If it takes more than half an hour to do with CSS, you’re legally allowed to render a different tree.

Sad music corner :(

I’m no expert on either, just that they are so simple. Last expression in a file is the export, and it can be a function.

Sadly ESnext is full of short sighted decisions :(

And here I am thinking that utf-8 is ubiquitous. Apparently not!

No worries, me neither 😂 just that these structures need to obey laws, in case of monads they are left identity, right identity and associativity

I didn’t memorize them, looked up from here wiki.haskell.org/Monad_laws

ES modules were made to enable static analysis as opposed to whatever CSM was doing, but I think some happy medium could’ve been found.

Nix and Dhall are great.

Okay I read your post properly this time and you mention functors 🙏

I thought functors in Ocaml were parametric modules? And there was a backpack initiative in Haskell. Hope I’m not making these up :D

Just had one today, the contact form didn’t submit and had no indicator either, I deleted the offending character and it was all dandy!

It looks very cool but not sure how it’s a monad. Does it have flatMap (or similar) defined?

Non-branching code is easier to read and more predictable. Good example!

I don’t think it’s rude, you’re an open source maintainer and if I’m asking for tour time I should at least put the minimal effort.

Good rule of thumb! Similar to what I alluded to by "vibe check", for a piece of code that's not in a tight loop, how it expresses intent and blends in with the surrounding code is more important than raw performance.

This seems pretty similar in construction and lookup complexity, also a Map. In practice there could be differences due to JS engine optimization though.

If the array is constructed just to be searched too, I’d reach for set in that instance. But I guess there’s an internal vibe check, if the search space is small (<10 elements) I could still go for array since Set doesn’t have a nice constructor syntax.

Hm it kinda depends? JS sets are hashsets so constructing a set is O(n), which is same as a linear search. So if the set is constructed only to search, linear search would be better at n/2 avg time, assuming hashing is free and memory is infinite.

If set will be reused it’s a different story tho

But it takes effort and solves actual problems :(

Gotcha. So it seems like it was only an assumption that effect events can’t be called asynchronously

Is this screenshot the error, or the correction?