On Wed, Mar 14, 2012 at 08:34:34PM +0100, I. Szczesniak wrote: > On Mon, Mar 5, 2012 at 11:36 PM, Wendy Lin <[email protected]> wrote: > > ast-ksh 2012-02-14 locale.sh fails for me on SUSE LINUX12.1 AMD64. > > This happens for gcc and clang builds regardless of optimiser settings > > (debug, nodebug, -flto ...). Can anyone reproduce the test failure: > > > > /home/wenlin/ast_clang3/arch/linux.i386/src/cmd/ksh93/ksh > > ./src/cmd/ksh93/tests/shtests --locale > > LD_LIBRARY_PATH_64=/home/wenlin/ast_clang3/arch/linux.i386/lib: > > LD_LIBRARY_PATH=/home/wenlin/ast_clang3/arch/linux.i386/lib:/usr/lib64/mpi/gcc/openmpi/lib64 > > LD_LIBRARY_PATH_32=/home/wenlin/ast_clang3/arch/linux.i386/lib: > > LC_ALL=C LANG=C VMALLOC_OPTIONS=abort > > SHCOMP='/home/wenlin/ast_clang3/arch/*/bin/shcomp' > > ./src/cmd/ksh93/tests/locale.sh > > test locale begins at 2012-03-05+22:58:52 > > locale.sh[276]: LANG change not seen by function -- expected > > $'(libshell,3,46)\nAn error occurred.\n(libshell,3,46)' or > > $'(debug,message,libshell,An error occurred.)\nAn error > > occurred.\n(debug,message,libshell,An error occurred.)', got > > $'(libshell,1,46)\nAn error occurred.\n(libshell,1,46)' > > test locale failed at 2012-03-05+22:58:53 with exit code 1 [ 31 tests 1 > > error ] > > test locale(shcomp) begins at 2012-03-05+22:58:53 > > shcomp-locale.ksh[276]: LANG change not seen by function -- > > expected $'(libshell,3,46)\nAn error occurred.\n(libshell,3,46)' or > > $'(debug,message,libshell,An error occurred.)\nAn error > > occurred.\n(debug,message,libshell,An error occurred.)', got > > $'(libshell,1,46)\nAn error occurred.\n(libshell,1,46)' > > test locale(shcomp) failed at 2012-03-05+22:58:54 with exit code 1 [ > > 31 tests 1 error ] > > I can confirm the failure of locale.sh on Suse 12.1 but not on the > Suse 11.x series. Might be a problem with glibc. Werner, could you > investigate, please? This cripples builtin localisation of ksh scripts > on Suse 12.1.
Don't know what's happen, but here it works (OK I've changed in locale.sh the locale from C_EU.UTF-8 to an existing de_DE.UTF-8 locale and also changed the option for wc from -C to -m as the wc from the coreutils does not know about -C): ~/rpmbuild/BUILD/ksh93> LD_LIBRARY_PATH=$PWD/arch/linux.i386-64/lib \ PATH=$PWD/arch/linux.i386-64/bin:$PATH SHELL=$PWD/arch/linux.i386-64/bin/ksh \ $PWD/arch/linux.i386-64/bin/ksh src/cmd/ksh93/tests/shtests --locale src/cmd/ksh93/tests/locale.sh test locale(POSIX) begins at 2012-03-15+08:15:11 test locale(POSIX) passed at 2012-03-15+08:15:12 [ 31 tests 0 errors ] test locale(shcomp) begins at 2012-03-15+08:15:12 test locale(shcomp) passed at 2012-03-15+08:15:12 [ 31 tests 0 errors ] even with some other locale this works flawless: ~/rpmbuild/BUILD/ksh93> LANG=de_DE.UTF-8 LD_LIBRARY_PATH=$PWD/arch/linux.i386-64/lib \ PATH=$PWD/arch/linux.i386-64/bin:$PATH SHELL=$PWD/arch/linux.i386-64/bin/ksh \ $PWD/arch/linux.i386-64/bin/ksh src/cmd/ksh93/tests/shtests --locale src/cmd/ksh93/tests/locale.sh test locale(de_DE.UTF-8) begins at 2012-03-15+08:21:22 test locale(de_DE.UTF-8) passed at 2012-03-15+08:21:22 [ 31 tests 0 errors ] test locale(shcomp) begins at 2012-03-15+08:21:22 test locale(shcomp) passed at 2012-03-15+08:21:22 [ 31 tests 0 errors ] ~/rpmbuild/BUILD/ksh93> LANG=en_US.UTF-8 LD_LIBRARY_PATH=$PWD/arch/linux.i386-64/lib \ PATH=$PWD/arch/linux.i386-64/bin:$PATH SHELL=$PWD/arch/linux.i386-64/bin/ksh \ $PWD/arch/linux.i386-64/bin/ksh src/cmd/ksh93/tests/shtests --locale src/cmd/ksh93/tests/locale.sh test locale(en_US.UTF-8) begins at 2012-03-15+08:21:31 test locale(en_US.UTF-8) passed at 2012-03-15+08:21:32 [ 31 tests 0 errors ] test locale(shcomp) begins at 2012-03-15+08:21:32 test locale(shcomp) passed at 2012-03-15+08:21:32 [ 31 tests 0 errors ] all done with ksh93u+ 2012-02-29 on openSUSE factory within a clean build chroot environment with an own build user (not root). At https://build.opensuse.org/package/show?package=ksh&project=shells you'll find ksh93u+ 2012-02-14 even for openSUSE 12.1. I'm currently working on ksh93u+ 2012-02-29 but found the the plain upstream version also shows a small memory leak but it seems that my patch I've also reported upstream a few weeks ago does fix this one as well. Werner -- "Having a smoking section in a restaurant is like having a peeing section in a swimming pool." -- Edward Burr _______________________________________________ ast-users mailing list [email protected] https://mailman.research.att.com/mailman/listinfo/ast-users
