Sean Werkema's Blog

I swear there is no part of the Node ecosystem that isn’t a flaming tire fire. Does anyone ever design anything before implementing it anymore?

Try this example:

• NPM is a package manager. But you can’t install packages from local source (i.e., on this computer) in older versions because NPM was only really designed to work with a single remote repository. (New versions at least let you add multiple remote repositories, as a somewhat hacky extension.)
• There’s a workaround hack feature called npm link that lets you set up a symbolic link — a pointer — to a local package. It doesn’t work like a standard link command, so you’ll have some relearning to do if you want to use it.
• But nevermind that learning, because npm link has no shortage of problems and bugs and failings and workarounds and workarounds to the workarounds. Also, it has disastrous data-destroying bugs on Windows.
• But nevermind the bugs, because it’s not compatible with the normal npm install commands anyway, so the one use case you’ll have for it — installing packages from a local directory to make sure your design works — isn’t really usable with it anyway.

None of that would have been an issue with a little more advance thinking and a little less advance typing. Two small design changes early on — (1) allow multiple repositories to be listed in .npmrc and check them in order, and (2) allow the file:// protocol to point to a local directory of .zip files or tarballs as a repository — would have been enough to obviate all of this. NPM scopes, link, yalc, lerna, and a dozen other things built on top of the initial bad design would have been completely unnecessary.

But, no, can’t spend time on that! — you gotta move fast, be first, break things, change the paradigm, own the market, declare your IPO, and retire like a king. Never, ever design or plan anything, because that takes too long! Just shove something barely-passable out the door, and if enough people use it, make a slightly better layer on top to hide the roughest parts, and then if that gets used, another on top of that, and after enough layers, you might not even notice the sand you built on. After all, when you’re changing the world, who cares if anything actually works?

How is Sharknado vs. Polar Bear Vortex not yet a thing? It’s a 1000% awful idea! You know darned well we would all tune in with huge bowls of popcorn to watch the insanity of Bruce Campbell using a bazooka that shoots tornados of hyperevolved laser sharks to attack a swirling vortex of flying mutant polar bears. Come on, Syfy, the script writes itself!

The first person to create some kind of lint tool for Smile will be shot.

I understand the “desire” to have code be “uniform,” but linting tools, automatic-formatting tools, and other such ilk are an excellent example of not understanding the problem.

There is no “one correct formatting” for source code — in any language! — and style guides just make the problem worse by making you think there is such a thing. The goal of well-written source code is to communicate to other human beings — not the computer — the intent of the program. But intent is art, not science, and like all arts, it’s subjective. And la forme suit le fond, as the French say.

For comparison, the A.P. style guide, or the Chicago Manual of Style, tell you how you should write prose. They also disagree on really basic things. Should you use an Oxford comma, or not? Should you or shouldn’t you use contractions? Are sentence fragments ever acceptable? How long a sentence or paragraph is too long? Or too short?

e e cummings wrote his famous poetry — and his name — in all lowercase. Jesus wept is both the shortest and one of the most meaningful sentences in the Bible. Shakespeare’s most famous sonnet’s first four lines are loaded with what modern authors would consider spelling errors and a sentence fragment — “Shall I compare thee to a Summers day? / Thou art more louely and more temperate: / Rough windes to fhake the darling buds of Maie, / And Sommers leafe hath all to fhort a date…“

The point is that language changes, and language is mutable. You can freely alter it to serve your purposes. Writers always have, and always will.

The same is true in programming languages, especially in a language like Smile that is itself designed to be mutated. There’s no one correct way to write anything — in any programming language. There are only ways that are easier to read than other ways — and what’s easy and pleasant to read depends heavily on your experience and background. To many inexperienced programmers, the formatting below is confusing and heresy; to more experienced programmers, this code reads like art:

switch (field.Kind) {
    case FieldKind_Primary:    flags |= INITIALIZED | BOLD;    break;
    case FieldKind_Secondary:  flags |= INITIALIZED;           break;
    case FieldKind_Disabled:   flags |= INITIALIZED | GRAYED;  break;
    case FieldKind_Password:   flags |= HIDDEN_TEXT;           break;
    case FieldKind_Hidden:                                     break;
    default:                   throw new InvalidFieldError(field.Name);
}

I could show you what that looks like after it has been through a code-formatter that would put those switch cases on three lines each, with fixed indentation, but I’m not a big fan of ruining beautiful things. Which is exactly what a linter or code-formatter would do to it.

In literature, as in any of the arts, there are always exceptions to every rule. There are reasons not just to break every rule but to beat it to a brutal, bloody death with heavy bats and spiked shoes. Of course, to break the rules properly, you must first know them, and know when to apply them, and when not to. But Picasso didn’t become famous by painting still-lifes of fruit. Monet didn’t become famous through photographic precision of people’s faces. O’Keefe didn’t become famous by painting romantic landscapes. The best artists know the rules, and then break them, establishing new rules that are then broken by the next generation.

Well-designed source code in the 1960s left extra un-punched holes on the punch card between words and numbers to make it easier to fix errors. Well-designed source code in the 1970s put broken while statements inside switch cases because of how much more efficiently it ran. Well-designed source code in the 1980s made sure to write char *x, not char* x, because of how dangerous the “more obvious” form could be. Fashions change, as requirements change, and as preferences change. Today’s “perfect formatting” is tomorrow’s “ugly punch cards.”

So stop with the linters. Stop with the code formatters. Stop making prescriptive software that exists just to enforce your opinion. Because the best code is art, and the best art could care less about your rules.

Since there’s no longer Twitter in my life, you get to hear my tiny rants here instead.

The most useful script you’ll ever have when working with NodeJS:

rm -rf node_modules
npm cache clear --force
rm -f package-lock.json
echo crossing fingers...
npm install

Save this as npm-is-effed-again.sh. Run as often as needed. (And it’s often needed.)

Here’s my thread explaining why I left, and why I just might not bother ever going back:

Something that I haven’t really explained well is why I believe Smile code is shorter, better, and simpler than nearly everything else out there.  I’ve talked about it at a high level, and I’ve shown “Hello, World” and simple test programs, which are nice, but none of that really gives you the feel of coding in Smile.

So here’s an example of some unit tests written using a simple unit-test suite that I’ll be including out-of-the-box with the Smile distro.  Writing unit tests is a little closer to the day-to-day activities of Real Programmers, and it helps you to see a little better what code in a language feels like:

#include "testing"

tests for the arithmetic unit {

    it should add small numbers {
        x = 5 + 7
        assert x == 12
    }

    check the boundary conditions {
        x = 5 + 7
        assert x != 0 (adding positive numbers should never produce zero)
    }
}

Continue reading →