Ruby is great, I find it super expressive and concise. But sometimes you find corner cases which are very hard to explain. During an interview on the philosophy of Ruby, Matz said:
Sometimes people jot down pseudo-code on paper. If that pseudo-code runs directly on their computers, it’s best, isn’t it?
– Yukihiro Matsumoto, creator of Ruby
So clearly it’s important for this language to be readable by humans, I guess that’s the reason why this language isn’t as mathematically strict as many others. It aims to be easily understood and it doesn’t even have a standard specification.
The lack of a spec sometimes makes it impossible to reason about whether something is wrong or intended within the language. And I have one example of this.
Take the following program:
def some_method(x) x.class end a = false b = 42
Sometimes you hit confusing features but you quickly find a good explanation:
some_method(a || b) # => Integer (ok, as expected) some_method a || b # => Integer (ok) some_method((a || b)) # => Integer (ok) some_method a or b # => FalseClass (ok?)
This may be confusing but makes perfect sense. While you can use both the
operator and the
|| they’re not exactly the same. The
have a much lower precedence, so the last call is equivalent to
some_method(a) or b. Here’s another blog post by Preston Lee
on these operators if you’re very interested. So far Ruby is clear enough.
On the following example, a good explanation isn’t found that easily:
some_method((a or b)) # => 42 (ok) some_method (a or b) # => 42 (ok) some_method(a or b) # SyntaxError (Hmm)
That’s certainly confusing, there’s no ambiguity in the last line so I would expect its syntax to be as valid as the other two lines.
I really have no idea if this is a bug or expected behavior. It might happen that this isn’t documented or described anywhere and if that’s the case the only person who can answer that question is Matz.
Postface and further read
The claim of this article, that Ruby lacks a strict unambiguous specification could be argued. So for those who aren’t aware of this topic, find below a brief walkthrough with links to sources to catch up.
There’s a ISO Ruby spec, but it got deprecated pretty much since it was released. This document is useless today.
Rubinius creator Evan Phoenix created RubySpec in 2006, a project to test Ruby implementations with actual code but this never got adopted by the MRI. Despite the project being dead now, the same idea was continued by The Ruby Spec Suite. And that’s the closest thing to a spec we currently have.