On 9/26/13 12:27 PM, Daniel Macks wrote: > On Thu, 26 Sep 2013 12:07:28 -0700, Alexander Hansen > <alexanderk.han...@gmail.com> wrote: > >> The symptom is as below: >> >> -------- Original Message -------- >> Subject: [Fink-beginners] installing octave >> Date: Wed, 25 Sep 2013 02:04:32 +0200 >> From: Werner Bäni <wba...@sunrise.ch> >> To: fink-beginn...@lists.sourceforge.net >> CC: d...@fincproject.org >> >> Hello >> >> When trying to install octave I get the following error message: >> >> imake -DUseInstalled -I/sw/lib/X11/config >> imake: No such file or directory >> imake: Cannot exec /usr/bin/llvm-cpp-4.2. Stop. imake: Exit code 1. >> Stop. >> ----------------------------------- >> >> This is fixed in the 10.9 packaging for xmkmf (since that also uses >> Xcode 5), via the use of a Fink package for Apple's llvm-gcc-4.2 in >> the guise of the "llvm-gcc42" package. Currently this is currently >> only available for Xcode >= 4.7, but in the interests of not making >> xmkmf support any more complicated than necessary I'd propose widening >> the availability of llvm-gcc42 to include earlier Xcodes so that we >> can build that on 10.7, and then we can use the same xmkmf packaging >> for 10.7-10.9. > > Can we *please* just kill off xmkmf? :) > > Okay, seriously, I'm not sure how it "hardcodes" this path/compiler, but > seems like it could be patched to use something more generic, or a > special wrapper that is more generic. > dan > > -- > Daniel Macks > dma...@netspace.org > >
I did a test where I modified xmkmf-compiler.patch not to use just "llvm-cpp-4.2" and "llvm-gcc-4.2" rather than specifying the full paths. That seems to provide the minimum amount of fuss here. We've got another problem, though. The imake executable from xmkmf really needs to know where the build tools are at runtime. Our "llvm-gcc" virtual package will account for the presence or absence of /usr/bin/llvm-*. In principle we could use a Depends: llvm-gcc | llvm-gcc42 to make sure that people with fresh Xcode 5 installs will have %p/bin/llvm-*, and people who currently have /usr/bin/llvm-* won't be forced to install llvm-gcc42. Unfortunately, the llvm-gcc virtual package is BuildDependsOnly (my bad there, I'm afraid). Since we just released a new fink, we could in principle let xmkmf fail validation for now, pending the next fink release. Thoughts? I'd like to update this soonish. -- Alexander Hansen, Ph.D. Fink User Liaison My package updates: http://finkakh.wordpress.com/ ------------------------------------------------------------------------------ October Webinars: Code for Performance Free Intel webinars can help you accelerate application performance. Explore tips for MPI, OpenMP, advanced profiling, and more. Get the most from the latest Intel processors and coprocessors. See abstracts and register > http://pubads.g.doubleclick.net/gampad/clk?id=60133471&iu=/4140/ostg.clktrk _______________________________________________ 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
