News for hackers. Stuff that matters.
Re:Progress on the Gilectomy by nine_k on Friday June 02, 2017 @02:05AM (#14467811)
Is there a link explaining the suggestions succinctly?
Re:Progress on the Gilectomy by Animats on Friday June 02, 2017 @03:06AM (#14468044)
- Multiple interpreters with mostly non-shared data in the same process.[1].

- VecPy, for number-crunching.[2]

- Newthreading, which I wrote in 2010.[3]

[1] https://lwn.net/Articles/650489/ [2] https://www.andrew.cmu.edu/user/dfarrow/proposal.html [3] http://www.animats.com/papers/languages/newthreadingintro.ht...

Re:Progress on the Gilectomy by simcop2387 on Friday June 02, 2017 @03:48AM (#14468221)
To be fair at least towards #1, that's a really difficult one to get right. Perl has had that kind of threading for a long time and it's almost always considered a bad decision to try to use them. I'm sure it could be done better, but perl did it via a complete copy on write setup in order to create the mostly non-shared data. But you end up with a lot of locking and other contention for the shared data because there's no good way to expose the fact that it's shared and changing it requires such restrictions. Instead the method usually used is to take advantage of most OS's built in COW when forking a new process and communicating via sockets, or some other method instead that forces you to deal with the limitations (copying data, or if you used shared memory you have to do explicit sharing and locking/serialization yourself).