Thanks. I could only guess here, so I'll leave addition of llvmgcc.mk for someone else. (just please don't add a dash in the HB_COMPILER name).
Brgds, Viktor On 2009 Nov 22, at 14:16, Tamas TEVESZ wrote: > On Sun, 22 Nov 2009, Viktor Szakáts wrote: > >>> could you also add this (llvm-gcc) target to the mix? for all intents >>> and purposes, it is the very same as the gcc target, except compilers >>> being llvm-gcc and llvm-g++ instead of gcc and g++, respectively. >> >> Can you try if it works by setting HB_CCPREFIX=llvm- >> and using gcc target? If this seems ok, I may add it, > > no, it doesn't: > > HB_BUILD_DEBUG=yes > HB_CONTRIBLIBS=no > HB_COMMERCE=yes > HB_INSTALL_PREFIX=/home/ice/w/xhb/hbci/inst/linux/gcc/c > HB_COMPILER=gcc > HB_BUILD_MODE=c > HB_BUILD_OPTIM=no > HB_PLATFORM=linux > HB_CCPREFIX=llvm- > === > ! Building Harbour 2.0.0beta3 from source - http://www.harbour-project.org > ! MAKE: make 3.81 /bin/sh > ! HB_INSTALL_PREFIX: /home/ice/w/xhb/hbci/inst/linux/gcc/c > ! HB_BUILD_DEBUG: yes > ! HB_BUILD_OPTIM: no > ! HB_BUILD_MODE: c > ! HB_CONTRIBLIBS: no > ! HB_HOST_PLAT: linux (x86_64) HB_SHELL: sh > ! HB_PLATFORM: linux (x86_64) > ! HB_COMPILER: gcc > ! Component: 'zlib' found in /usr/include > ! Component: 'pcre' found in /usr/include > ! Component: 'openssl' found in /usr/include > ! Component: 'gpm' explicitly disabled > ! Component: 'slang' explicitly disabled > ! Component: 'curses' found in /usr/include > ! Component: 'x11' found in /usr/include > ! Component: 'wattcp/watt-32' not supported on linux platform > ! REVISION: 12974 > make[2]: Nothing to be done for `all'. > make[1]: Nothing to be done for `all'. > llvm-gcc -I. -I../../../../../include -g -DHB_TR_LEVEL_DEBUG > -I/usr/include -I/usr/include -o_hbhbpdf.o -c ../../../_hbhbpdf.c > [...] > llvm-gcc -I. -I../../../../../include -Wall -W -g -DHB_TR_LEVEL_DEBUG > -ostrwild_dyn.o -DHB_DYNLIB -fpic -c ../../../strwild.c > ( llvm-ar rcs ../../../../../lib/linux/gcc/libhbcommon.a expropt1.o > expropt2.o hbffind.o hbfopen.o hbfsapi.o hbgete.o hbwince.o hbhash.o > hbprintf.o hbstr.o hbdate.o hbmem.o hbtrace.o hbver.o hbverdsp.o hbarch.o > reserved.o strwild.o ) || ( rm -f ../../../../../lib/linux/gcc/libhbcommon.a > && false ) > llvm-gcc -I. -I../../../../../include -Wall -W -g -DHB_TR_LEVEL_DEBUG > -onortl.o -c ../../../nortl.c > ( llvm-ar rcs ../../../../../lib/linux/gcc/libhbnortl.a nortl.o ) || ( rm > -f ../../../../../lib/linux/gcc/libhbnortl.a && false ) > llvm-gcc -I. -I../../../../../include -Wall -W -g -DHB_TR_LEVEL_DEBUG > -ohbpp.o -c ../../../hbpp.c > llvm-gcc -I. -I../../../../../include -Wall -W -g -DHB_TR_LEVEL_DEBUG > -ohbpp_dyn.o -DHB_DYNLIB -fpic -c ../../../hbpp.c > llvm-gcc -Wall -W -g -L../../../../../lib/linux/gcc > -o../../../../../bin/linux/gcc/hbpp hbpp.o -lhbnortl -lhbcommon -lm > ../../../../../lib/linux/gcc/libhbnortl.a: could not read symbols: Archive > has no index; run ranlib to add one > collect2: ld returned 1 exit status > make[3]: *** [hbpp] Error 1 > rm hbpp.o > make[2]: *** [descend] Error 2 > make[1]: *** [pp] Error 2 > make: *** [src] Error 2 > > but using this: > > --- gcc.mk 2009-11-22 14:00:24.000000000 +0100 > +++ llvm-gcc.mk 2009-11-22 14:03:30.000000000 +0100 > @@ -3,9 +3,9 @@ > # > > ifeq ($(HB_BUILD_MODE),cpp) > - HB_CMP := g++ > + HB_CMP := llvm-g++ > else > - HB_CMP := gcc > + HB_CMP := llvm-gcc > endif > > OBJ_EXT := .o > > it does (HB_COMPILER=llvm-gcc). i have no idea what the essential > difference is, but this way it does, HB_CCPREFIX way it doesn't. i've > also modified postinst.sh like > > Index: postinst.sh > =================================================================== > --- postinst.sh (revision 12974) > +++ postinst.sh (working copy) > @@ -49,7 +49,8 @@ > [ "$HB_COMPILER" = "cygwin" ] || \ > [ "$HB_COMPILER" = "djgpp" ] || \ > [ "$HB_COMPILER" = "icc" ] || \ > - [ "$HB_COMPILER" = "sunpro" ] > + [ "$HB_COMPILER" = "sunpro" ] || \ > + [ "$HB_COMPILER" = "llvm-gcc" ] > then > if [ -n "${HB_TOOLS_PREF}" ]; then > > hb_mkdyn="${HB_INST_PKGPREF}${HB_BIN_INSTALL}/${HB_TOOLS_PREF}-mkdyn" > > this, and the installed tree seems mostly ok (what not ok is the > dynlib link generation, which is probably due to hb-mkdyn.sh also in > need of some massaging). > > full build logs of both cases are available if needed (my build tests > only do hb core, ie. no contribs or anything). > > -- > [-] > > mkdir /nonexistent > _______________________________________________ > Harbour mailing list (attachment size limit: 40KB) > Harbour@harbour-project.org > http://lists.harbour-project.org/mailman/listinfo/harbour _______________________________________________ Harbour mailing list (attachment size limit: 40KB) Harbour@harbour-project.org http://lists.harbour-project.org/mailman/listinfo/harbour
