On 31-mrt-2006, at 0:38, Christopher Barker wrote: > Ronald Oussoren wrote: >> I'd prefer not to have seperate Intel and PPC binaries. Would merging >> seperate >> Intel and PPC builds be of any use? > > Yes, though I'm not sure what you mean by merging. > > Do you mean lipo'ing them together? or having separate builds in > one *.mpkg?
The former, to seperate builds in one package wouldn't be to usefull as you mention below. > > Ideally, we could get it all truly Universal, then one could build > Universal Py2App bundles with it and all. > > I think the big trick with SciPy is the Fortran components. SciPy, > Fortran and gcc 4 have been having trouble for a while. If that's > fixed, > I have no idea about how to build Universal binaries with gfortran or > g95. Does anyone know? IIRC both are frontends for GCC. You could probably build them using Apple's version of GCC (downloadable from Apple's opensource site) and then get the multiple-arch bit for free. The other alternative is a script that compiles using a compiler for intel, then with a compiler for ppc and finally uses lipo to merge the two. > > > While I'm on the topic, I'm looking into building Universal > binaries of > some other packages that depend on other libs. For instance, > matplotlib > depends on libpng and libfreetype. In the past, I've built static libs > of those, and then linked MPL against them, getting a nice > self-contained package. However, I have no idea how I would go about > building Universal static libs, or two sets of libs, or???? That depends :-) For my universal python build you definitely want to end up with Universal static libs. If configure doesn't check for CPU features (like sizeof(int) and byteorder) you can most likely get away with adding 'CFLAGS="-arch i386 -arch ppc -isysroot ..." LDFLAGS="-arch i386 - arch ppc -isysroot ..."' to the arguments of configure. Otherwise you either have to patch the sources (like we did for python), or create two builds and them lipo them together. I have a script to do the latter somewhere. Ronald > > Any suggestions? > > -Chris > > > > > > -- > Christopher Barker, Ph.D. > Oceanographer > > NOAA/OR&R/HAZMAT (206) 526-6959 voice > 7600 Sand Point Way NE (206) 526-6329 fax > Seattle, WA 98115 (206) 526-6317 main reception > > [EMAIL PROTECTED] > _______________________________________________ > Pythonmac-SIG maillist - Pythonmac-SIG@python.org > http://mail.python.org/mailman/listinfo/pythonmac-sig _______________________________________________ Pythonmac-SIG maillist - Pythonmac-SIG@python.org http://mail.python.org/mailman/listinfo/pythonmac-sig
