On 24 May 2011 20:39, Dave McGary <[email protected]> wrote: > Hi Michael, Thanks for your help, and for your quick reply. > > I agree, the -arch ppc is a big part of the problem, maybe all of it. If
Well, the problem seems to me to be that setuptools specifies "-arch ppc" when lipo doesn't support the PPC arch on your system. I am not an expert in those things, though, so maybe I'm misinterpreting something. > in fact ARM files are being generated, then the problem is worse than I > imagined. I think that issue was caused by "crossing the beams", using > one compiler, mixed with other tools and sources. I think one tool > generated PPC code, and another could not deal. The iPhone has an ARM CPU, so if you use an iPhone compiler as you did when creating the symlink to the iPhone version of gcc-4.0 (or crossing the beams as you put it), it would likely have generated ARM binaries. > Unfortunately all that does not help much when trying to understand the > root cause of the build issues. > I have been unable to find the letters ppc anywhere in the cx_freeze code, > so the root problem probably is not there. As I said, I believe it's setuptools and not cx_Freeze that is putting the "-arch ppc" in there. OK, I've just asked my good friend google for some help with this, and it seems there's an issue about this in the setuptools issue tracking system, but that says the problem is with distutils and not setuptools: http://bugs.python.org/setuptools/issue122 and after refining the search a bit, I found the corresponding distutils issue: http://bugs.python.org/issue11623 This has a possible workaround as the last comment: ARCHFLAGS='-arch i386 -arch x86_64' /usr/bin/python2.6 setup.py install See also: http://stackoverflow.com/questions/5366882/installing-pil-on-os-x-snow-leopard-w-xcode4-no-ppc-support > My best guess at this point is that the issues are all in the Python build > tools for OSX. They should not be building FAT bits for OSX 10.6. > > OK, I think I got it figured out. It sucks, but it is simple. > > I am using python 2.6.6 (job requirement). Therefore I am using the > python.org Mac OSX 10.3 binary install. Hence the FAT output. This > SHOULD NOT happen when building on OSX 10.6, which is an Intel only OS > (there is no FAT, there is no PPC, in this universe). > > All of the OSX binaries of python.org builds (including python 3.x) are > 10.3 compatible, which means they may not be 10.6 compatible. ARGH! > > It is possible that at some point between 2.6.6 and 3.2, the python.org > folks figured all this out, and everything works if I can just install the > right binaries. > > I'm not sure about the reaction to straying from 2.6.6. If it solves all > issues, and does not generate others, all will be good... > > I may have to build python from sources to use 2.6.6. Not looking forward > to that. Try the workaround listed above first. > I am moving this to python.org for now... > > I will try to reply to the other posting of this issue which I saw > yesterday here on cx_freeze also. > > Thank you! No problem. Hope this message helps. -- Michael Wood <[email protected]> ------------------------------------------------------------------------------ vRanger cuts backup time in half-while increasing security. With the market-leading solution for virtual backup and recovery, you get blazing-fast, flexible, and affordable data protection. Download your free trial now. http://p.sf.net/sfu/quest-d2dcopy1 _______________________________________________ cx-freeze-users mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/cx-freeze-users
