1. 32
    1. 5

      A bit of an off-topic comment (I’m still in the middle of reading the paper): I find it funny that a paper from Microsoft Research is typeset using the Linux Libertine, Linux Biolinum, and Inconsolata typefaces, rather than something in-house (at least I’m 95% sure it’s Libertine, the italic p is quite distinctive).

      1. 6

        Yea that looks like the ACM journal template, which would explain why. They are probably working on a submission and published this as a preprint without changing templates.

        1. 1

          This is a great area of research for FP. Performance really matters, and you simply cannot beat the performance of in-place memory updates. Many more people would be inclined to use FP if they didn’t have to take the inevitable performance hit.

          1. 2

            That’s the reason why Haskell exposed genuine in place memory updates using a guaranteed pure interface using the ST monad - basically all the algorithms on the Vector type use mutable vectors under the hood, with no way for that fact to leak (thanks to a very cool and simple type system trick).