Haskell and the elegant attack.

Welcome back to the Happy Haskell HAckers Tour. As we approach our next stop, let me firstly apologise for the recent detours. Our bus driver, Yoneda, mixed up the maps and started following the GHC After Dark1 script. But now we’re back on track, and coming up on your left is our monument to the Haskell literals:

-- | motto
--
-- We may need to think more about this (SPJ)
--
motto :: String
motto = "avoid success at all costs"

For a long time, through the darkness of pointfree, the incompatibility of lhs, and the golden age of DerivingVia, this motto has guided Haskell language design. If you peer closely, you’ll notice some scratchings of the original work. These are various attempts at interpretation involving commas and parenthesis, by our bright but cheeky novitiates, but we’ve done our best to restore the work to its original intent.

Archivist retrieval digs have recently confirmed the original context of these words, and now believe it is not a motto at all, but part of a larger computation2, one that has been recovered3 from a lost functional pearl4

-- | The literals
--
-- Ok, ship it (SPJ)
--
literals :: [String]
literals =
  mconcat $
    fmap List.unwords <$>
    (combinations <$>
     [1 .. 5] <*>
     pure
     (List.words
      "avoid success at all costs"
     ))

Whilst we do not have time to cover all of the technical details5, what has previously been described as the Haskell motto, is, in fact, simply a seed string that generates a lexicographically ordered list of aphorisms (the literals).

Note how the seed value is embedded directly in the computation and cannot be changed. One can imagine inserting another motto into the computation, but this would, literally, not be Haskell.

Note also that any combination disrespecting the lexicographic ordering of the seed literal is considered heresy, and will regardless be unavailable to the compiler. Problematic examples that have caused confusion in the past include “success avoid” and “costs avoid success.”

But enough with the theory, if a volunteer would like to step up and spin the literal combinator, that big button marked random-1.2.0, let’s meet the Haskell literals.

-- >>> literal 5
-- avoid success

Or, as Rudyard Kipling (an early adopter of category theory) wrote on the steps of Wimbledon, “meet with triumph and disaster, and treat those two impostors just the same (or at least, up to isomorphism).”

Quoting from the lost pearl, “maybe we should avoid labels such as success being placed on language design. Either they will turn out to be correct, in which case the label is redundant, or they will eventually need replacement, in which case the label becomes an impediment to adopting better ways.”

-- >>> literal 1
-- success

My favourite literal, and one of the singletons. The singletons are best understood as interjections6, so an accurate translation is actually Success!.

In the lost pearl, it reads literally as “joy of haskell.” Some other interpretations include “I can’t believe it compiles!”, “Huzzah!”, “Poggers! Let’s goooo!” and, the dryer, “With Haskell, if you can get it to compile, it usually runs.”

-- >>> literal 0
-- avoid

Or Avoid!, the head of the literals. To quote the ancients7, “In a functional language with lazy evaluation, we can go a step further and eliminate the recursive definitions … in favour of a more modular solution.” A more modern version is “a bug can only exist if it’s representable.” When laziness is inherent and exposed in language design, we can all get to the point quicker and have some down time.

Avoid!

The Elegant Attack Proclamation

-- >>> literal 30
-- avoid success at all costs

Now Yoneda and I could sit here all day, watching the combinator spin8 its way through Hask -> Hask, but that’s our day job, and some of you may be getting bored, so let’s skip to the last of the literals.

In a recent enunciation9, literal 30 was described in these terms:

Haskell embodies a radical and elegant attack on the entire enterprise of writing software.

The elegance comes from how the literals combine to form this emerging narrative. For example, this is not the elegant attack but embodies it:

Haskell is a non-commercial (avoid all costs10), volunteer army (success at all costs11) who would like coders (forall coders.12) everywhere (avoid success at13) to have rich (success at all14) lives, with less tedium (Avoid!) and the best (Success!) tools.

Please, a round of applause, and spare change if you have it, for our real-life combinatorial choreographers of language design, who juggle and balance the literals day in and day out, in their quest for software perfection.

And until next time, when we visit the GHC foundry, where tar balls are rendered and frozen, and libraries machined to isomorphic perfection, Yoneda and I will leave you be. Feel free to spin the literal combinator some more, and ponder the Haskell literals and the elegant attack they may formulate.

Finally, literal 7 (avoid all)15



  1. Unfortunately, the GHC After Dark tour is solidly booked due to unprecedented demand. ↩︎

  2. The code is not the exact original, but has had some doctrinal modification, including the qualified List obligation, the applicative noise injunction, the trailing operator cascade and, of course, monadic purification. The use of [String] may grate the modern ear, but the old interface is retained to allow interaction with the other Haskell List String mysteries. ↩︎

  3. Found in a recursively-defined Windows backup directory, of all places. ↩︎

  4. GitHub - tonyday567/lits: The Haskell literals ↩︎

  5. Just as an aside, to quote from the lost pearl: “For the combinations operation, we use a functional algorithm similar in spirit to Knuth’s algorithm R. The imperative version is also known as Knuth’s revolving-door algorithm, but in the functional equivalent the door does not so much revolve as remain in a fix point in relation to the grey code generated, saving the computation costs of actually revolving the door.”" ↩︎

  6. An interjection or exclamation in English is similar to BangPatterns, but with StrictData also turned on, that can be used to interupt concurrent conversation. ↩︎

  7. R.S. Bird; John Hughes (1987). The alpha-beta algorithm: An exercise in program transformation. ↩︎

  8. or remain a fix point in the natural transformation of category theory to software design, if you prefer. ↩︎

  9. The Elegant Attack Proclamation can be found at haskell.foundation, just above “Learn about Haskell.” ↩︎

  10. literal 20: Avoid all costs Haskell, as a project, is poor and shall remain so. A price point of zero ensures wide adoption and a collegiate approach to the craft of coding. ↩︎

  11. literal 29: Success at all costs For all that they are, this bunch of misfits is pretty focused on the tasks of writing software. They are not in the mix to take some small slice of existing commercial arrangements. ↩︎

  12. literal 3: All! Also written forall coders. ↩︎

  13. literal 15: Avoid success at Haskell is a general purpose language. Although we are quite good at parsing, do not box us in. Success at a particular endeavour does not necessarily mean we should specialise16, but instead maybe work on our weaknesses. ↩︎

  14. literal 21: Success at all We would like coders to have balanced lives, with less boiler-plate, and for teams of coders to use and embrace diversity, so that success may be shared widely. ↩︎

  15. For passengers who have specifically complained, again, apologies. For those complaining that they did not sign up for any tour business, this is, in reality, not a bus tour but an allusion to one. A sketch to set a satirical scene in which to portray the craft of software design. In my defence, the use of allusion and metaphor used to be stock in trade for a functional programmer back in the day. If you read any old functional pearl you will find an author playing with the English language alongside their functional language expressions. Put the fun back in function I say but, regardless, you are free to click away at any time. I won’t name names, but I understand that, for specific cultures, satire is seen as equivalent to sarcasm. 40 years of SNL has robbed this culture of seeing any joy or point to parody, and that is sad. Treating subject matter whimsically is not equivalent to treating it as a joke. To the passenger who alluded to doxing Yoneda and I, on the main charge of sloppy writing, if I understand the complaint, and described uncertainty surrounding the fictional status of the piece as unbearable, I say good day ma’am, but your ride is over. ↩︎

  16. As, Heinlein, one of our early adopters of literal 15, scribbled on the side of a punch card, “specialisation is for insects.” ↩︎