Re: [Cooker] SDL12 build fix with gcc 3.3
Stefan van der Eijk [EMAIL PROTECTED] writes: [...] Is this the same thing I'm seeing with Mesa (hasn't been rebuilding since gcc 3.3): http://eijk.homelinux.org/build/cooker/i586/problem/Mesa-5.0.1-3mdk gcc -O2 -fomit-frame-pointer -pipe -march=i586 -mcpu=pentiumpro -g -march=pentium -DUSE_X86_ASM -DUSE_MMX_ASM -DUSE_3DNOW_ASM -Wall -Wno-uninitialized -O3 -fexpensive-optimizations -fomit-frame-pointer -finline-functions -funroll-loops -DNDEBUG -DHW_LOG_ENABLED -DGLX_LOG -DDYNAMIC_MODULE -DSTDC_HEADERS=1 -DHAVE_FCNTL_H=1 -DHAVE_LIMITS_H=1 -DHAVE_SYS_IOCTL_H=1 -DHAVE_UNISTD_H=1 -DHAVE_SYS_TYPES_H=1 -DHAVE_SYS_MMAN_H=1 -DHAVE_SYS_PERM_H=1 -DHAVE_ASM_SIGCONTEXT_H=1 -DHAVE_LINUX_NEWAGP=1 -DSIZEOF_LONG=4 -DRETSIGTYPE=void -DHAVE_VPRINTF=1 -DHAVE_STRDUP=1 -DHAVE_STRERROR=1 -DHAVE_STRSTR=1 -DHAVE_IOPL=1 -DHAVE_LIBDL=1-DGLX_LOG -DGLXEXT -DXDEBUG -DGLX_SO_PATH='/usr/X11R6/lib/modules/glx-3.so' -DMESA31 -DBUILD_COLOR_TABLE_EXT=1 -DGLX_MODULE -I/home/cooker/RPM/BUILD/Mesa-5.0.1/glx/Mesa-3.2.1/src -I/home/cooker/RPM/BUILD/Mesa-5.0.1/glx/Mesa-3.2.1/include -I/home/cooker/RPM/BUILD/Mesa-5.0.1/glx/Mesa-3.2.1/src/X -I/home/cooker/RPM/BUILD/Mesa-5.0.1/glx/Mesa-3.2.1/src/X86 -fPIC - c 3dnow.c 3dnow.c:159:35: pasting gl_3dnow_rescale_normals_raw and ( does not give a valid preprocessing token 3dnow.c:159:35: pasting gl_3dnow_normalize_normals_raw and ( does not give a valid preprocessing token [...] BitchX was not rebuilding because of a similar problem. Please check out: http://gcc.gnu.org/onlinedocs/gcc-3.3/cpp/Concatenation.html#Concatenation Quoting: two tokens that don't together form a valid token cannot be pasted together. For example, you cannot concatenate x with + in either order. If you try, the preprocessor issues a warning and emits the two tokens. Whether it puts white space between the tokens is undefined. It is common to find unnecessary uses of ## in complex macros. If you get this warning, it is likely that you can simply remove the ##. I'm rebuilding Mesa and will fix if this is really the same problem, which is highly probable. PS : please could you remove quoted text that is unneeded -- Guillaume Cottenceau - http://people.mandrakesoft.com/~gc/
RE: [Cooker] SDL12 build fix with gcc 3.3
-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 Is any of this also the cause of goom/synaethesia not compiling as detailed in message [Cooker] gcc issue with compiling goom in MythTV/mythmusic? -BEGIN PGP SIGNATURE- Version: PGPfreeware 7.0.3 for non-commercial use http://www.pgp.com iQA/AwUBPuiOhvc1NpCTlP0JEQLVgACfZp7to91JFfZuCxbfd2u0Po/LTVwAnjFu JwX5oA5SkCeg2yaw9tlWGnBf =Yx3B -END PGP SIGNATURE-
Re: [Cooker] SDL12 build fix with gcc 3.3
Hi! On Wed 2003-06-11 at 09:43:18 +0200, [EMAIL PROTECTED] wrote: This is at least needed on powerpc to build SDL12 with gcc 3.3. AFAIK, this is need for all platforms. gcc 3.3 has abandoned support for multi-line strings, after they were already deprecated since 3.0: http://gcc.gnu.org/gcc-3.3/changes.html --- SDL-1.2.5/src/video/SDL_stretch.c.bak 2003-05-09 20:04:49.0 +0200 +++ SDL-1.2.5/src/video/SDL_stretch.c 2003-05-09 19:57:47.0 +0200 @@ -261,9 +261,9 @@ break; default: #ifdef __GNUC__ - __asm__ __volatile__ ( - call _copy_row - + __asm__ __volatile__ ( + call _copy_row + : =D (u1), =S (u2) You either have to concatenate them as shown here or end each line with '\' or '\n\', depending, if you need the new-line or not. Bye, Benjamin.
