-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 On 11/16/11 5:59 PM, Max Horn wrote: > > Am 16.11.2011 um 17:56 schrieb Alexander Hansen: > > [...] > >>> >> >> Right, and the point I was trying to make is that fink-0.31.4 >> *doesn't work* for people who do a fresh install of Xcode 4.2, >> because of the absence of gcc-4.2 in the PATH. Bootstrap >> immediately fails, as per >> >> http://pastebin.com/5ZndHJP1 >> >> I'm not exactly sure why falling back to gcc-4.2 was suggested as >> a "solution", given that it is _absent_ in Xcode 4.2. > > My guess would be that it was suggested and accepted as a solution > because at that time nobody was aware that gcc-4.2 is absent in > Xcode. E.g. I myself tested Xcode 4.2 on 10.6 and had no reason to > suspect that gcc-4.2 could be a "leftover" (I wish they would use a > real package managment system for their installers... dpkg would > not leave "traces" like that. Nasty). > > There, simple as that :). >
Yeah, true enough. I guess the philosophy is "We'll let you keep old stuff around if you want." > > The fact that it *is* absent is extremely troubling for multiple > reasons. On the one hand, the current bootstrap failure, of course > (which we can and need to fix by making a new release of fink > ASAP). > > But on the other hand, this also means that Xcode 4.2 on 10.6 now > contains no reliable compiler at all :-(. Just today I wasted an > hour tracking down a bug only to discover that I had recompiled > some code with "gcc", which is llvm-gcc, which causes gmp 5.0.2 > miscompilation, resulting in some gmp test code not terminating. > Yuck. And clang miscompiles other code... Anyway, so now we can > decide whether to make gcc = llvm-gcc-4.2 default, or clang... both > will cause pain for some. Yuck again. > > > > Bye, Max What exactly do we want to do to fix bootstrap? The quickest change to implement is probably to roll back the g(cc|++)-4.2 stuff, but I'm not sure whether or not that will be best in the long term. Something that occurred to me is: maybe we could have a couple of extra compiler wrappers in fink, e.g. "gcc-alt" and "g++-alt". On 10.5 and 10.6 with Xcode 3.2, those would just point to "gcc" and "g++", respectively. On 10.6 with Xcode 4.2 we'd have "gcc" -> llvm-gcc "gcc-alt" -> clang "g++" -> llvm-g++ "g++-alt" -> clang++ if we went with LLVM as the default. (and changed appropriately for 10.7 and if we decided to make clang the default) That way packages could just use, e.g. SetCC: gcc-alt if they don't get along with LLVM, and on 10.5 and 10.6/Xcode3 this would be effectively a no-op, and the normal gcc would be chosen. This would make life easier for maintainers. - -- Alexander Hansen, Ph.D. Fink User Liaison http://finkakh.wordpress.com/ -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.10 (Darwin) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/ iEYEARECAAYFAk7FwfwACgkQB8UpO3rKjQ/YoACeOK7Ay3nsMPFxX1ae7J6fdKdL h94AoKJvqKQX4E33jHr+QAYOg4LkOAe3 =xl/E -----END PGP SIGNATURE----- ------------------------------------------------------------------------------ All the data continuously generated in your IT infrastructure contains a definitive record of customers, application performance, security threats, fraudulent activity, and more. Splunk takes this data and makes sense of it. IT sense. And common sense. http://p.sf.net/sfu/splunk-novd2d _______________________________________________ Fink-devel mailing list Fink-devel@lists.sourceforge.net List archive: http://news.gmane.org/gmane.os.apple.fink.devel Subscription management: https://lists.sourceforge.net/lists/listinfo/fink-devel
