We're trying to sort out why transcode isn't compiling on OSX:
From: achurch@achurch.org (Andrew Church) To: transcode-devel@exit1.org Subject: Re: [transcode-devel] Re: Compiling transcode Date: Fri, 27 Apr 2007 08:52:42 JST X-WWMF-Version: 3.2.20070423.2@fries Reply-To: transcode Developers Mailing List <transcode-devel@exit1.org>
gcc -DHAVE_CONFIG_H -I. -I. -I.. -D_REENTRANT -DLINUX -I.. -Wall -D_LARGEFILE_SOURCE -D_LARGEFILE64_SOURCE -D_FILE_OFFSET_BITS=64 -g -O2 -no-cpp-precomp -D_INTL_REDIRECT_MACROS -MT tcmemcpy.lo -MD -MP -MF .deps/tcmemcpy.Tpo -c tcmemcpy.c -fno-common -DPIC -o .libs/ tcmemcpy.o /var/tmp//ccMwDlN6.s:43:missing or invalid immediate expression `0b111' taken as 0
What version of binutils are you using? (Sorry if you already mentioned this--I haven't had a chance to review the thread.) This looks like a case of an outdated assembler version not being able to parse binary constants.
I responded I don't know how to determine what OSX is using natively. How is that done? I then installed the port of FSF's binutils, reconfigured transcode and recompiled, but the build failed in exactly the same way. How do I know if the FSF binutils is being used?
/var/tmp//ccMwDlN6.s:206:Alignment too large: 15. assumed.
This should only happen on x86 if you're compiling to a.out, which is ancient, or (again) if your assembler is too old. (On the other hand, the use of .balign rather than .align here would avoid such potential trouble, and I'll look into making that change on CVS HEAD when I have a chance.)
Note that you can disable assembler code entirely by passing --disable-mmx to ./configure (but this can slow transcode down considerably depending on what you do with it).
I responded I tried to compile with --disable-mmx, and also --disable-sse and --disable-sse2, individually and collectively, and it had absolutely no effect on the build -- it fails in exactly the same way. Suggestions would be much appreciated. Dave