John Coene

@john-coene.com

Go, Javascript, Vapour, R, working on http://opifex.swiss

They believe LLMs are superhuman tools that generate correct and performant code that require no supervision or review so they decide to use those "superhuman" tools to generate... Python and JavaScript!

Fine #rstats built-ins NAs are great but I genuinely don't think they make any sense for booleans: it's in the name ffs, it's a boolean, it can be one of 2 values. And this doesn't make sense either. as.logical("1") $> NA as.logical("true") $> TRUE I've got TRUE, FALSE, and MAYBE???

Please enjoy this graph from a FT article explaining that AI hasn’t added any value (no, debt and circular investment isn’t value) It has produced More Stuff, it’s just the Stuff isn’t useable or fixable. Welcome to the age of everyone making apps no one uses

Bild

I actually forget how environments factor in but in the S origins (w/o lexical scope), functions were isomorphic to lists. The first n-1 elements was a named list corresponding to argument names and defaults, and the last element was the body expression. So ls[2] would be like function(pos=-1) NULL.

We made a fake repo with fake bounties, and the bots are applying fake PRs, so we know who is fake, and we can ban them from the Coolify repo. IQ over 1000

BildBild

Oh boy do I love S7! Treat new_class() as struct, props now have their types, and everything is a method 💯🎉🔥 I use it to write #rstats code like I write Go code; it's verbose, nothing like idiomatic R, it'll make your hair curl but I love it!