With Jank, Jolt and Babashka FFI, Janet's use cases are shrinking. There are many near alternatives. While I'm not sure where the pros and cons are (how fast everyone's start up times are vs. larger program execution)(Jolt is straight vibe-coded, I think Jank accepts LLM code now)(how comfy and performant is everyone's CFFI?), there are less clear reasons to use Janet now. Fennel still has all the targets which embed Lua. I'm really curious how other people make the decision. I'm sort of sticking to it out of inertia, having fun experiments etc. but the experiments verge into Racket territory with many DSLs. I'm very curious where the dice will fall going forward.
Aaron Hsu left Scheme for APL, unable to fit the paradigms together. I feel Clojure-like syntax gets us much of the way there. As projects proliferate and speed up, what commonalities and synergies will they all find? Will we find a new Platonic ideal before the industry collapses?
Very weak data point: As someone who reads about this stuff far more than I actually have the time to use it…
I recognize all but Jolt
I recall Fennel being relatively modern
I recall Janet being relatively established
Jank seemed exciting but not like something I'd think to build on
I use projects like these for fun, so the less LLM use the better
Not exactly a Stack Exchange survey, but I've been surprised in the past by Internet strangers who have the same "soft associations" as I do—I'd always assumed they were highly individual to me.
Overall, I hope that when I need/want a tool like Janet (sometime in the next few years, optimistically), I look it up and find it's still around :)
IMO and IME Janet is squarely in the 'for fun' category. LLMs are kinda antithetical to it for most of the target audience. When one is not looking at or writing code, language stops being a distinguishing value and people just go for whatever heavyweight thing.
I'll personally never use Jolt. Jank always felt a bit more 'serious' because involving LLVM means we're out of lightweight and 'small enough to fit inside my head' land. Fennel is great, but it is 'just a compiler' and Lua's runtime semantics rule, for better and worse.
Janet still has value to me (although I've wrote it mfar less than I've read about it) as a human-made, fairly well-thought out language with solid fundamentals. Ecosystem is wobbly as expected, but I kinda keep expecting better tooling. Maybe I'll take another shot to see how things have improved.
veqq | 8 hours ago
Good job! This is really nice, but the universe is strange.
With Jank, Jolt and Babashka FFI, Janet's use cases are shrinking. There are many near alternatives. While I'm not sure where the pros and cons are (how fast everyone's start up times are vs. larger program execution)(Jolt is straight vibe-coded, I think Jank accepts LLM code now)(how comfy and performant is everyone's CFFI?), there are less clear reasons to use Janet now. Fennel still has all the targets which embed Lua. I'm really curious how other people make the decision. I'm sort of sticking to it out of inertia, having fun experiments etc. but the experiments verge into Racket territory with many DSLs. I'm very curious where the dice will fall going forward.
Aaron Hsu left Scheme for APL, unable to fit the paradigms together. I feel Clojure-like syntax gets us much of the way there. As projects proliferate and speed up, what commonalities and synergies will they all find? Will we find a new Platonic ideal before the industry collapses?
nil | 3 hours ago
Very weak data point: As someone who reads about this stuff far more than I actually have the time to use it…
Not exactly a Stack Exchange survey, but I've been surprised in the past by Internet strangers who have the same "soft associations" as I do—I'd always assumed they were highly individual to me.
Overall, I hope that when I need/want a tool like Janet (sometime in the next few years, optimistically), I look it up and find it's still around :)
e3bc54b2 | 2 hours ago
IMO and IME Janet is squarely in the 'for fun' category. LLMs are kinda antithetical to it for most of the target audience. When one is not looking at or writing code, language stops being a distinguishing value and people just go for whatever heavyweight thing.
I'll personally never use Jolt. Jank always felt a bit more 'serious' because involving LLVM means we're out of lightweight and 'small enough to fit inside my head' land. Fennel is great, but it is 'just a compiler' and Lua's runtime semantics rule, for better and worse.
Janet still has value to me (although I've wrote it mfar less than I've read about it) as a human-made, fairly well-thought out language with solid fundamentals. Ecosystem is wobbly as expected, but I kinda keep expecting better tooling. Maybe I'll take another shot to see how things have improved.