Book contradicts itself

[Pyriphlegethon]

In chapter 2.3 Dining Philosophers the map function is used to call a closure for every philosopher. The closure however is only executed for its side effects (it doesn't actually calculate anything). The chapter on iterators however states that one shouldn't use map for side-effects, but simply use a for loop.

[petrochenkov]

The closure however is only executed for its side effects (it doesn't actually calculate anything).

How is that? The closures calculate thread handles which are then collected by collect().

[steveklabnik]

Thank you for the report!

Yes, @petrochenkov is correct. If we used a for loop here, we would drop each handle in turn, which would make things not paralell. The purpose is to map closures to handles.

Maybe we can make this more clear somehow?

[Pyriphlegethon]

We wouldn't necessarily drop the handles, we could still insert them into a vector. But you're correct, the map does in fact calculate something. Maybe we should add that using map without using its result is bad? Although the compiler will complain about that anyhow.

[sfackler]

Just using map alone is bad because it doesn't actually work :). The thing the book is warning against doing is something like

things.iter().map(|thing| thing.update()).count();

Where the call to count consumes the iterator, making all of the update calls.

[steveklabnik]

Yeah, given that the compiler gives a warning if we remove it, I think that we can chalk this one up to misunderstanding. If anyone wants to somehow improve this, please send a PR. Thanks!