Hey,
> Could it be that the bug was fixed in the meantime? Or that this one > was, say, built directly via Guix whereas the other one was built > through Cuirass? Mystery! That's strange. There's nothing really special about how Cuirass builds its stuff. It's a plain "build-derivations" call in the "cuirass remote-worker" process. This process only builds stuff and report it using (simple-zmq). A memory corruption in that module also seems unlikely, as there are almost 30 instances of this process running nicely for days. Thanks, Mathieu