2 ms·
There's more to report than simply an awesome library. Two days ago, someone allegedly unfamiliar to the project (Paul Banks) appeared from the ether, applied
by Dowwie 11y ago
There's more to report than simply an awesome library. Two days ago, someone allegedly unfamiliar to the project (Paul Banks) appeared from the ether, applied a few "simple" patches, and improved the performance of libmill 7 fold (for OSX at least). That just doesn't happen, ever, in open source.
https://twitter.com/sustrik/status/666360179802402816 https://twitter.com/sustrik/status/666360179802402816
- thechao 11y agoThe twitter message you linked to says this: The speedup was caused by replacing setjmp by _setjmp though. I know this may sound mysterious, but for BSD/OSX... "yes". It is "well known" (if you're the sort of person who writes coroutine runtimes [1]) that _setjmp is far faster than setjmp. It's morally equivalent to 'sigsetjmp', but more portable, if less standard. [1] https://github.com/jaroslov/coro/blob/master/coro.hpp https://github.com/jaroslov/coro/blob/master/coro.hpp