Comment by smcl | Hacker News Reader

Comment by smcl | Hacker News Reader

Parent Full thread

Yeah I don't think anyone likes over-engineered architecture astronaut code with too many layers and unnecessary abstractions, whether that's been formatted in the Whitney style or in a more conventional one. I think what I can't get over are the short identifiers (and filenames) and the way it just looks like a wall of text without any breathing space, though looking at another example someone posted[0] it seems there's a bit more whitespace and structure than I remember.

> Learn to read and it turns out not to be unreadable?

There's the thing, if someone learns Russian they can converse with 140 million people in Russia, similarly Ukrainians and Belarusians will be fine and they could probably make themselves understood through the Caucasus and Central Asia. If you learn to read C written in the Arthur Whitney style you can "converse" with a fairly small number of people who like the Whitney style[1]. So taking the example a bit further, I learned the Cyrillic alphabet in an afternoon and through knowing another Slavic language I can roughly parse the meaning of many Russian things I read (audio is another thing entirely, I can only pick out a handful of Czech/Russian homophones). If I had gotten up to speed with the ngn/k codebase would I be productive on one of the projects you wrote in a similar style, or is there a similar productivity wall where I'd have to first learn some idioms local to your codebase?

Sorry for the questions, I know people who like this style probably have to answer these questions fairly frequently. I am genuinely just quite curious though.

[0] = https://codeberg.org/ngn/k/src/branch/master/h.c

[1] = is there a proper name for this or is it ok to refer to it as "Whitney style"?

I don't know if there's a proper name for it. At least people who are aware of the style would probably recognize Whitney's name so that's the best term I've come up with yet.

For me, the end goal isn't Whitney style, but I've been pursuing effective programming all my life. When I learned to code, I wasn't talking to anyone except my computer, and that alone was exciting enough to make it a life-long hobby and profession for me.

Do you know what brought me to Hacker News? Arc. And Paul Graham's writings about Lisp. The message was never about a popular language that everyone speaks. If anything, it was rather the opposite: pg saw in lisp a powerful and expressive (if niche) language that makes you competitive against larger players who stick to boring mainstream languages. I wasn't interested in competition or startups, but merely in powerful ways to make the computer do what you want. I don't particularly care if I'm the only person on planet earth willing to wield that power; it's for my own enjoyment. Programming for me isn't about "product" so "productivity wall" isn't something I think about, and complaining about productivity wall would be a bit like complaining that getting fit for Tour de France is time consuming, why don't they just drive it by car?

That said, I think there are people who find K, APL, and related languages very productive in their niche. I'm definitely not speaking for everyone.

Anyway, it is the curiosity and desire to discover a powerful way to command the computer that has driven me to study Haskell, APL, PostScript, forth, TLA+, lisps, BQN, K, Erlang, and more. "Whitney C" is just one milestone along the journey, and I don't know where the journey will eventually lead; I'm just not happy with any existing language right now.

So the answer is no, learning Whitney C will not make someone immediately productive with my code, just as learning Java does not make you immediately productive C++. They are different languages. However, anything you learn can shrink the productivity wall; knowing APL or BQN, K, and Whitney C might make it easier to grasp whatever I come up with next. That applies to all of programming in general though; the more you know, the more you know, and some of that knowledge will almost always transfer. There will be familiar patterns and ideas.

I also think people seriously overestimate the productivity wall. As you say, one can learn to read cyrillic in an afternoon. Kana in a weekend. But learning Russian or Japanese is significantly more work than learning the script. In terms of scope, I'd say learning APL or Whitney C is closer to learning kana than it is to learning Japanese.

(EDIT: I also find it ironic that programmers are ostensibly excited about learning new things, yet at the same time programmers really love to complain about languages that look alien won't give a few weekends to learn them)

Good to know re the name, I typed it out a couple of times and wondered if I was just doing something stupid. I really wasn't trying to formulate an attack on this style, if someone or some organization uses it and it works then more power to them. I was really just trying to understand a bit better, but it's possible that is something that I can only really get by experience.

So I find APL, J, K and friends quite fascinating (and J is on my list to try) but I haven't seen much hostility to them. People understandably get a bit intimidated by how different it is but they usually still seem curious. The real hostility is reserved for Whitney C. In this case I don't think it's like - if you'll forgive me for abusing that human language metaphor a bit - an English speaker learning Russian, more like

    ifAnEnglishSpeakerEncountered
    aLocaleWhereEnglishIsWrittenA
    ndStructuredDifferentlyToWhat
    theyWereUsedToSomethingLikeTh
    is.

I can understand why their instinct is to recoil in horror and think "I already know a more standard English, I'm currently learning Russian and Japanese ... I have little patience for trying out this alternate form of English". It's obviously an exaggerated/contrived example, but this is genuinely how that C code appears to an outsider at first blush (or at least it did to me and a couple of my friends).

That said your replies have piqued my interest, I'm gonna have to properly dig through that ngn/k repo some day. If I turn into a Whitney C guy then I'm holding you responsible :D

Yep. It's mostly a knee jerk reaction.

After accustomization with a style that doesn't force explicit declarations of identifiers and their types, verbose type conversions, line breaks and indentation after every statement and brace, etcetra., one could definitely make a different (and similarly exaggerated) human language metaphor. For example, take some English text and feed it through a parser. Feels good to read?

    (S (NP Parsing)
       (VP refers
           (PP to
               (NP (NP the activity)
                   (PP of
                       (S (VP analysing
                              (NP a sentence)
                              (PP into
                                  (NP its component categories and functions))))))))
       .)

That's a bit how mainstream languages feel after using something that hasn't been forced into such an artificial form :) If you're willing to let go of that, you can write sentences and clauses on the same line, almost like prose!

Hehe that's actually a nice way to put it. So it's a little bit like the red pill, you can't really go back after going embracing Whitney C :D

Reply on news.ycombinator.com