GitLab

In command_process_destroy_cell(), we are supposed to convert client-ward destroy cells into truncated cells, so the client can know at what hop in the path the circuit broke. But we don't actually do this currently, because the client behave was always to close the circuit, so we were wasting bandwidth with a two-round-trip circuit close.

If we're going to start trying to re-extend circuits, though, we need this functionality. Should we re-enable the ifdef 0'ed stuff (and remove the circuit_mark_for_close())?

[Automatically added by flyspray2trac: Operating System: All]