Why Continuations Are Coming to Java

Lack of Continuations / coroutines has been a huge pain in my rump for a while now; it is fantastic that they're finally adding support for it.

C yes, assembly--where? Its mapping to machine code is relatively straightforward, and as is the mapping from machine code to ELF. Or do you mean all the magic the CPU does to make it go fast?

We had a weird network issue in a production system, and because the interplay of different components was very difficult to debug I briefly investigated each component. This lead me to run Jersey (a JAX-RS implementation, ie. a REST library) with a debugger.

Jersey is written in a continuation passing style. That's the only time I've seen the style outside academic discussions of Scheme. I found the code flow difficult to grok and thought continuation passing didn't fit Java very well. That may all be just due to my lack of experience, though.

Granted, the alternatives to handling concurrency each have their own problems. Maybe this is a good idea; it will be interesting to see in practice.

Interesting that, AFAIK, computer science considers continuations a more generic mechanism than coroutines. Continuations allow continuation-passing style, creating from scratch other control flow mechanisms (like return from a subroutine), exceptions and coroutines (and maybe something else which can't be reduced to items from this list, I'm not sure). Wonder if all of that can be done with coroutines?

You are just rearranging electrons and holes anyway, and nobody understands what's happening on the quantum level of billions of tiny structure elements in silicon (plus "stuff"). When I program I always envision something like a gigantic planet sized steam punk engine, and wonder about the equally gigantic apparent disconnect between my code and what is actually happening. I'm moving matter around on an incredibly tiny scale, along unseen paths, rearranging it in different patterns, billions of times per second in the middle (CPU), collecting some in pools to which I occasionally refer (memory). It also helps me to remind myself that I'm not doing something abstract, that every thing I do has a cost and takes energy. The more stuff (matter) I have to move around, the more costly. And whether I engage a giant sub-machine for a given task or manage to build a small one that achieves the same matters.

Yes, plenty of magic.

And then I look at biological systems, not even the brain yet, just biochemical pathways. Also lots of structure, mixed in with lots of... mixing, and lots of molecules mix and sometimes match. The whole thing is much slower than man-made electronics - but many orders of magnitude more parallel. And probabilistic. And I wonder how to go from programming the man-made stuff to describing those machines in a similar way. Right now we are only tinkering at the far edges of it. Whenever somebody posts about a revolutionary new programming languages I think to myself, you are still just working with the exact same machine. It's like when you zoom in to a flat surface, that under a microscope looks ragged. The only reason we see a huge difference between various programming languages is because we are waaayyyyy zoomed into one paradigm.

In a biological system the elements doing the work are many different molecules with distinct shapes. In CPUs there only are shapeless electrons, so unlike the molecules which meet and match all the time based on probabilities the electrons don't actively contribute to the computing themselves. Also, the molecules are almost all active, all the time, in computers our "code" lies dormant and waiting in pools (memory) doing nothing most of the time. In comparison, only an incredibly tiny fraction of our human-made computing elements are actually doing something at any point!

In nature/biology shape and structure on the nano level are the major design factor. Shapes of molecules and shapes of the structures used to separate or guide or hold them. Most of the process is determined by those shapes. In human-made computing we are extremely limited in what shapes we use under the hood. We use higher frequencies and limiting ourselves to few goals, otherwise nature would outcompute us by many orders of magnitude (and that's only true because we decide to ignore most of the computation going on in biological systems as not relevant "it's just random noise with no purpose"). I see a disconnect between how we program and what is going on. We think it's a vital "abstraction". I think that abstraction is a blessing, sure, but also a great hindrance. In the end those shapes and structures matter a lot. If somehow we could get a translation into more and more flexible (nano) shapes/structures than now, where we have a 100% fixed structure and only electrons (so, no shapes)... even biological systems are actually quite limited in what kinds of shapes they use, there was a path-dependency on which random path evolution took. I think it would be possible to far exceed biological systems, but we would have to get down to the shapes (molecules and nano-structures). That is far away from now, where it takes us years to come up with one fixed structure and shapeless moving point-forms inside of it.