Gloat is a Glojure AOT automation tool.
I worked with James Hamlin to get Glojure AOT going last summer and have been moving it forward since. I've also been working with marcingas (nooga) to get Gloat/Glojure/let-go all cooperating.
As far as JVM-free Clojure-like, Janet is really nice. I've been using it in production for a while: https://janet-lang.org/ There's also Fennel if you want the Lua vm and libraries.
While Janet pulls from a few inspirations, the syntax is pure Clojure. I always figured that it was trying fix up the bumpy parts in Fennel to enable a programming style that was more consistently Clojure-like and functional than could be done in Fennel, since Fennel ultimately has to use Lua's semantics because Fennel compiles to Lua.
It's a bit harder as Fennel can produce pure functions, whereas basically everything in Janet's standard library outputs mutable structures etc. Janet style tends towards (elegant) imperative programming.
This is the kind of clojure port that I always was looking for. Mostly because I thought go's core library and channels abstractions hits a simpler/nicer base API which would with the core & async apis (not to mention scratches my big beautiful binary itch)
Thanks for your work will definitely check it out again once I get over renewed love for cpp (26)
Edit how did glojure go under my radar also a great project from the looks
I have played with the idea of making a “old school PHP” style DSL that takes advantage of the Go runtime and packages under the cover. I say old school PHP because PHP used to be a web focused DSL its no longer the case, I feel like it would make for an interesting easy to use backend language similar to PHP but with the full power of Go behind it. Clojure is an excellent choice.
There seems to be a surge in compile to Go projects recently. To me this signals that the runtime / stdlib of Go is one of the best out there (when going the GD'd route), but the surface level (syntax) is too simple/verbose and lacks the expressiveness developers want.
So far Lisette (http://lisette.run) seems to be the best/most active version of a compile to Go language out there.
Cool, I think at this point, since there's a few Clojure Go dialect, what I'd like to understand most is which one is fully hosted on Go, has as good interop as JVM Clojure has with Java, where I can both use what I make from Go, and also use Go from Clojure, and even create mixed projects would be nice. And also, truly details the differences, beyond interrop, what is not the same and differs.
I think Gloat/Glojure has the best hosted runtime story because of its AOT to Go src pipeline - you can grab anything Go at compile time. OTOH let-go can roundtrip any Go value including structs, functions and channels but it does not allow pulling arbitrary go libs without wrapping them up first - they'd need to be built into the runtime for this.
I started publishing amd64 and arm (32) binaries for Plan 9 in GH releases. I tested the amd64 one on 9front and everything seems to work. The CLI is not very Plan9-y but I'm happy to make the port more native at some point :)
Sure, I should clarify: The 7ms here is measured at the point where let-go starts executing user code. It takes 7ms to initialize the compiler, load all stdlib namespaces and compute all vars. So it's not "time to first instruction", it's "time to running your code".
I think I've pulled the latest today but will double-check and update them again tomorrow. I'm still puzzled why it doesn't run the tak function. Btw. Have you tried running my benches? I'm very curious about your results!
I'm relieved that clogo is not an option for you. It has some strong, malodorous campground toilet connotations to my foreign-language-trained ears. Loogo? Pottygo? Go number 2? Yikes to all of those.
I think you could make a framework on top of this. It doesn't yet run unmodified Clojure libs like hiccup but it wouldn't be hard to roll something relatively simple and solid in let-go. IMO
The JVM itself - yes, but then you need to wait for Clojure itself. Clojure is famously slow because it interns a bunch of stuff and computes var references every time it boots. I never intended for let-go to be a race car but small distro, low memory usage and snappy start are its priorities at the moment.
jvm clojure's startup tax is exactly why it never cracked the serverless/cli niche, and this pure-go implementation is the kind of unhinged side project, good luck
Thanks! I've been looking into custom runtimes for AWS Lambda and I think it's not going to be hard to have let-go lambdas. Even though the language is not 100% compatible it should be enough for request handlers and data processing. Throughput isn't on par with JITed impls but it's not bad enough to be impractical IMO
absolutely sick of reading through obviously AI-slopped READMEs. it's your project, take a little pride and tell me why i should like it quickly instead of asking your agent to rattle off a list of features -- it's severely boring & offputting.
apologies if i was blunt - readme sloppage is a particular annoyance of mine that is quickly becoming common. i'm not against vibecoding, far from it. but a readme is a part of a project that humans immediately touch - seeing it littered with em-dashes signals carelessness.
Job offer I couldn’t refuse that didn’t have Clojure.
Now I work for a fully remote team, can work anywhere in the world, at any moment I want, leading the data / cloud team for a distributed timeseries database.
Can’t complain. :)
Clojure has had a huge, fundamental impact on my way of approaching software development. I actually came from a Haskell / C++ background, but the way Clojure treats data still has a fundamental impact on how I reason about data, architecture and simplicity.
