[externalreality]

I tried to ask the author the follow: (kept getting deleted as spam). Perhaps he will see it here but its unlikely due to the fact there are many comments as it is.

Hi John,

Are you familiar with Jackson Structured Programming?

https://en.wikipedia.org/wiki/Jackson_structured_programming

Notice how the focus in on using control flows that are derived from the structure of the data being processed and the processed data. Notice how the JSP derived solution in the Wikipedia example lack if-statements.

Pattern matching allows ones to map control flow to the structure of data. What are your thoughts on that? I think inversion of control has other benefits but I don't think it has much to do with elimination of `if` conditionals, the pattern matching does that.

Also, I noticed one thing:

In the article you mention `doX :: State -> IO ()` as being called for its value and suggest that if you ignore the value the function call has no effect. Isn't it the case that a function of that type usually denotes that one is calling the function for its effect and not for any return value? Its value is just an unevaluated `IO ()`.

[mbrock]

The return value of the function is a description of an effect. Calling the function doesn't cause the effect to happen. That's why you could, for example, call the function many times and get a list of IO actions which you then execute in parallel or backwards or whatever. Hence "inversion of control".

[externalreality]

I was debating whether on not to put that last sentence because I knew that it would lead to a technical discussion that was aside from the meaning of the question. My question is more -- why choose an `IO ()` as an example of something being called for its value (especially since the article isn't aimed at a Haskell audience)

[eru]

Yeah, that's probably not a wise decision on part of the author. The IO monad is nifty but of minor importance in the grand scheme of things, and distracts when making a mostly language independent point.