Menu

#1451 build of bf_test.exe fails for 4.6-branch on windows

closed-fixed
nobody
None
5
2015-03-23
2014-07-31
No

Today I check out the recent stable branch source (Latest ChangeLog date 2014-07-30) and tried to build binaries on windows using mingw 32 and 64 bit. The binaries of gnuplot.exe, wgnuplot.exe and wgnuplot_pipes.exe were built flawlessly. However, "make" stopped at building bf_test.exe.

gcc -L/c/Programs/gplibs32/lib -Wl,--enable-auto-import -Wl,--enable-runtime-pseudo-reloc-v2 -Wl,--allow-multiple-definition -s -I/c/Programs/gplibs32/include -D_FILE_OFFSET_BITS=64 -D__USE_MINGW_ANSI_STDIO -O2 -pipe -I. -I../../src -D_Windows -DHAVE_CONFIG_H -DGNUPLOT_SHARE_DIR=\"share\" -DDEVELOPMENT_VERSION -DUSE_MOUSE=1 -DWIN_IPC -DWITH_HTML_HELP -I/c/PROGRA~2/HELPWO~1/include -DHAVE_LIBGD -DHAVE_LIBPNG -DHAVE_LIBGD -DHAVE_GD_H -DHAVE_GD_GIF -DGIF_ANIMATION -DHAVE_GD_PNG -DHAVE_GD_JPEG -DHAVE_GD_TTF -DHAVE_CAIROPDF -DWXWIDGETS -DHAVE_LUA -DHAVE_ICONV -U_Windows -o ../../demo/bf_test.exe ../../src/bf_test.c dbinary.o alloc.o
C:\Users\Tatsu\AppData\Local\Temp\cc40n8gJ.o:bf_test.c:(.text+0x35): undefined reference to MyFPrintF' C:\Users\Tatsu\AppData\Local\Temp\cc40n8gJ.o:bf_test.c:(.text+0x12f): undefined reference toMyFPrintF'
c:/mingw/bin/../lib/gcc/i686-w64-mingw32/4.8.2/../../../../i686-w64-mingw32/bin/ld.exe: C:\Users\Tatsu\AppData\Local\Temp\cc40n8gJ.o: bad reloc address 0x10 in section `.text.startup'
collect2.exe: error: ld returned 1 exit status
make: *** [../../demo/bf_test.exe] Error 1

(The errors are the same between 32 and 64 bit build.)

I have confirmed that the above errors do not occur using the source Latest ChangeLog date of which is 2014-07-15.
(Unfortunately I could not get the source tree at 2014-07-28.)

The change on 2014-07-30 might affects the build.
I execute diff to src/eval.c and sys/syscfg.h which are changed between 07-15 and 07-30.
However, I could find the origin.

(As a reference I attached a diff file.)

Interestingly, the error does not occur 5.0 cvs branch although the similar changes are done.

Tatsuro

1 Attachments

Discussion

  • Tatsuro MATSUOKA

    I got the source at 2014-07-15 and replace the src/eval.c or sys/syscfg.h in the latest source.
    When I replaced src/eval.c, no error was occurred. While when I replaced sys/syscfg.h, the same error was occurred. The origin of the error is the change of sys/syscfg.h.

     
  • Tatsuro MATSUOKA

    The attachment file is working diff from syscfg.h at 2014-07-15. This modification works for both 32 and 64 bits. On 64 bit build, build OS are correctly represented.

    The first part of modification on 2014-07-28 seems to be origin the build of bf_test.exe error. I cannot find the reason why modification on 2014-07-28 affects build of bf_test.exe.

     
  • Ethan Merritt

    Ethan Merritt - 2014-07-31

    I have reverted the 2014-07-28 change to syscfg.h except for the line

    define OS ...

    The file bf_test.c should be removed or rewritten, since it is now
    causing problems on both Windows and linux.

     
  • Tatsuro MATSUOKA

    I have confirmed that build on MinGW platfrom works both 32 and 64 bit. Thanks!

     
  • Ethan Merritt

    Ethan Merritt - 2014-08-15
    • status: open --> closed-fixed
     
  • Ethan Merritt

    Ethan Merritt - 2014-08-15

    bf_test rewritten for version 5

     

Log in to post a comment.