I did have some issues with how Clojure is managed and do not always subscribe to Rich’s vision (I think core.spec makes no sense, a heavily macro based global state registry is fundamentally not how I would design this, and malli is infinitely better. same for core.async vs manifold), but that is a minor detail in what was a transformative experience for me.
I believe I am not alone when I say this.
I’m still following things from a distance. Considering the current thread, I’m actually very interested in yank, which is Clojure on LLVM, and have been sponsoring that project for a few years. That would be very nice if it could enter stable state, I may take another look again.
> I did have some issues with how Clojure is managed
Yes there was some drama a few years back and then Rich wrote his post 'Opensource is not about you'. It was a good post.
Opensource is not easy and you might argue the reason why Clojure is so stable and backwards compatible is because of the way it's managed.
Luckily we didn't end up with a scenario where Rich completely stopped. I think there was a recent case of an opensource maintainer (who works in academia) stopping PRs due to an entitled user. Can't remember the project.
But equally, is the current form of stewardship fit for purpose for the next 10 years of Clojure, i.e. to increase adoption by businesses? Don't know. Maybe something can be learned from how Linux is managed. I think Linus experienced similar bottleneck issues back then.
Macros with syntax quote,
Reader conditionals,
Destructuring,
Multi-arity functions,
Atoms, channels & go-blocks a'la core.async,
Regular expressions (the Go flavor),
Simple json, http and os namespaces,
Many functions ported from clojure.core,
REPL with syntax-highlighting and completions,
Simple nREPL server that seems to work with BetterThanTomorrow/calva,
The readme clearly has abundant emdashes and emojis everywhere, the code itself is obviously vibed. Not really sure what you're objecting to, to be honest.
Yeah, it has em dashes alright. But the emojis you're referring to were committed in January 2023 (0c4925c). But that's besides the point I guess. What is your point?
Try out this Wasm browser REPL https://gloathub.org/repl/
Gloat is a Glojure AOT automation tool. I worked with James Hamlin to get Glojure AOT going last summer and have been moving it forward since. I've also been working with marcingas (nooga) to get Gloat/Glojure/let-go all cooperating.
Nice! I recently played around with a Lisp syntax for Go semantics: https://codeberg.org/veqq/Joe
As far as JVM-free Clojure-like, Janet is really nice. I've been using it in production for a while: https://janet-lang.org/ There's also Fennel if you want the Lua vm and libraries.
Thanks! Joe looks good! As for Janet - never tried it myself but I always thought it's doing its own thing instead of trying to be Clojure.
While Janet pulls from a few inspirations, the syntax is pure Clojure. I always figured that it was trying fix up the bumpy parts in Fennel to enable a programming style that was more consistently Clojure-like and functional than could be done in Fennel, since Fennel ultimately has to use Lua's semantics because Fennel compiles to Lua.
It's a bit harder as Fennel can produce pure functions, whereas basically everything in Janet's standard library outputs mutable structures etc. Janet style tends towards (elegant) imperative programming.
For what types of work do you Janet in production?
This is the kind of clojure port that I always was looking for. Mostly because I thought go's core library and channels abstractions hits a simpler/nicer base API which would with the core & async apis (not to mention scratches my big beautiful binary itch)
Thanks for your work will definitely check it out again once I get over renewed love for cpp (26)
Edit how did glojure go under my radar also a great project from the looks
I have played with the idea of making a “old school PHP” style DSL that takes advantage of the Go runtime and packages under the cover. I say old school PHP because PHP used to be a web focused DSL its no longer the case, I feel like it would make for an interesting easy to use backend language similar to PHP but with the full power of Go behind it. Clojure is an excellent choice.
Thanks for kind words! Please don't forget to drop me an issue or two when you eventually get to it :)
Micro nit: it says 7ms cold start and then 6ms just few lines lower.. maybe it gets faster as you read README
Fixed, thanks! It's 6-7ms on my machine. Median seems to be around 6.5ms :)
There seems to be a surge in compile to Go projects recently. To me this signals that the runtime / stdlib of Go is one of the best out there (when going the GD'd route), but the surface level (syntax) is too simple/verbose and lacks the expressiveness developers want.
So far Lisette (http://lisette.run) seems to be the best/most active version of a compile to Go language out there.
That does look pretty good. I don’t love the mixing of PascalCase / camelCase and snake_case, though.
Thats just a pragmatic design decision for Go interop, as Go went nuts with public being UpperCase.
Theres no camelCase tho? Snake case and PascalCase fir Go modules interrop.
I'll be happy is Lisette succeeds, but I like Soppo (https://github.com/halcyonnouveau/soppo) for staying closer to Go and not introducing this mismatch.
