🤣🤣🤣🤣💚💚💚💚🛤️🛤️🛤️🛤️

Oh yeah absolutely that's what I usually do but I found the perspective interesting 😁

💯

💯

Ah sorry, well NIF is faster so ports are slower usually 😁

Hating on the one we went to together... I see you.

I set it to `warning` in `test.exs` usually.

But on top of that iirc at Remote I implemented a notify system for output during test running, similar to flaky tests. I wouldn't fail the build but notify a channel of volunteers that triaged these.

Got as to 0 in time.

😁

Thanks for sharing! 💚

Haha thank you I really should 😅💚

That said most of these lookups I've seen eventually want to have an inverse lookup - which map gives you easily thanks to introspection.

That said, my main argument is the simplicity - it does less but it does the job. Seeing it I know it's "just" a lookup.

That's what I mean - the added flexibility there is YAGNI to me. Because you can do these things, but you don't need to. I have to read the entire case statement to make sure no magic is going on.

If it's a map I know no magic is going on.

That's why it should be the default.

🙈🙈🙈🙈🙈🙈🙈🙈🙈

I guess I could do a new blog and migrate stuff later 😅🙈

😅

The essence is:
* Maps are simpler, a case may do pattern matches and other stuff while a map clearly communicates that this is a key -> value lookup
* Maps offer introspection, I can get all keys, all values, ask if a key is there etc
* I can generate everything else based on the map

stuck in "need to migrate my blog off of wordpress" 😭

This would be an example in a blog post where I'd argue for "the simplest construct that gets the job done" but we'll see 😬

Y'all are tempting me to write the blog post calling out using data structures for data structure stuff 😅

This should be a map 👀

Amount of tests depends on surroundings, Impact, context etc.

I can see arguments for 0, 2 and "iterate all cases".