On Mon, Feb 16, 2015 at 10:50 AM, Vladimir Komendantskiy < komendantsky@gmail.com> wrote:
Yet I wonder why do you have a requirement to run code in the interpreter on an automnomous vehicle? Wouldn't compiled code run more efficiently and require less resources?
Because the ability to change the programming in the field is important. A lot of the configuration is pure functions that get iterated over the state to produce new state, and expressing those in Haskell is where I think this can win. Ok, there'd be an advantage to writing it in Haskell in the first place, but that's a hard sell. On the other hand, replacing a text file in INI or XML or YAML or some ad-hoc format that specifies what is, when all is said and done, a pure function that takes the state of the hardware as input with an actual Haskell function, that I think I can sell.