News for hackers. Stuff that matters.
Re:Guy Steele: "How to Think about Parallel Programming: Not" by silentbicycle on Saturday January 15, 2011 @10:33AM (#2107122)
> To the extent that I have a programming style, I find myself increasingly thinking of it as Primitive-Oriented Programming. If you want your final program to have some desirable property, be it correctness, scaling, parallelizability, or even more complicated things like one I work on where the interface must have the property "able to be remoted across this certain protocol for various reasons", the best way to achieve that in general is to create primitives with this property, create ways to combine them that maintain the property, and build up from there.

Do you have any experience with the APLs? They also strive for this sort of functionality - they're based around a set of primitives which have implicit looping and high potential for parallelism. I have been experimenting a lot with k (kona) lately; it and J (http://jsoftware.com) are both definitely worth a look.

Re:Guy Steele: "How to Think about Parallel Programming: Not" by silentbicycle on Saturday January 15, 2011 @12:51PM (#2107374)
Also, I have no doubt that GLS is influenced by APL - he mentions in an interview (it was either in _Coders at Work_ or on SE Radio (http://www.se-radio.net/2006/11/episode-36-interview-guy-ste...); I recommend both) that APL was one of the first programming languages he learned.
Re:Guy Steele: "How to Think about Parallel Programming: Not" by jasonwatkinspdx on Sunday January 16, 2011 @05:11AM (#2109106)
If you're interested in this thread I highly recommend Backus's "Can Programming be Liberated from the Von Nueman Style". It covers both the evils of accumulators, as well as creating languages with provable algebraic properties. Certainly I don't think anyone today would really enjoy programming in FP or FL, but I do think his thoughts are ripe for modernizing.