“let data = io.ReadAll(file)?”
That’s Pascal case any way. Looks sort of camel case with the “io” being lower case.
Yes. PascalCase for Go, snake_case for Lisette. I find it quite good actually as i can see what hooks into the Go ecosystem instantly.
Ofc you can write your own wrapper, but so far i have not bothered.
An alternative is Joker: https://joker-lang.org
I think it is brilliant and completely underappreciated :)
What I appreciate about Joker is how smoothly it wraps Go libs. It seems that they have covered everything that Go has to offer.
I'm trying to avoid adding too much though, I like that let-go fits in 10MB :)
Cool, I think at this point, since there's a few Clojure Go dialect, what I'd like to understand most is which one is fully hosted on Go, has as good interop as JVM Clojure has with Java, where I can both use what I make from Go, and also use Go from Clojure, and even create mixed projects would be nice. And also, truly details the differences, beyond interrop, what is not the same and differs.
I think Gloat/Glojure has the best hosted runtime story because of its AOT to Go src pipeline - you can grab anything Go at compile time. OTOH let-go can roundtrip any Go value including structs, functions and channels but it does not allow pulling arbitrary go libs without wrapping them up first - they'd need to be built into the runtime for this.
Amazing that it can run in plan9!
It would be awesome if go was a first class language on 9front. I.e. ships with it.
I've been messing around with a social network for plan9:
https://youtube.com/watch?v=q6qVnlCjcAI&si=MBCeM0QdA0WsKAe7
It's all in rc and awk. There are places where go (or clojure!) would have been nice.
Love the demo for 9social! I'm not a user myself but I'm always inspired by the Plan 9 ways :)
Thanks for checking it out!
I started publishing amd64 and arm (32) binaries for Plan 9 in GH releases. I tested the amd64 one on 9front and everything seems to work. The CLI is not very Plan9-y but I'm happy to make the port more native at some point :)
Ah, awesome!
I'll head over to the releases.
You should see how fast libc gets mmaped() into the VM and the first instr runs :)
Sure, I should clarify: The 7ms here is measured at the point where let-go starts executing user code. It takes 7ms to initialize the compiler, load all stdlib namespaces and compute all vars. So it's not "time to first instruction", it's "time to running your code".
PR most welcome for https://github.com/chr15m/awesome-clojure-likes
Added https://github.com/chr15m/awesome-clojure-likes/pull/27 Thanks for letting me know :)
do you know about Glojure?
https://github.com/glojurelang/glojure
Yes, I know about this one. I'm even comparing against it in my benchmarks :)
You need to update the go-joker numbers, I removed the GIL yesterday or so and did some changes to the IR. ;)
I think I've pulled the latest today but will double-check and update them again tomorrow. I'm still puzzled why it doesn't run the tak function. Btw. Have you tried running my benches? I'm very curious about your results!
I got tak fixed. There was an internal logic error, IIRC. A bunch of stuff got done.
Updated. Looks like my VM gets a beating from wazero but that's kind of expected :D
https://github.com/gloathub/glojure is the actively maintained fork.
I wish this had a better language name than just "lets-go". How about "clogo" ?
Naming is hard but I like let-go, it's a layered pun :)
(let [...] (go ...))
It lets you let in Go!
Clogo is a no-go for me. Maybe you have some other ideas? I'll consider good ones :)
FWIW I like let-go a lot as the name.
I'm relieved that clogo is not an option for you. It has some strong, malodorous campground toilet connotations to my foreign-language-trained ears. Loogo? Pottygo? Go number 2? Yikes to all of those.
And now, as bonus content, for something completely different: https://youtu.be/M3-51DhOzHE?
I laughed, audibly :D
Gojure is simplest. Just a bit close to Glojure which is already taken.
Goclo, Gophoj,Gorjure ..ok my tongue is paining now verbally trying out combos.
Actually, look what I just found https://github.com/tcard/gojure 12 years!
Changing name to Gojure now would only add to the confusion :D
I am finding i need "Rails" but i like single binary deployment of Go and fast/low resource usage like Go.
Is it possible for now?
I think you could make a framework on top of this. It doesn't yet run unmodified Clojure libs like hiccup but it wouldn't be hard to roll something relatively simple and solid in let-go. IMO
This is beautiful, makes me wish I'd made it.
Excellent work, thank you for sharing it with us ^_^
Thank you! If you ever want to contribute PRs and issues are warmly welcomed :)
The JVM can boot up in microseconds you just have to tune it a bit.
The JVM itself - yes, but then you need to wait for Clojure itself. Clojure is famously slow because it interns a bunch of stuff and computes var references every time it boots. I never intended for let-go to be a race car but small distro, low memory usage and snappy start are its priorities at the moment.
Joker, Janet, Glojure and now Let-go. Am I missing something?
