It feels similar to database migrations: sure, you can change the schema, but if the code is changed accordingly at the same time, the migration will happen smo

It feels similar to database migrations: sure, you can change the schema, but if the code is changed accordingly at the same time, the migration will happen smoothly, without having downtime. The fact that other languages don't feature restartless updates is as strange, from a point of view, as it is to imagine using a database that forces you to have downtime. Of course, we go around that by having multiple servers that restart in sequence… but to think that people were able to debug and update spacecraft software while it was running, without downtime, is pretty impressive (the story is narrated here: http://flownet.com/gat/jpl-lisp.html http://flownet.com/gat/jpl-lisp.html).