mimblr.com

While I didn’t find much of interest, I have to admit that the line: [Monads] have been turned into a tool of intellectual terrorismmade me laugh.

He may mean this paper: http://homepages.inf.ed.ac.uk/wadler/papers/marktoberdorf/baastad.pdfIf that doesn’t help there are a number of other monad related papers on Wadlers page here: http://homepages.inf.ed.ac.uk/wadler/topics/monads.html

You might be interested in “implicit parameters” in Haskell, for passing around configuration.

I assume you’re joking. My cat meows at the door because it wants to go out. That does clearly show an “if I do this, then that happens” form of thinking.

What do you mean “give it access to Monad”?Giving something access to a piece of configuration information should just involve adding some function parameters. I suppose if you want to be really fancy about it, you could use the Reader monad, but that’s usually excessive. Just add a parameter for a record consisting of your configuration data, and pass it down to wherever it’s needed. It’s really not as painful as you think.The article itself is pretty incoherent. I have no idea what it means by “monad bundle”, and it seems pretty obvious that whoever wrote it has spent little to no time whatsoever programming in Haskell and actually using monads.Monads are not generally intended to be combined, and when they are, the result is generally something nontrivial, and probably deserves its own module. In order for the composition of two monads to be a monad, you need something called a distributive law.Haskell programmers almost never use this method of combining monads, and you won’t find it in the standard libraries. If you’re interested, Conal Elliot’s TypeCompose defines an appropriate class for distributive laws, and an instance of Monad for the composite.There are monad transformers, for building up more complex monads from simpler ones, but this is not the same thing as combining monads, though it does in some sense combine the effects of certain monads, and again, it’s not something to be taken lightly. You generally stack a bunch of monad transformers to get exactly the monad which you’re interested in, or perhaps something somewhat more general, then newtype that and hide the fact those monad transformers ever existed for users of your library. For detail on that, see my article How to use monad transformers.In general, it just doesn’t make sense to mix computations from different monads. You usually apply some ‘run’-function to carry out the computations built up in one monad and then use the results in a completely different one. For example, in a compiler, you might use the IO monad to read in a file, and then pass the resulting string off to a parser written in some parsing monad, get back a syntax tree, then manipulate that with some pure functions into, say, some C code, and eventually do some output in the IO monad. None of this really involves mixing effects from different monads at the same time — for example, doing a bunch of I/O in the middle of some backtracking parser might be useful in rare cases, but for the most part would be pretty unpredictable and dangerous. If you really want that, there’s a monad transformer version of Parsec, but I’d recommend against doing it unless you were really sure you couldn’t do the I/O before or after parsing. (Which in my experience, basically never happens, and you can always factor the I/O out. At worst, you can have your parser return an IO action to be carried out afterward.)The only reasonable places to use unsafePerformIO is in adding new features to the evaluation model of the language, or new primitive data structures. Consider it similarly to modifying the Haskell compiler. There really is a better way to do everything else.I think a lot of these complaints also come from trying to use fancy monads for simple tasks which are better handled with pure functions and parameter passing.A lot of the simpler monads are just ways of factoring out certain patterns of parameter passing. The flip side of that coin is that you don’t know which monad you want until you understand how to write the thing without using a monad. I see far too many beginners jumping into solutions which use some complicated combination of monad transformers without first seeing what the code looks like written in a direct style.

If you’re just throwing out ideas, why get so pissed off? I’m just throwing out ideas myself.I’m just saying that as internet rewards go, Karma is a pretty weak one.On some forums I’ve been too, you get “reputation”, little green bars under your username, avatar and post count. There are two main differences-Everyone can see your “rep” whenever you post, and it’s like a little badge you wear. (Same with post count)People on the forums develop a pecking order, with stars and nobodies.In the case of reddit, nobody can see your karma unless they really want to. And aside from say the 911 was an inside job guy and maybe Captain Obvious, there are no “stars”, at least that I know of. As far as I can recall we’ve never talked before, and likely won’t again. Its a lot more anonymous than most forums. So far as I can tell, there’s no social ladder to climb. I would love it if you could credibly prove that no one else here cares about karma.Well, I would love for you to prove everyone, or even many people, do. I can’t prove they don’t, you can’t prove the opposite. We’re both just throwing out ideas.But let me ask you this- who do you know for a fact that karma whores? I can’t think of anyone. Why do they do it? I can’t think of any reason.