Re: [Fink-devel] llvm/llvm-gcc42-2.5-1 on fink tracking

2009-02-09 Thread Peter O'Gorman
Jack Howarth wrote: > The libLTO.dylib in Xcode 3.2.1 only supports dead code > elimination at -O4 whereas the libLTO.dylib in llvm 2.5 > provides additional optiminizations like inlining across > code files. I am not sure about the utility of a fink providied libLTO. The llvm-gcc release notes

Re: [Fink-devel] llvm/llvm-gcc42-2.5-1 on fink tracking

2009-02-09 Thread Jack Howarth
Peter, You totally misunderstand the purpose of bundling the libLTO.dylib with the llvm-shlibs package. It is only there to give the user the option to manually replace Apple's libLTO.dylib with that from the llvm-2.5 sources if they want to explore the complete functionality of LTO. I never sai

Re: [Fink-devel] llvm/llvm-gcc42-2.5-1 on fink tracking

2009-02-09 Thread Jack Howarth
Peter, That is not what Chris Lattner said... http://lists.cs.uiuc.edu/pipermail/llvmdev/2009-January/019687.html Remember that the libLTO.dylib included in Xcode 3.1.2 is generated from the llvm source tree, specifically in llvm-2.5/tools/lto, and definitely is a work in progress. I would be

Re: [Fink-devel] llvm/llvm-gcc42-2.5-1 on fink tracking

2009-02-09 Thread Peter O'Gorman
Jack Howarth wrote: > Peter, >You totally misunderstand the purpose of bundling the > libLTO.dylib with the llvm-shlibs package. It is only there > to give the user the option to manually replace Apple's > libLTO.dylib with that from the llvm-2.5 sources if they > want to explore the complete f

Re: [Fink-devel] llvm/llvm-gcc42-2.5-1 on fink tracking

2009-02-09 Thread Peter O'Gorman
Jack Howarth wrote: > Peter, >That is not what Chris Lattner said... He says that /usr/bin/ld and /Developer/usr/bin/ld differ, which it seems, they do, ld -v for /Developer/usr/bin includes the string "llvm version ..., Apple Build ...", which does not appear for /usr/bin/ld -v. He also says