Yes! Haskell is a delight for optimization because you can use algebraic reasoning and calculation to derive more efficient code. For example, if I had the following code:

    (concat . reverse . map reverse) strs

I could easily replace it with the following code, which is simpler, faster, and reduces memory churn:

    (reverse . concat) strs

You can prove that these two implementations produce the same results (see [1] for a step-by-step proof). After you do the proofs for a while, you'll start seeing similar optimizations everywhere – and claiming them.

[1] https://gist.github.com/tmoertel/7968466

Sure they do! Optimizing is more than just low-level efficiency. It's also about using appropriate data structures and algorithms, and using advantageous APIs (like buf_concat here) when they're applicable.

This looks like a case of someone using a profiler to optimize a haskell program. A profiler is nearly always the best place to start when optimizing a program, regardless of the language.

Look. I love Haskell. I usually formulate solutions to programming problems in my head in Haskell before I write code in whatever language I have to deal with. Constructing programs using combinators? Great. Assuming referential transparency everywhere? Great. It's a great language.

But there comes a point in time where you have to project that stuff all away and get down to the brass tacks of what's happening in your system, leaving behind the Platonic heaven you've constructed for yourself. As the patch proves without a doubt, this is very possible to do in Haskell. But sadly, the general ethos of the Haskell community does not bend in that direction. I am a lover of the language, but at the same time a vocal critic of the community in that respect.

Shouldn't we optimize all our programs?