On 18/08/17 19:45, Ignas Brašiškis wrote: >On 18/08/17 09:51, Andrew Haley wrote: >> But we're being asked to believe that, all of a sudden, >> language support for coroutines is critical! > > Who said that really? On 17/08/17 16:05, Avi Kivity wrote: > I will probably switch to clang soon, despite losing concepts and a > subjective preference for gcc, because of this feature. On 18/08/17 19:45, Ignas Brašiškis wrote: > I did not used word "critical" and was wondering why it is so. See above. :-) > Also unlike 50 years of existence now seems that syntax are getting > constant in many programming languages and does not feel like spit > out hack. It didn't seem like a hack then, either. > OOP also existed before classes via form of function pointers and > structs, but it got its form from smaltalk era onwards, that > everyone is familiar. I'm not sure I agree with you about that. There wasn't much time between these things: they were developed at around the same time, in the 1960s. > Really take boost.coroutine2, some Macro ridden coroutine libraries, > or any c coroutine task library and compare what N4680 draft proposes. > Which of them is feel more natural and easier integrate to base C++? > > No one suggests it is "critical". But do not think that interest is > special or strange, it just isn't. Oh, please. SIMULA had coroutines working perfectly well, and they were very useful, and that was *exactly* fifty years ago! I am just amused, that is all. But there is a serious question behind all of this: why now? Coroutines are not particularly difficult to implement, and could have been done at any time. I'm not opposed to doing it now! -- Andrew Haley Java Platform Lead Engineer Red Hat UK Ltd. <https://www.redhat.com> EAC8 43EB D3EF DB98 CC77 2FAD A5CD 6035 332F A671