Re: [Cooker] SDL12 build fix with gcc 3.3
Benjamin Pflugmann wrote: Hi! On Wed 2003-06-11 at 09:43:18 +0200, [EMAIL PROTECTED] wrote: This is at least needed on powerpc to build SDL12 with gcc 3.3. AFAIK, this is need for all platforms. gcc 3.3 has abandoned support for multi-line strings, after they were already deprecated since 3.0: http://gcc.gnu.org/gcc-3.3/changes.html --- SDL-1.2.5/src/video/SDL_stretch.c.bak 2003-05-09 20:04:49.0 +0200 +++ SDL-1.2.5/src/video/SDL_stretch.c 2003-05-09 19:57:47.0 +0200 @@ -261,9 +261,9 @@ break; default: #ifdef __GNUC__ - __asm__ __volatile__ ( - call _copy_row - + __asm__ __volatile__ ( + call _copy_row + : =D (u1), =S (u2) You either have to concatenate them as shown here or end each line with '\' or '\n\', depending, if you need the new-line or not. Bye, Benjamin. Is this the same thing I'm seeing with Mesa (hasn't been rebuilding since gcc 3.3): http://eijk.homelinux.org/build/cooker/i586/problem/Mesa-5.0.1-3mdk gcc -O2 -fomit-frame-pointer -pipe -march=i586 -mcpu=pentiumpro -g -march=pentium -DUSE_X86_ASM -DUSE_MMX_ASM -DUSE_3DNOW_ASM -Wall -Wno-uninitialized -O3 -fexpensive-optimizations -fomit-frame-pointer -finline-functions -funroll-loops -DNDEBUG -DHW_LOG_ENABLED -DGLX_LOG -DDYNAMIC_MODULE -DSTDC_HEADERS=1 -DHAVE_FCNTL_H=1 -DHAVE_LIMITS_H=1 -DHAVE_SYS_IOCTL_H=1 -DHAVE_UNISTD_H=1 -DHAVE_SYS_TYPES_H=1 -DHAVE_SYS_MMAN_H=1 -DHAVE_SYS_PERM_H=1 -DHAVE_ASM_SIGCONTEXT_H=1 -DHAVE_LINUX_NEWAGP=1 -DSIZEOF_LONG=4 -DRETSIGTYPE=void -DHAVE_VPRINTF=1 -DHAVE_STRDUP=1 -DHAVE_STRERROR=1 -DHAVE_STRSTR=1 -DHAVE_IOPL=1 -DHAVE_LIBDL=1 -DGLX_LOG -DGLXEXT -DXDEBUG -DGLX_SO_PATH='/usr/X11R6/lib/modules/glx-3.so' -DMESA31 -DBUILD_COLOR_TABLE_EXT=1 -DGLX_MODULE -I/home/cooker/RPM/BUILD/Mesa-5.0.1/glx/Mesa-3.2.1/src -I/home/cooker/RPM/BUILD/Mesa-5.0.1/glx/Mesa-3.2.1/include -I/home/cooker/RPM/BUILD/Mesa-5.0.1/glx/Mesa-3.2.1/src/X -I/home/cooker/RPM/BUILD/Mesa-5.0.1/glx/Mesa-3.2.1/src/X86 -fPIC -c 3dnow.c 3dnow.c:159:35: pasting gl_3dnow_rescale_normals_raw and ( does not give a valid preprocessing token 3dnow.c:159:35: pasting gl_3dnow_normalize_normals_raw and ( does not give a valid preprocessing token 3dnow.c:159:35: pasting gl_3dnow_transform_normals_raw and ( does not give a valid preprocessing token 3dnow.c:159:35: pasting gl_3dnow_transform_normals_no_rot_raw and ( does not give a valid preprocessing token 3dnow.c:159:35: pasting gl_3dnow_transform_rescale_normals_raw and ( does not give a valid preprocessing token 3dnow.c:159:35: pasting gl_3dnow_transform_rescale_normals_no_rot_raw and ( does not give a valid preprocessing token 3dnow.c:159:35: pasting gl_3dnow_transform_normalize_normals_raw and ( does not give a valid preprocessing token 3dnow.c:159:35: pasting gl_3dnow_transform_normalize_normals_no_rot_raw and ( does not give a valid preprocessing token 3dnow.c:173:37: pasting gl_3dnow_rescale_normals_raw and ; does not give a valid preprocessing token 3dnow.c:173:37: pasting gl_3dnow_normalize_normals_raw and ; does not give a valid preprocessing token 3dnow.c:173:37: pasting gl_3dnow_transform_normals_raw and ; does not give a valid preprocessing token 3dnow.c:173:37: pasting gl_3dnow_transform_normals_no_rot_raw and ; does not