chapter 9 why suddenly remove optimization code entirely in e-coroutines-pin example ?

[baiwfg2]

I think introducing Pin is to solve the crash problem in c-coroutines-problem example. But you said Let’s remove our optimization entirely in executor.rs, which is quite weird to me.

Then how do we verify that crash issue is solved by Pin usages without the code triggering the crash ?

[cfsamson]

Sorry for the late response. I'm working on a new edition and putting most of my time towards that. Pin wasn't introduced to solve that crash, but to prevent us from making that kind of mistake. The optimization we made is not possible to do safely the way we did it. It's hard to come up with non-contrived reasons to introduce Pin, but this optimization is a real example of something that could make sense to do (a lot of futures and runtimes resolve at first poll), but can't be done the way we did it due to the self-referential nature of the futures generated by async/await.

I could have tried to make it work another way with Pin but it would add a lot of code and the main motivation was to show how Pin would give us compile time errors in this case.

Again, sorry for the slow response on this one.

[harshRn]

Hi @cfsamson big fan of your work! I really think chapter 9 as a whole could use some kinda re-work. I think you explained pinning and the stuff around it extremely well to the point where I don't need to go to some other resource to learn more about it. But the stuff discussing the integration of pinning and how it fits into the picture seems rather artificial . I wish I had a better adjective to describe what I am feeling about it😅 . But I think you already know this, that's why you call your example 'contrived'. Your book is perhaps the most holistic non-operating system book about asynchronous programming, and I think adding just 'a lot of code' to make it fit better would be worth it.
very much looking forward to the second edition! Cheers!

[cfsamson]

Hi @harshRn and thanks for the kind and constructive feedback. I'm very happy to hear you enjoy the work that went into this book.

I'll try my best to come up with something better here even though it might require a little bit more code. I just don't want to make it too complex to explain my point. I have some loose thoughts that I'll try out when I get to this chapter to make this part of the example a little bit less contrived without adding too much complexity. I'll see what I can do 👍