13 Matching Annotations
  1. Jan 2022
    1. We can do that, using the @typedef JSDoc:

      By this point I might argue that just writing TS is less verbose ("cleaner", as the author mentions above) but I understand that preferences vary.

    2. Workable, but less than perfect.

      would've loved more detail on things like choice of IDE etc. Vscode has pretty decent TSC support. better than webstorm even, imo.

    3. In actuality, TypeScript is not one language. It’s an infinity of languages, each one slightly different, and each one “generated” by a specific configuration file.

      Really good point, and I really like this way of thinking about TS.

      I think this behaviour comes from the fact that TS is trying to bend itself to JS; there's infinite variability in the kinds of projects written in JS so of-course one would need a configuration space that's wide enough to accomodate that.

    4. everybody should move to dynamic typing

      I appreciate that later in the article the author realizes that they were falling into a black or white fallacy because this is another example of the same.

      clearly all software contexts are not the same. What's appropriate for an MVP is clearly different from what is needed for code deployed on a space-mission.

    5. my code was “clean”

      might be more accurate to say it was less verbose (compared to Java/C++).

      and having terser code is also important for sure, just that saying code is "clean" opens up scope for too much subjectivity imo

    6. This was because dynamic typing doesn’t give the safety net that static typing does. So you write more tests. Which is a good thing.

      Tests are a good thing, but not all devs are even aware of the loss of safety net and so there's no guarantee that using a dynamic language results in more tests.

      Using tests and static-types together would mean that there's whole categories of bugs that get eliminated while also getting more coverage for less lines of test-code.

    7. because you’re dealing with data, and not trying to abstract it away, then the design becomes simpler. Much easier to reason and deal with.

      this seems like more of a pro-FP argument than an anti-static-types argument.

      the destroy-all-software talk on "Boundaries" / "Functional core, reactive shell" illustrate this approach beautifully.


  2. May 2021
    1. but that we’re looking to science for answers that ultimately require human moral intervention

      science can provide the tools, but not the motivations to actual do something. For that we need philosophy, morals and ethics. Hume's Guillotine somewhat formalizes this distinction.

  3. Apr 2021
    1. There is such a thing as a price umbrella, but there's also one for customer happiness.

      I think this is meant to mean that Amazon sets the benchmark not only for price but also for customer-happiness. So if you're coming in as a new retailer, you need to either make an offering on the higher-end or the lower-end.

      and offering something on the lower-end of customer happiness is probably not a winning strategy.

    2. ex nihilo

      translates to 'out of nothing'

  4. Mar 2021