@bascule @voidspace @alex_gaynor We've also had years of real-world high-concurrency users rousting out bugs and bottlenecks.
@rich0H @envygeeks @voidspace yeah, that's what I thought. I like shared heaps FOR SPEED!
-
-
@bascule@envygeeks@voidspace I prefer IPC like sendfile(2) over shared heaps. That's personal though, not "this is better than that". -
@rich0H@envygeeks@voidspace there's a lot more complexity and inefficiency with that kind of approach, but it has some merits -
@bascule@envygeeks@voidspace inefficiency? MMU dependant I guess. scales across machines transpaently, but both valid. Case by case imo. -
@rich0H@envygeeks@voidspace having separate heaps per process and requiring context switches to copy data between procs is slow -
@bascule@envygeeks@voidspace sendfile(2). Zero copy. If you're multicore there's no context switch. -
@rich0H@envygeeks@voidspace o_O I'm a big fan of sendfile but it's not really comparable to a lock-free data structure on a shared heap -
@bascule@envygeeks@voidspace I'm not saying "Shared heaps are bad". I'm saying they're not a magic bullet, and it's case by case. -
@rich0H@envygeeks@voidspace they're definitely more efficient. I'd challenge you to write something like Disruptor without one - 9 more replies
New conversation -
-
-
@bascule@rich0H@voidspace Need for Speed: Python Edition. Coming soon to a programmer near you.Thanks. Twitter will use this to make your timeline better. UndoUndo
-
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.