Dear Marc,

What's the performance issues about make-parameter that you pointed out recently about?


Ah btw on the same note, roughly how much more work is needed for capturing the continuation with |continuation-graft| or |call/cc|, than for allocating a closure?

Just to get an idea of how expensive continuations are.

I made a trivial call/cc test on my laptop and got something like 4 million call/cc:s per second, so clearly for trivial usecases the performance is extremely good. Would be happy to understand more how it plays out in complex code/execution environments, I mean, are we talking rather a million or a thousand per second?

And.. invoking a continuation is roughly as cheap as invoking a closure right (maybe plus some phasing in and out stack content)?


Thanks,
Mikael