Matlab is used for working with matrixes (MATrix LABoratory), and instead of looping through the whole matrix, one can specify a condition on the elements of the matrix and then perform a calculation only on these elements:

x = [1 2 3]; x(x =< 2) = 0;

would produce a new matrix (‘list’) with elements [0 0 3].

As you see, this syntax does not explicitly use an if at all. A kind of selector is used instead.

in LINQ, assuming some a function netstat exists, this becomes something like

nestat(resolve=false) .Where(line => line.Contains(“tcp”)) .Select(line => line.Split(” “)[5]) .Distinct() .Select(ip => new { IP=ip, Octets=ip .Split(“.”) .Select(int.Parse) .ToArray() }) .OrderBy(t => t.Octets[3]) .OrderBy(t => t.Octets[2]) .OrderBy(t => t.Octets[1]) .OrderBy(t => t.Octets[0]) .Select(t => t.IP)

I’m guessing this will look totally unreadable, so I’ll leave a readable version over at reddit where I found this.

See http://www.reddit.com/r/programming/comments/arsu8/chaining_vs_nesting/.

You may be interested to know that this isn’t remotely unique to Haskell, or functional programming.

price > 50 ifTrue: [ "expensive" ] ifFalse: [ "cheap" ].

This could of course be simplified to the following with suitable semantics:

(price > 50) then: “expensive” else: “cheap”.

Which I think reads perfectly well.

mapM_ (putStrLn . show) . nub . sort . awk 5 $ filter “tcp” netstat_n

No pollution of the intervening classes, no syntactic noise from parentheses.

class Array def if(predicate) predicate ? self.first : self.last end end

http://gist.github.com/281687

You could introduce a new syntax, for instance using a comma to separate two inputs that should be directed to a single consumer

this thing is utterly terrible, and Smalltalk solved the problem of expressing conditionals via means of method dispatch some 30 years ago:

(price > 50) ifTrue:['expensive'] ifFalse:['cheap']

It would make more sense to have if be a method on a boolean expression, but I can’t think of a clean way to encode that without having a clean way to feed a function multiple inputs.

you know it’s possible to give a function more than one parameter don’t you?

Your example:

(print (sort_n (uniq_c (sort (awk (grep (netstat_n) "tcp") 5)))))

can be written something like

(-> (netstat_n) (grep "tcp") (awk "5") sort uniq_c sort_n print)

… you’re “threading” the first form through the remaining forms.