As I said elsewhere, I would patch Nim's allocator to support shared memory but 
you might as well store your important data in a custom seq that uses 
persistent memory. It should be straight-forward and then you have a 
"reloading" feature simply by running a new binary against the persistent data.

Reply via email to