Erik Dominikus

I am not who I thought I was.

[Erik Dominikus @ErikDominikus]

@andywingo Hi Andy, just a quick and friendly yes-no question: Does debbugs (or anything) send you any kind of notification when someone sends an email to [email protected]? Thanks. :)

[Erik Dominikus @ErikDominikus]

@williamyaoh Stories and feelings have a better chance on changing people than arguments and reasons do. Do not make them think; make them feel. Share how your feelings change as you get more familiar with Haskell. Humans are not as rational as they believe they are. :)

[Erik Dominikus @ErikDominikus]

@lexi_lambda If you merely want to memorize, then the work is a waste. But if you want to understand, to know, to master, then perhaps the work is not a waste, but a necessity. Or, to avoid rediscovery, one can research off the beaten path, but not everyone likes the loneliness. :)

[Erik Dominikus @ErikDominikus]

Programmers should specialize into language creators and language users, not into front-end and back-end.

[Erik Dominikus @ErikDominikus]

@buggymcbugfix @Iceland_jack Thus, what you usually call a "function" is actually a name that refers to a function (not the function itself), and what you usually call an "anonymous function" is actually a function (the function itself). E.g. "do_something" _is_ not a function; it _refers_ to a function.

[Erik Dominikus @ErikDominikus]

@buggymcbugfix @Iceland_jack Recall that the definition of a function is a triple of domain-codomain-pairing; a function does not have a name. We pick a name to refer a function. The name is not a property of the function.

[Erik Dominikus @ErikDominikus]

@smdiehl (For future people: I was joking.)

[Erik Dominikus @ErikDominikus]

@smdiehl Haskell's method of standardization: Develop GHC like crazy so that alternative implementations would take too much effort.

[Erik Dominikus @ErikDominikus]

@smdiehl I'm afraid that if we want both capitalism and privacy, then we may need a business that sells temporary/fake identities.

[Erik Dominikus @ErikDominikus]

@smdiehl This seems true for all languages. If one's intention is unclear, its expression will be unclear in all languages. On the other hand, sometimes clear intention is not clearly expressible due to language limitations.

[Erik Dominikus @ErikDominikus]

@deeptech_id Karena mereka mengembangkan pikiran. 🙂

[Erik Dominikus @ErikDominikus]

@deeptech_id Pada tahun 2019: Racket, Haskell, Prolog, TypeScript.

[Erik Dominikus @ErikDominikus]

@slickbrick__ @jack It distributes power. It makes oppression impractical. It is an example of using the laws of nature to defeat human-made laws. But unfortunately it is too slow and uses too much energy.

[Erik Dominikus @ErikDominikus]

@FSE_UMD Does collective bargaining require involving the government? Why not direct action? Can't people unite without unionizing, and gain power without principal-agent problem?

[Erik Dominikus @ErikDominikus]

@Iceland_jack I think the problem is finding people who care about the same thing as we do, not finding a place to brainstorm. We can brainstorm via email or chat just fine. The problem is finding who that email should be sent to.

[Erik Dominikus @ErikDominikus]

@jashkenas Hawthorning https://t.co/ta0bX4X4eh

[Erik Dominikus @ErikDominikus]

@mekajfire Sense and reference?

[Erik Dominikus @ErikDominikus]

@nomeata * -> * may be easier to explain if we think of Haskell as two lambda calculuses (LCs): the value-level LC and the type-level LC describing the value-level LC, with the caveat that the type-level LC is a very limited LC.

[Erik Dominikus @ErikDominikus]

@nomeata I think of a type = a set of values. A functor = a way of uniformly extending all such sets. A monad = an idempotent functor (F x = F (F x) = F^n x for all n >= 1). Unfortunately, that is true in the category of sets, but not in Haskell.

[Erik Dominikus @ErikDominikus]

@nomeata I think of a * -> * as a type-level function in the same way a t -> t is a value-level function. I think of Haskell's type system as lambda calculus that is "crippled" (which is the price to pay for type inference).