When you make a golden master on Floppies or CDROM, you have to test very thoroughly, and do the production in batches for fear of a bug being discovered. I went through 14 releases over 5 years, so there weren't a lot of bugs. The thing i had going for me was using Modula-2 which is like C with all the runtime and strong typing checks. For some reason Modula-2 didn't catch on like Pascal its 10 year predecessor, and i was one of only 100 people or so in the USA who bothered to use it. Exactly matches C feature-wise, but incredibly code bases are only about half the length because of the superior code sharing brought about by the clever module system. the ES 2016 JS module system is a bastard child of the Modula-2 system; they bungled a whole bunch of details.

I wrote my own module system a handful of years ago, and in the time since it's been the single biggest multiplier on my productivity. Along the same lines, I feel the

ns

macro in Clojure is a secret weapon. And yeah, I'm not fond of that JS module system. Thankfully, I've never needed to use it. What became of the Modula family?

ivan do you host that module system on github somewhere? I would love to learn more about the philosophy

https://github.com/cdig/take-and-make It's designed assuming you'll use it with CoffeeScript, but it should work fine with any compiles-to-JS (though it might not be quite as useful). The docs explain how to use it, but not necessarily why it's nice (and thus offers a productivity win, for me). If you're curious about that, shoot me a DM, since it's pretty weedy.

great! I’ll check it out, thanks!

Prof. Wirth wrote Pascal, a huge success and the UCSD pascal system was a contender for the IBM PC. However MS got the contract, and that finished that off. 10 years later was Modula-2, then later was Oberon. Oberon was used to make a whole OS from bare hardware, an incredible achievement for a team of only two dozen people. They sold hardware for a while, called the Lilith machine. It was inspired by Wirth's sabbatical at Xerox PARC; his version of the Alto machine. Anyway Oberon had a design flaw where Wirth had removed enumerated types which are omnipresent in Modula-2, and the thousands of people who had built up huge code bases around the world couldn't easily migrate their code, because when you remove a key feature from a language, it means a massive refactoring effort for minimal benefits. So he derailed his user base by not creating a bridge from his prior work. Academics live in their own world, and they get a fresh crop of graduate students on which to experiment every year, but commercial companies build a code base and sell it for decades, and depend on continuity, and any major change has to have a clean migration path or it doesn't happen. Which is why COBOL is still used in places, because there was no perceived easy way out short of a complete rewrite. The whole love of Linux to me seems ill-founded; the Oberon system with dynamic linking and the fantastic simplicity is less than 1/10th the complexity of Linux. What really bothers me about the end of the Lilith/Oberon project is that i haven't seen a practical tool that i wanted to use come out of academia since. Some brutal disconnection process has occurred between Academics and industry. I have no idea why, but i can sure see the results.

I'm surprised you're also a fan of dynamic linking, what with the vitriol you heap on OO 😉. But that goes to show that these ideas like OO and FP are, indeed, more granular than the typical zealotry allows. Thanks for sharing these anecdotes!

What's bungled in the JS module system, in your view?