There is one being made in C http://mino-lang.org/
jank is another one
Very cool project!
jvm clojure's startup tax is exactly why it never cracked the serverless/cli niche, and this pure-go implementation is the kind of unhinged side project, good luck
Thanks! I've been looking into custom runtimes for AWS Lambda and I think it's not going to be hard to have let-go lambdas. Even though the language is not 100% compatible it should be enough for request handlers and data processing. Throughput isn't on par with JITed impls but it's not bad enough to be impractical IMO
absolutely sick of reading through obviously AI-slopped READMEs. it's your project, take a little pride and tell me why i should like it quickly instead of asking your agent to rattle off a list of features -- it's severely boring & offputting.
Thanks for feedback. Here's a pre-AI-slopped README https://github.com/nooga/let-go/blob/98c2e2ebf38519bceb4f799...
You can also refer to the HN post itself - it says why I think it's cool.
This version is infinitely better.
apologies if i was blunt - readme sloppage is a particular annoyance of mine that is quickly becoming common. i'm not against vibecoding, far from it. but a readme is a part of a project that humans immediately touch - seeing it littered with em-dashes signals carelessness.
i appreciate you taking my feedback with grace.
I would like to point out, again, that em dashes are very much used by humans that run macOS or iOS — like in this case.
Also Linux, where it's easy to configure a compose key to mnemonically type all the Unicode goodies you can think of.
Android's software keyboards generally make it easy, too.
No worries at all. I understand your point. I'll look into fixing this!
Why did you feel the need to slopify your README? The original version read much, much better.
I genuinely don’t understand why people do this.
Good question, perhaps I really was just careless. I'll look into fixing the README.
It’s all good. Your project is awesome (and I say this as someone who has done Clojure fulltime for 5 years and nowadays write mostly Go).
What made you stop using Clojure? Lack of Clojure jobs? Or something else?
Job offer I couldn’t refuse that didn’t have Clojure.
Now I work for a fully remote team, can work anywhere in the world, at any moment I want, leading the data / cloud team for a distributed timeseries database.
Can’t complain. :)
Clojure has had a huge, fundamental impact on my way of approaching software development. I actually came from a Haskell / C++ background, but the way Clojure treats data still has a fundamental impact on how I reason about data, architecture and simplicity.
I did have some issues with how Clojure is managed and do not always subscribe to Rich’s vision (I think core.spec makes no sense, a heavily macro based global state registry is fundamentally not how I would design this, and malli is infinitely better. same for core.async vs manifold), but that is a minor detail in what was a transformative experience for me.
I believe I am not alone when I say this.
I’m still following things from a distance. Considering the current thread, I’m actually very interested in yank, which is Clojure on LLVM, and have been sponsoring that project for a few years. That would be very nice if it could enter stable state, I may take another look again.
Thanks for clarifying.
> I did have some issues with how Clojure is managed
Yes there was some drama a few years back and then Rich wrote his post 'Opensource is not about you'. It was a good post.
Opensource is not easy and you might argue the reason why Clojure is so stable and backwards compatible is because of the way it's managed.
Luckily we didn't end up with a scenario where Rich completely stopped. I think there was a recent case of an opensource maintainer (who works in academia) stopping PRs due to an entitled user. Can't remember the project.
But equally, is the current form of stewardship fit for purpose for the next 10 years of Clojure, i.e. to increase adoption by businesses? Don't know. Maybe something can be learned from how Linux is managed. I think Linus experienced similar bottleneck issues back then.
> I’m actually very interested in yank
I think you mean Jank: https://jank-lang.org/ ?
I'm quite excited about Clojure for GO projects.
EDIT: clarity.
I've never heard of manifold before and I'm curious what you prefer about it to core.async.
I've read through their website and README but I feel like I'm missing what separates it from other async libraries.
Can you explain a few examples where it shines compared to core.async?
I did. And it looks like you did too. Which is why you answered your own question in the second half of your comment, quite amusingly. "Probably" LOL!
So...you didn't ?
https://github.com/nooga/let-go/tree/d9dc094822b2983ebf44604...
In 2023 he had a working Clojure compiler with:
Macros with syntax quote, Reader conditionals, Destructuring, Multi-arity functions, Atoms, channels & go-blocks a'la core.async, Regular expressions (the Go flavor), Simple json, http and os namespaces, Many functions ported from clojure.core, REPL with syntax-highlighting and completions, Simple nREPL server that seems to work with BetterThanTomorrow/calva,
The readme clearly has abundant emdashes and emojis everywhere, the code itself is obviously vibed. Not really sure what you're objecting to, to be honest.
Yeah, it has em dashes alright. But the emojis you're referring to were committed in January 2023 (0c4925c). But that's besides the point I guess. What is your point?