Hacker News new | past | comments | ask | show | jobs | submit login

Map-reduce is effectively the fruit of realising that, as long as your processing can be described as a combination of Functor (map) and Monoid (reduce) operations, then you can trivially parallelise your computation by letting the implementation of `map` and `reduce` handle all the parallelism for you.



A question. Does map-reduce assume a commutative monoid, or does it need to order the output of map?

Edit: seems like it's the latter.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: