On 2008-12-23 14:03-0500 Hazen Babcock wrote: > Apologies for leaving out all the details. I've tried to assemble them in the > attached files. > > The problem is: > hl:/usr/local/share/plplot5.9.1/examples$ ./c/x21c > > Plotting Options: > < 1> ps PostScript File (monochrome) > < 2> psc PostScript File (color) > > Enter device number or keyword: 1 > Enter graphics output file name: x21.ps > Segmentation fault > hl:/usr/local/share/plplot5.9.1/examples$
I replicated your cmake.out with the following cmake options. cmake -DCMAKE_INSTALL_PREFIX=/home/software/plplot_cvs/installcmake -DENABLE_DYNDRIVERS=OFF -DDEFAULT_NO_BINDINGS=ON -DDEFAULT_NO_DEVICES=ON -DPLD_ps=ON ../plplot_cmake >& cmake.out Do you confirm those? I am pretty amazed you are getting a segfault for such a simplified build. After my make install I changed to the installed examples tree and did the exact commands you did above with no segfault. I then applied valgrind and got a perfect result. So I am having extraordinary trouble replicating the segfault that you are seeing for virtually an identical platform (both tests were for 64-bit hardware using the same [except for presumably small distro patches] compiler version). Could you try again with the same cmake options as above except for the install prefix? As I told you off list, make sure you have a clean source tree and an initially empty build tree and install tree. Also use VERBOSE=1 for the make commands (e.g., make VERBOSE=1 and make VERBOSE=1 install) to make their output useful and comparable to mine. While in the installed examples/c directory could you also output the result of the pkg-config command that is in the Makefile? For me I have PKG_CONFIG_ENV = \ PKG_CONFIG_PATH=/home/software/plplot_cvs/installcmake/lib/pkgconfig so pkg-config is run as follows: PKG_CONFIG_PATH=/home/software/plplot_cvs/installcmake/lib/pkgconfig \ pkg-config --cflags --libs plplotd The result for my platform is -I/home/software/plplot_cvs/installcmake/include/plplot -L/home/software/plplot_cvs/installcmake/lib -lplplotd -lm -lcsirocsa -lcsironn -lqhull -lfreetype The resulting ldd -r results are as follows: softw...@raven> ldd -r ./x21c linux-vdso.so.1 => (0x00007ffff11ff000) libplplotd.so.9 => /home/software/plplot_cvs/installcmake/lib/libplplotd.so.9 (0x00007f02e8d03000) libm.so.6 => /lib/libm.so.6 (0x00007f02e8a80000) libcsirocsa.so.0 => /home/software/plplot_cvs/installcmake/lib/libcsirocsa.so.0 (0x00007f02e8877000) libcsironn.so.0 => /home/software/plplot_cvs/installcmake/lib/libcsironn.so.0 (0x00007f02e866d000) libqhull.so.5 => /usr/lib/libqhull.so.5 (0x00007f02e8419000) libfreetype.so.6 => /usr/lib/libfreetype.so.6 (0x00007f02e8195000) libc.so.6 => /lib/libc.so.6 (0x00007f02e7e42000) libz.so.1 => /usr/lib/libz.so.1 (0x00007f02e7c2b000) /lib64/ld-linux-x86-64.so.2 (0x00007f02e8f72000) What are your equivalent results? Could others here also try the same simplified build to see if they can replicate the segfault found by Hazen? I am completely stumped as to how to reproduce this. Alan __________________________ Alan W. Irwin Astronomical research affiliation with Department of Physics and Astronomy, University of Victoria (astrowww.phys.uvic.ca). Programming affiliations with the FreeEOS equation-of-state implementation for stellar interiors (freeeos.sf.net); PLplot scientific plotting software package (plplot.org); the libLASi project (unifont.org/lasi); the Loads of Linux Links project (loll.sf.net); and the Linux Brochure Project (lbproject.sf.net). __________________________ Linux-powered Science __________________________ ------------------------------------------------------------------------------ _______________________________________________ Plplot-devel mailing list Plplot-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/plplot-devel