3 ms·
Most of the problem is the `return` keyword. In a proc or lambda, it returns from the proc or lambda. In a block, it returns from whatever is enclosing the bloc
by raganwald 18y ago
Most of the problem is the `return` keyword. In a proc or lambda, it returns from the proc or lambda. In a block, it returns from whatever is enclosing the block, like a proc, lambda, or method.
If you had a block that was an actual object, it could outlive its enclosing environment. Calling return within it would reincarnate its enclosing environment. Calling return twice would fork time.
In other words, if a block was an object, the return keyword would create a continuation. This is a very expensive, very heavyweight feature in Ruby, and the whole point of blocks is to provide something cheap and lightweight.
Note that a proc or lambda has no such problem, since a return within it returns to whatever invokes the #call method of the proc or lambda.
Imagine how much simpler things would be if Matz had worked this all out and decided that the return keyword was more trouble than it was worth?
- aaronblohowiak 18y agoI respectfully disagree that the return keyword is more trouble than it is worth. "return false if simple_question?" ... rest of the body is a very useful thing to do. granted, you might say it is "programming by GOTO", but it lowers the indentation level, and so keeps me happy. To the GP: we can also .to_proc on blocks.
- raganwald 18y agoSorry I didn't see this when you posted it. The problem is not `return :foo if bar()`, it's something like `pizzles.each { |p| return :foo if p.bar() }`. In Ruby, that returns from the surrounding method, not from the block where it's defined. You can decide for yourself if this behaviour of return more trouble than it's worth. I didn't say it was, I said that this definition is what forces a block to be something different from a proc, which makes the language more complicated.