Comparing boost.coroutine2 (stackful) to stackless coro implementation...

[germandiagogomez]

Well, needless to say, but, unless I am really confused... it does not make even sense to compare stackful to stackless coroutines and conclude that co2 implementation is faster. It does not even mention that they implement different things and that stackful is more costly by its own nature and capabilities.

[jamboree]

Are you mentioning the comparison at Performance section?

It's just for showing the overhead of context-switch. What comparison do you want to see?

[germandiagogomez]

Are you mentioning the comparison at Performance section?

It's just for showing the overhead of context-switch. What comparison do you want to see?

That was my point: you are comparing a stackless context switch to a stackful context switch and concluding that your library is faster with no mention about the differences and difference in power (if I am not wrong and I misunderstand something).

[jamboree]

I don't think I have made any conclusion here, anyway, it's just a very basic comparison of context-switch overhead. For the differences in stackless/stackful coroutines, you can see my answer here.

[ozra]

I, for one, like the benchmark and the comparison of techniques. That's what I want to know: what can I gain from fitting my code in stack-less vs. full blown stack-switched co's. Please keep it.

[germandiagogomez]

@ozra There is nothing wrong in keeping the benchmark itself. It is just that it is misleading how things are explained. :) Just my two cents.