3 ms·

I'm no FRP expert but Conal Elliot seems to agree with the now deleted comments: https://github.com/ReactiveCocoa/ReactiveCocoa/pull/1344 https://github.com/Re

by myko 12y ago

I'm no FRP expert but Conal Elliot seems to agree with the now deleted comments: https://github.com/ReactiveCocoa/ReactiveCocoa/pull/1344 https://github.com/ReactiveCocoa/ReactiveCocoa/pull/1344 (Conal Elliot is one of the authors of what is, AFAIK, the earliest paper on FRP: http://conal.net/papers/icfp97/ http://conal.net/papers/icfp97/)

seanmcdirmid 12y ago
Here is Conal's comment duplicated for prosperity: > All of these changes so far look good to me. They definitely helps reduce confusion! Under "Here are some resources for learning about FRP", I'd drop the "Principles of Reactive Programming on Coursera" and the two Elm references, as they're both quite different from (the original) FRP. Based on the Coursera description, I'm guessing that it is about Rx. Elm is a system conceived independently from (not even "inspired by") FRP, which bears some superficial similarities, but doesn't have either of the two fundamental properties I mention in the StackOverflow posts (precise & explicit denotation and continuous time). I've spoken with Elm's author, and we'll hopefully get that confusion sorted out. He is definitely right about Rx, but Meijer also never claimed as much (since they are also WG 2.8 members, I'm sure they talk a lot). One of the tenants of FRP is continuous time, and you lose that once you update only at discrete events, which IMHO, is a very sane choice to make. This divergence with classical FRP first appears with Cooper et al's FatherTime work and the later Flapjax work by the same authors. Conal has made some effort to combine the pull (sampling) and push (demand-driven) versions of FRP: http://conal.net/papers/push-pull-frp/ http://conal.net/papers/push-pull-frp/ But still, it makes a lot of sense to go full on discrete. Krishnswami provides some good results in this direction: http://www.mpi-sws.org/~neelk/popl074-krishnaswami.pdf http://www.mpi-sws.org/~neelk/popl074-krishnaswami.pdf I'm not sure, however, what Conal thinks of his work.

by myko 12y ago

seanmcdirmid 12y ago

Here is Conal's comment duplicated for prosperity: > All of these changes so far look good to me. They definitely helps reduce confusion! Under "Here are some resources for learning about FRP", I'd drop the "Principles of Reactive Programming on Coursera" and the two Elm references, as they're both quite different from (the original) FRP. Based on the Coursera description, I'm guessing that it is about Rx. Elm is a system conceived independently from (not even "inspired by") FRP, which bears some superficial similarities, but doesn't have either of the two fundamental properties I mention in the StackOverflow posts (precise & explicit denotation and continuous time). I've spoken with Elm's author, and we'll hopefully get that confusion sorted out. He is definitely right about Rx, but Meijer also never claimed as much (since they are also WG 2.8 members, I'm sure they talk a lot). One of the tenants of FRP is continuous time, and you lose that once you update only at discrete events, which IMHO, is a very sane choice to make. This divergence with classical FRP first appears with Cooper et al's FatherTime work and the later Flapjax work by the same authors. Conal has made some effort to combine the pull (sampling) and push (demand-driven) versions of FRP: http://conal.net/papers/push-pull-frp/ http://conal.net/papers/push-pull-frp/ But still, it makes a lot of sense to go full on discrete. Krishnswami provides some good results in this direction: http://www.mpi-sws.org/~neelk/popl074-krishnaswami.pdf http://www.mpi-sws.org/~neelk/popl074-krishnaswami.pdf I'm not sure, however, what Conal thinks of his work.