Schism – A self-hosting Scheme to WebAssembly compiler

by tosh • 8 years ago

194 36 8 years ago

Is there any chance Google might put more effort into this in the future? Scheme in the browser would be awesome. I'm guessing it is just a Google employee's pet project?

DannyBee• 8 years ago

This is just a random googler side project. That's what the disclaimer means. Any effort being put into it is by that person :)

throwaway76452• 8 years ago

Sadly that's my assumption too.

DogestFogey• 8 years ago

Scheme in the browser is awesome. Check out Biwascheme[1] and some of the games people have made. It's also on repl.it[2]

[1] http://www.biwascheme.org/ [2] https://repl.it/languages/scheme

sdiupIGPWEfh• 8 years ago

Been waiting for someone to do this. Glad to see there's still plenty of room for contributions.

shaunxcode• 8 years ago

Yes! Beneath the paving stones - the beach!!

alexozer• 8 years ago

Under 1000 lines, too?

kuwze• 8 years ago

I love the fact that they chose s-expressions for the textual representation of WASM[0][1].

[0] https://developer.mozilla.org/en-US/docs/WebAssembly/Underst...

[1] https://github.com/WebAssembly/design/blob/master/TextFormat...

bcherny• 8 years ago

Could you explain for the uninitiated?

icebraining• 8 years ago

WASM supports two formats; a binary format, which is the recommended for actually shipping to the browser, and a text format, mostly for humans to read and edit programs.

The text format uses Lisp-style syntax, with parenthesis, e.g.:

    (func (param i32) (param i32) (result f64) ... )

This syntax is called S-expressions, and it has strong adherents and opponents.

kodablah• 8 years ago

There are some shortcuts and other possible ways to represent the same code/items using, as your link mentions, the linear representation and import/export shortcuts. While not the canonical text spec, I find the grammar at https://github.com/WebAssembly/spec/tree/master/interpreter#... The clearest to understand.

ThatGeoGuy3• 8 years ago

First off: Neat project, I think this could actually turn into something super cool, especially if it could be transpiled into from another source. Think like CHICKEN, except instead of Scheme -> C -> binary, we could get $LANG -> Schism -> WASM (here, my $LANG of choice could even be CHICKEN)

However, I find the choice of R6RS... odd. Is it not easier to target R7RS-small or R5RS as a baseline to begin with? I know some people have problems (or too much love) with R6RS for what may be described as religious reasons, but I find it an extremely surprising choice here given that R7RS-small is much simpler and easier to get right.

Likewise, the majority of R7RS-compatible libraries would be available for free. That would add a lot for a high level language targeting WASM. Seems a strange choice, maybe the author has documented this somewhere?

eholk3• 8 years ago

I haven't documented the choice of R6RS anywhere, so here is as good of place as any.

The main reason was familiarity. R6RS is the Scheme I know best. I learned on R5RS and moved on the R6 when it was the new hotness. I vaguely was aware of R7 (I have friends that were on the standards committee), but didn't really know much more than that it had happened.

There's a technical reason for choosing R6RS as well though. To me the defining characteristic of R6RS is the library system. In my mind, this has a pretty clear mapping to WebAssembly modules. Scheme libraries import some set of identifiers and export another set. WebAssembly modules are the same way, so I basically set it up so that Scheme exports and imports map directly only WebAssembly exports and imports.

I'd definitely be open to supporting other Scheme versions in the future.

e12e• 8 years ago

I wonder how writing a wasm backend for larceny would compare to writing a new scheme.

http://www.larcenists.org/

ThatGeoGuy3• 8 years ago

There's a technical reason for choosing R6RS as well though. To me the defining characteristic of R6RS is the library system. In my mind, this has a pretty clear mapping to WebAssembly modules.

Well, this is a clear win for R6RS over R5RS, but R7RS has a much more flexible and well-agreed upon module system. One of the things about R6RS against its favour is that the module system never caught on with the major Scheme implementations (Gauche, Gambit, CHICKEN, Racket), and everyone kind of rolled their own. It's only in recent times that I've notice a lot more activity converging around R7RS.

Nonetheless, thanks for responding. Familiarity is a reasonable enough reason for choosing a standard.

na852• 8 years ago

Where can a person who only dabbles in Lisp read about the meaningful differences between R5/6/7 Schemes?

ThatGeoGuy3• 8 years ago

Your best bet is likely http://www.scheme-reports.org/

This is the main page that the RnRS committee uses to post publications.

VikingCoder2• 8 years ago

...why didn't they use github hosting, so we can actually see it running?

mirekrusin2• 8 years ago

Why didn’t you post pull request with it yet so we can all see?

VikingCoder2• 8 years ago

Configuring your github repro so that it CAN be hosted is not something you can change through a pull request.

eholk3• 8 years ago

Sorry I didn't set this up initially, but here's a super barebones playground: https://google.github.io/schism/playground.html

Keep in mind there's barely any error checking, so it may or may not work for you. I hope to improve this over time, and am happy to accept patches to do that as well.

bjoli• 8 years ago

I can't wait to play with this when I get home! I loathe doing web development. I just can't wrap my head around it, so whenever WASM gets in a position where it can easily work with the DOM (preferrably with a decent GC), I will be all over it :)

The project is in an interesting phase as well. Doing CPS transforms and implementing macros (the only thing really hard about writing a scheme).

pjmlp• 8 years ago

It looks quite cool.

For awhile I had the idea of applying "An Incremental Approach to Compiler Construction" to WASM, but never followed up on that.

Congratulations.

igorkraw• 8 years ago

First thought I had: We finally have scheme running in the browser: https://brendaneich.com/2008/04/popularity/

Then the comments pointed me to this

http://www.biwascheme.org/

ah well

tomcam3• 8 years ago

Somewhat unrelated question: would WebAssembly make a good intermediate language? It looks more machine-independent than I expected.

davidgrenier• 8 years ago

I'm confused, isn't that what WebAssembly has specifically been designed for? Unless you mean something else by "intermediate language" than what I read as "intermediate representation".

radarsat1• 8 years ago

I think he means, not just for web stuff. Which, of course, exists: https://github.com/AndrewScheidecker/WAVM

tomcam3• 8 years ago

Thank you for the correction. I did indeed mean “intermediate representation“. I had thought WebAssembly was x86-specific

tarcyanm• 8 years ago

I wonder if they're aware of Ian Piumarta's Maru and Cola (http://www.piumarta.com/software/) - which are also tiny and self-compiling (although use C and IA32 as a basis, not JS).

eholk3• 8 years ago

I hadn't seen those before. Thanks for the link, they look pretty interesting!

qwerty456127• 8 years ago

Cool! BTW, there is so much LISP on HN, I just wish it could actually get some momentum in the real world of production programming...

lilactown• 8 years ago

I just got my whole team to switch to Clojure(Script), so... ;)

na852• 8 years ago

HN seems to be mostly javascript web developers with a minority of all other tech fields sprinkled on top.

Schism – A self-hosting Scheme to WebAssembly compiler

Related Stories