Dear Maintainer, this SIGSEGV and also the valgrind "invalid read" below ledger::journal_t::clear_xdata can also be seen in a Debian 11 Bullseye VM.
Because valgrind points already to a use-after-free, and the output "Done." is printed, might the issue here be, that python does destruct the "session" variable before the "journal" while it is still kind of connected at C++ side? And therefore maybe an upstream issue? Kind regards, Bernhard
