On Mon, 2017-02-06 at 08:50 +0100, Andreas Schneider wrote: > I have now seen that the version with debug informations does not > crash. > So it might have something to do with the optimisation?
Well optimization is done independently of creating and storing debug information. To test without optimization I edit the file src/Makefile at the line CFLAGS = -g -O2 -fdiagnostics-color=auto -I/usr/include/glib-2.0 ... to read CFLAGS = -g -O0 -fdiagnostics-color=auto -I/usr/include/glib-2.0 ... (It's quite extraordinary, but I've never noticed any performance change when running like that! But debugging is much easier). It'll take me a little while to work out what to do next (we *may* be chasing some bug that is not actually in Denemo, but the fact that it happens while exercising code that I have recently altered is highly suspicious of some un-initialized pointer). Bear with me. Meanwhile it might be worthwhile to generate another make clean && make 2>&1 | tee make.log to be sure that there are no warnings being ignored. Richard _______________________________________________ Denemo-devel mailing list Denemo-devel@gnu.org https://lists.gnu.org/mailman/listinfo/denemo-devel