[ The Types Forum, http://lists.seas.upenn.edu/mailman/listinfo/types-list ]
> On Mar 20, 2018, at 6:49 AM, Neel Krishnaswami > <neelakantan.krishnasw...@gmail.com> wrote: > > [ The Types Forum, http://lists.seas.upenn.edu/mailman/listinfo/types-list ] > > On 19/03/18 13:52, Zachary Palmer wrote: >> The first is prior work involving typed function destructors other than >> application. Cloud Haskell seems close, as the runtime essentially allows a >> restricted form of transmission of lambdas over a network and that requires >> the closures to be serialized. I can't help but think, though, that >> there's some core theory I'm missing on the topic. Is anyone familiar with >> any work of that sort? > > I do not know what a "function destructor" is, but the theoretical > account of the ability to send functions over the network that is > closest to what Cloud Haskell is doing is given in Tom Murphy's > PhD thesis, "Modal Types for Mobile Code” In the same spirit the work of Heather Miller et al. [1] relevant. It explains the Scala support for Apache Spark (no worries, the paper does contain theoretical material), which sends closures to large sets of data in large clusters of computers. — Matthias [1] https://www.cambridge.org/core/journals/journal-of-functional-programming/article/programming-model-and-foundation-for-lineagebased-distributed-computation/B410CE79B21E33462843B408B716E1E5
