This is an old revision of the document!
Afaik usually there is no need to attach coredumps to bugs. For most bugs it is sufficient to get the backtrace. For most distros it is possible to provide a meaningfull backtrace: https://docs.xfce.org/contribute/bugs/start#backtraces
Afaik there are (at least) three different ways how to diagnose the cause of a crash
1.) Provide a failsafe way to reproduce the crash
If devs can reproduce the crash on their own, they can debug and hopefully fix it. However usually it is very hard to figure out why the crash happens for you, but not for others.
2.) Provide a meaningfull backtrace by using debug packages
Some distributions ship debug packages ( e.g <package>-dbg or <package>-dbgsym ). If your distribution does so, you can create a coredump and add debug-symbols to it. ( Usually the raw coredump alone is not of much value )
- - set “ulimit -c unlimited” in a console to get coredumps for binaries started from this console
- - set “ulimit -c unlimited” in .profile to get coredumps from everything
After that, you can use “gdb <programm> <coredump>” + “add-symbol-file ~/path/to/symbols.sym” + “bt” to get a meaningfull backtrace. E.g. here some manual for Ubuntu: https://wiki.ubuntu.com/DebuggingProgramCrash
3.) Compile the appliaction with debug symbols, install it and provide a backtrace
If all the other options fail, you can still compile stuff from source with debug symbols enabled (e.g. “./autogen.sh –enable-debug” if the package uses autotools ) You will need to install the used build environment and all dependencies of the application in dev-version .. the build-tool usually tells what is missing. Reproduce the crash with the self-compiled binary, either via gdb or via coredump, so you get a backtrace which has file-names/line-numbers.