Clojure Obscura | Hacker News Reader

Clojure Obscura

blog.fogus.me

57 pointsllambda11 comments

+1 great list Michael. I liked the contracts/Trammel link. I have never really got into contracts but this looks lite enough on ceremony that I can visualize using it. A long time ago I was very enthusiastic about the Little Schemer and Little Reasoner books, and sometime I need to make the plunge into core.logic.

I wish that I could (mostly) "go all in" with Clojure but there are also so many things I enjoy about Ruby. These two languages compete for my dev attention, and sometimes it is a brutal fight :-) Almost like there is too much good/useful/fun tech (cornucopia complex: http://blog.markwatson.com/2005/11/cornucopia-complex-ruby-o...)

Good to see the compilation target list is growing. The Clojure->Scheme->C target has previously been discussed on HN, sad to see it isn't being actively worked on now. The Clojure->C target is new to me though, and is active. The documentation is a bit light and I can't gleam how to use it from the tests. Has anyone here played with it and know what's going on?

https://github.com/schani/clojurec

It's extremely new, see discussion: https://groups.google.com/forum/?fromgroups#!topic/clojure/3...

Wow, his personal goals are exactly the type of thing I've been wanting. Will definitely follow this project.

Thanks for this: very interested to hear how live an issue it was that many more platforms supported C than LLVM.

What is it about Clojure that makes implementation towards other targets so appealing / approachable to devs?

Clojure was specifically designed to have multiple target implementations. That's the why it's approachable. As for why it's desirable, clojure is a high-level language and as such has certain niceties that are difficult to give up when forced to drop down to a lower level.

At least some of the recent Clojure implementations are based on ClojureScript's compiler, which from what I understand is relatively moduler: http://blog.fogus.me/2012/03/27/compiling-clojure-to-javascr...

Lisps in general are appealing to implement. Clojure's core language is about as small as Scheme's, and there are a lot of Scheme implementations too. The parts of Clojure that set it apart from other Lisps (STM, persistent data structures) aren't always ported. I don't think any of the recent approaches to targeting C (Clj->Scheme->C, Clj->C) include STM or persistent data structures. I'd be happy to be pleasantly surprised though.

persistent data structures are implemented in ClojureScript. So they don't need to be ported for new compilation targets of ClojureScript like C in clojurec, I think.

ClojureScript's core library has a dependency on Google Closure, and are optimized with knowledge of javascript vms. david nolen said that v8, for example, is blazingly fast at sequential instructions so they take great care to avoid function calls in critical paths.

here's some discussion that alludes to performance concerns in js vms: http://dev.clojure.org/jira/browse/CLJS-284

here's the original commit for PersistentVector, note how low level the code is: https://github.com/clojure/clojurescript/commit/e615f4cd326e...

here's more discusson by david nolen: https://groups.google.com/forum/?fromgroups#!searchin/clojur...

more discussion about allowing low level code access to js-specific special forms for numerics: https://groups.google.com/forum/?fromgroups#!searchin/clojur...

its clean and simple. the ClojureScript compiler is 2000 lines of clojure.

Reply on news.ycombinator.com