| msg53048 - (view) |
Author: Jeffrey Yasskin (jyasskin) *  |
Date: 2007-08-23 00:02 |
| I tested the change with Apple's gcc and gcc-4.2, but only on an Intel mac. http://gcc.gnu.org/ml/gcc/2005-12/msg00368.html says that -no-cpp-precomp is obsolete, and there weren't any warnings about long doubles, so I just removed those two, and because I'm paranoid, I added a test compile for -mno-fused-madd. |
|
|
| msg53049 - (view) |
Author: Mark Dickinson (mark.dickinson) * |
Date: 2007-08-23 04:15 |
| Also works for me on a PowerPC mac (OS X 10.4.10/G4), with gcc4.2 from macports: compiles successfully and all tests pass. This should fix bug 1450807. What happens with earlier versions of OS X? It looks as though the -Wno-long-double flag was added just to suppress warnings, so nothing dire should happen if it's removed. Might the -no-cpp-precomp still be required for 10.2 or earlier? Would it make sense to remove these two flags only for 10.3 and above, or only when using non Apple-supplied gcc? Or to test for existence of the -no-cpp-precomp flag? |
|
|
| msg55161 - (view) |
Author: Ronald Oussoren (ronaldoussoren) * |
Date: 2007-08-23 17:15 |
| Why does you patch test for -mno-fused-madd? If it isn't actually needed for something I'd rather not have that logic in the configure file, determining which bits of that file are actually needed for supported OS-es is hard enough as it is. -Wno-long-double is needed when compiling on old versions of OSX, where "long double" worked but wasn't actually useful. Keeping that option seems to be harmless. -no-cpp-precomp was necessary for several packages on older releases of OSX. I'll test on a 10.3 box when I have time (which is the oldest I can test on). |
|
|
| msg55209 - (view) |
Author: Jeffrey Yasskin (jyasskin) * |
Date: 2007-08-23 20:52 |
| In http://gcc.gnu.org/ml/gcc/2005-12/msg00368.html, Mike Stump <mrs@apple.com> said "This flag [-no-cpp-precomp] should be removed from the compile, it hasn't been needed in a long while (since gcc-3.1)." which was released in 2002. I'm happy to remove -mno-fused-madd if you say it's unnecessary. Ronald, do you mean "Removing that option [-Wno-long-double] seems to be harmless"? Keeping it harms compiles with non-apple gcc's, but I could add it to a block like -mno-fused-madd. |
|
|
| msg55379 - (view) |
Author: Ronald Oussoren (ronaldoussoren) * |
Date: 2007-08-28 17:57 |
| jyasskin: could you test apple_gcc_flags.diff? That's a version of your patch where configure tests if GCC supports -Wno-long-double and -no- cpp-precomp. NOTE: the patch doesn't update configure, you'll have to run autoconf after applying it. NOTE2: the patch is relative to python's trunk, but should apply to 2.5.x as well. |
|
|
| msg55380 - (view) |
Author: Jeffrey Yasskin (jyasskin) * |
Date: 2007-08-28 18:04 |
| I'm not going to get to this for about a week, but I'll test the patch then. |
|
|
| msg60019 - (view) |
Author: Jeffrey Yasskin (jyasskin) * |
Date: 2008-01-17 07:09 |
| Sorry for taking so long to get to this. gcc 4.2.1 still produces the following errors with this patch: gcc -c -fno-strict-aliasing -mno-fused-madd -no-cpp-precomp -g -Wall -Wstrict-prototypes -I. -IInclude -I./Include -I/opt/local/include -DPy_BUILD_CORE -o Objects/boolobject.o Objects/boolobject.c gcc: unrecognized option '-no-cpp-precomp' cc1: error: unrecognized command line option "-mno-fused-madd" make: *** [Objects/boolobject.o] Error 1 I see you dropped the test for -mno-fused-madd. Since gcc-4.2 doesn't support it, I meant I didn't mind if you dropped the flag entirely. It looks like -no-cpp-precomp stuck around because it doesn't actually cause the compile to fail, just to produce a warning message. |
|
|
| msg63646 - (view) |
Author: Jeffrey Yasskin (jyasskin) * |
Date: 2008-03-17 14:43 |
| I've fixed this in r61436 with a bunch of back pointers to the previous issues. If anyone on old versions sees problems, we can add the flags back conditionally. |
|
|