running nm on both slices (whether two binaries or just one) to make sure the expected symbol shows up for both architectures. This email sent to site_archiver@lists.apple.com