Jon Pretty

I think there's a tendency for FP purists to exaggerate the brokenness of software. Most people have lower standards. But perfect purity is the only position from which anyone can claim the absence of certain classes of bugs... and they do.

And they get confidence from the fact they write thousands of lines of code, and they never see certain kinds of bug. And from their perspective, amongst those thousands of lines of code, a single impure method call will look egregious.

Are there any Scala and Haskell bugs that programmers feel confident don't exist because the compiler assured them?

As one example, if you have no mutable state, there's no chance of a race condition caused by another thread mutating some state underneath you. In Haskell, you can't mutate state, though in Scala you rely on being careful not to use any mutable state.

So in Scala, the compiler guarantees it, iff you're careful, so it's not such a compelling guarantee. In Haskell, you're forced to write things in ways which many programmers would find awkward, but the guarantees are there.

As in my other tweets, I use Scala in a very immutable way, but not entirely immutable because actors save me some problems by preventing races. Still, I prefer to use a case class to maintain immutable state at the instance level because it frees my mind from mutability burden.

Yeah, that's a reasonably good approach, and Scala facilitates it well. Your state is at least all in one place. There is nothing to stop you mutating global state defined elsewhere, except convention, though.