[Joker_vD]

The really, really fundamental issue this proposal doesn't touch at all is that you also need to

- either version the data structures/classes/shapes of dictionaries/whatever that a function accepts/returns;

- or have converters between different data versions and use them inside your functions.

As I said in another topic on HN which was about that project that hoped to bring hot-code reloading in a C REPL or something: changing the code inside the running program is the least of the problems; flawlessly updating the data inside the running problem so that the new code could proceed to work on it — that's the hard problem (think e.g. about rolling back the update that threw away the bunch of fields).

[seanhunter]

Absolutely this. The edge cases are really the hard part here. As a real-world example, one of my worst ever days occurred when we essentially rolled out a routine update which changed how a date object was serialized. This had a very serious bug and needed to be rolled back. The problem was of course we "discovered" the bug when it serialized some important objects with broken dates. When we rolled back, of course the objects were still broken and also now we were on a previous version of the code that couldn't deserialize them. This included the tools that would normally be used to fix broken data so the data rollback which would normally be simple was anything but.

Unless these sorts of things are dealt with, any framework like this will just be solving the part of the problem that isn't really a problem.