give a valid preprocessing token 3dnow.c:173:37: pasting gl_3dnow_transform_rescale_normals_raw and ; does not give a valid preprocessing token 3dnow.c:173:37: pasting gl_3dnow_transform_rescale_normals_no_rot_raw and ; does not give a valid preprocessing token 3dnow.c:173:37: pasting gl_3dnow_transform_normalize_normals_raw and ; does not give a valid preprocessing token 3dnow.c:173:37: pasting gl_3dnow_transform_normalize_normals_no_rot_raw and ; does not give a valid preprocessing token make[2]: *** [3dnow.o] Error 1 make[2]: Leaving directory `/home/cooker/RPM/BUILD/Mesa-5.0.1/glx/servGL/mesaglx/X86' make[1]: *** [subdirs] Error 1 make[1]: Leaving directory `/home/cooker/RPM/BUILD/Mesa-5.0.1/glx/servGL' make: *** [all] Error 1 error: Bad exit status from /home/cooker/tmp/rpm-tmp.87821 (%build) smime.p7s Description: S/MIME Cryptographic Signature
Re: [Cooker] SDL12 build fix with gcc 3.3
On Thu 2003-06-12 at 00:16:54 +0200, [EMAIL PROTECTED] wrote: Benjamin Pflugmann wrote: AFAIK, this is need for all platforms. gcc 3.3 has abandoned support for multi-line strings, after they were already deprecated since 3.0: http://gcc.gnu.org/gcc-3.3/changes.html Is this the same thing I'm seeing with Mesa (hasn't been rebuilding since gcc 3.3): I doubt it. From what I have seen, gcc 3.3 gives very specific errors about the multi-line strings (i.e. that multi-line mentioned in the error message itself). http://eijk.homelinux.org/build/cooker/i586/problem/Mesa-5.0.1-3mdk gcc -O2 -fomit-frame-pointer -pipe -march=i586 -mcpu=pentiumpro -g -march=pentium -DUSE_X86_ASM -DUSE_MMX_ASM -DUSE_3DNOW_ASM -Wall -Wno-uninitialized -O3 -fexpensive-optimizations -fomit-frame-pointer -finline-functions -funroll-loops -DNDEBUG -DHW_LOG_ENABLED -DGLX_LOG -DDYNAMIC_MODULE -DSTDC_HEADERS=1 -DHAVE_FCNTL_H=1 -DHAVE_LIMITS_H=1 -DHAVE_SYS_IOCTL_H=1 -DHAVE_UNISTD_H=1 -DHAVE_SYS_TYPES_H=1 -DHAVE_SYS_MMAN_H=1 -DHAVE_SYS_PERM_H=1 -DHAVE_ASM_SIGCONTEXT_H=1 -DHAVE_LINUX_NEWAGP=1 -DSIZEOF_LONG=4 -DRETSIGTYPE=void -DHAVE_VPRINTF=1 -DHAVE_STRDUP=1 -DHAVE_STRERROR=1 -DHAVE_STRSTR=1 -DHAVE_IOPL=1 -DHAVE_LIBDL=1-DGLX_LOG -DGLXEXT -DXDEBUG -DGLX_SO_PATH='/usr/X11R6/lib/modules/glx-3.so' -DMESA31 -DBUILD_COLOR_TABLE_EXT=1 -DGLX_MODULE -I/home/cooker/RPM/BUILD/Mesa-5.0.1/glx/Mesa-3.2.1/src -I/home/cooker/RPM/BUILD/Mesa-5.0.1/glx/Mesa-3.2.1/include -I/home/cooker/RPM/BUILD/Mesa-5.0.1/glx/Mesa-3.2.1/src/X -I/home/cooker/RPM/BUILD/Mesa-5.0.1/glx/Mesa-3.2.1/src/X86 -fPIC -c 3dnow.c 3dnow.c:159:35: pasting gl_3dnow_rescale_normals_raw and ( does not give a valid preprocessing token 3dnow.c:159:35: pasting gl_3dnow_normalize_normals_raw and ( does not give a valid preprocessing token After quick look, the problems seems to be here (3dnow.c:26ff): extern void gl_##pfx##_rescale_normals_##masked## (NORM_ARGS);\ extern void gl_##pfx##_normalize_normals_##masked## (NORM_ARGS); \ ... IMHO, the last concatination (##) makes no sense. The concatination is used to create a new label, but the label ended with 'masked'. If I am not mistaken, simply removing the last '##' on every of these line should do the trick. But then, beware, my preprocessor-fu is not worth of mentioning (I never have used this concatination myself). Bye,
