Hi Interesting thread :-/
First up, ICA is not the be all and end all. It's just another tool in the armory in ensuring a good build. And it's not perfect either.. It's amazing how such a simple concept can apparently be troublesome for some folks to grasp. I was hoping my scripts would have been enough. Plus it's all been said before (in the archives). Dan N. seems to have the best handle on it but apparently he's lacking in CPU cycles. The Alphabetical proponents should tee him up access to some decent hardware :-) I hadn't run any ICA builds for a while but this thread prompted me to do so. Here are the results of a current DIY build (gcc-4.0.2, glibc-2.3.6) http://www.diy-linux.org/~greg/ICA/ AFAICT, the displayed differences are all explainable and therefore acceptable. As mentioned in the archives (and by Ken) the libstdc++ differences are due to randomness in the libstdc++ build due to symbol mangling and anonymous namespaces: http://linuxfromscratch.org/pipermail/lfs-dev/2003-August/036124.html http://gcc.gnu.org/ml/gcc/2003-08/msg00472.html This can be seen with something like so: $ cmp -lb cmp/iter1/usr/lib/libstdc++.a/locale-inst.o cmp/iter2/usr/lib/libstdc++.a/locale-inst.o 131079 67 7 106 F 131080 65 5 64 4 131081 101 A 105 E 131082 106 F 103 C 131083 105 E 103 C 131084 70 8 67 7 131085 63 3 105 E 131086 104 D 106 F 217192 67 7 106 F 217193 65 5 64 4 217194 101 A 105 E 217195 106 F 103 C 217196 105 E 103 C 217197 70 8 67 7 217198 63 3 105 E 217199 104 D 106 F $ nm cmp/iter1/usr/lib/libstdc++.a/locale-inst.o > 11 $ nm cmp/iter2/usr/lib/libstdc++.a/locale-inst.o > 22 $ diff -u 11 22 --- 11 2005-12-17 10:20:29.000000000 +1100 +++ 22 2005-12-17 10:20:34.000000000 +1100 @@ -1,7 +1,7 @@ 00000008 r .LC4 00000011 r .LC5 00000017 r .LC6 -00000000 t _GLOBAL__I_.._.._.._.._gcc_4.0.2_libstdc___v3_src_locale_inst.cc_6FCA3617_75AFE83D +00000000 t _GLOBAL__I_.._.._.._.._gcc_4.0.2_libstdc___v3_src_locale_inst.cc_6FCA3617_F4ECC7EF U _Unwind_Resume 00000000 t _Z41__static_initialization_and_destruction_0ii 00000000 V _ZGVNSt10moneypunctIcLb0EE2idE Anyhoo, this is all very much old hat :-) IMHO an interesting test these days would be to binary diff the end results of a system built natively versus a cross built system. Regards Greg -- http://www.diy-linux.org/ -- http://linuxfromscratch.org/mailman/listinfo/lfs-dev FAQ: http://www.linuxfromscratch.org/faq/ Unsubscribe: See the above information page
