WebAssembly is a Instruction Set Architecture (ISA) for a stack machine that can run in most modern browsers. Therefore, WebAssembly allows the web designer to customize the experience of the client, who is using his or her browser to visit an Internet page. Besides this, WebAssembly is blazing fast, when compared to Javascript, the only other programming language that the World Wide Web Consortium recommended to run natively on Internet browsers.
One thing that makes WebAssembly exciting is the use of the Cambridge Prefix Syntax, the same that makes Lisp so interesting and powerful. Since Lisp is designed to handle this kind of syntax, I invite you to help me in the construction of a compiler, which translates Low Level Lisp into WebAssembly. By using Lisp as the implementation language, we can get the tokenizer and the parser for free.
Implementing Lisp onto WebAssembly is so obvious that you may wonder why somebody did not have this idea long ago. In fact, a team at Google is working on the implementation of Schism, a dialect of Lisp, on WebAssembly. The members of this team intend to outfit Schism with all powerful tools for list processing and reflection that made Lisp famous. This is not what the author of Low Level Lisp intends to do. The power of Lisp comes with a price: Lisp compilers are very difficult to implement. Compilers for Chez Scheme, Bigloo, Racket, Gambit and Sbcl required years of work from bright people, such as Manoel Serrano, Kent Dybvig, Marc Feeley and Matthew Flatt.
I am fully aware that it would be difficult and unnecessary to compete with the Google team that is constructing a WebAssembly scheme. It will be unnecessary because they will eventually succeed, and everybody will have another open source Scheme dialect to use and deploy. However, besides a high level language, such as Lisp or Prolog, one needs low level languages to develop algorithms for gaining a better understanding of the underlining computer and performing things such as memory manipulation.
WebAssembly uses the Cambridge Prefix Notation, as Lisp, but it does not accept those concise expressions, which make Lisp programmers so productive. The idea is to design a Low Level Lisp (LLL) compiler that accepts succinct expressions, as in Scheme, but implement only those operations that can be translated directly into WebAssembly.
Let us examine a concrete example. People often use a naive definition of the Fibonacci's function to perform benchmarks. In WebAssembly, such a definition is given below.
(func $fib (param $n i32) (result i32)
(if (result i32)
(i32.lt_s (get_local $n) (i32.const 2))
(then (i32.const 1))
(else (i32.add (call $fib (i32.sub (get_local $n)
(i32.const 1)))
(call $fib (i32.sub (get_local $n)
(i32.const 2)))) )))In Low Level Lisp, the same definition would not require tags on constants or even on local variables, therefore it could become compact and more to the point, as shown below.
(define (fib n)
[if (<fx n 2) 1
(+fx (fib (-fx n 2))
(fib (-fx n 1))) ])Although succinct, the Low Level Lisp definition
of naive Fibonacci has all the necessary elements
to reconstruct the wat code. For instance, since
the -fx function that performs fixed point
addition accepts only i32 numbers, the Low
Level Lisp compiler can infer that the arguments
of (-fx n 2) should be (get_local $n)
and (i32.const 2).
In a typical WebApplication, three languages would be at play. On the server side, a Bigloo program would take care of list processing and Artificial Intelligence. On the browser, WebAssembly generated by the LLL compiler would deal with text processing and numerical calculations, while Javascript could handle input output operations and insertion of the LLL generated html snippets into the document.
In the documentation of this project, the interested reader will find detailed descriptions of miniature web pages designed pursuant to the LLL philosophy. The documentation also provides a brief tutorial of Bigloo, one of the languages supported by Manoel Serrano's Hop web programming environment. Finally, there is the ongoing work on an LLL compiler.
The first link shows the greet script, described in chapter 9 of the documentation.
http://medicina.tips/big/greet.k?xnome=%E9%82%93%E5%B0%8F%E5%B9%B3
The wasmdemo application is a very simple demonstration of how to deploy a web page that loads an wasm program. This example is discussed on chapter 4 of the documentation.
http://medicina.tips/big/wasmdemo.html
This demo shows a naive factorial benchmark that was programmed in Low Level Lisp and compiled to wasm through the wasCm compiler.
http://medicina.tips/big/nfib.html
For security reasons, Firefox and other browsers do not let you load local files containing Javascript code. To enable JavaScript in local files follow the steps below:
- In Firefox, open up a new tab, and type about:config in the address bar, then press Enter
- Click the accept-the-risk button
- Type unique in the search box, then press Enter
- Switch the privacy.file_unique_origin preference to false
I could not find a way of opening a local file in Google Chrome. However, you can start a local http server inside the folder where your application resides. Then you can open the localhost page:
~/src/wasCm/wasm-samples$ http-server . -p8080
Starting up http-server, serving .
Available on:
http://127.0.0.1:8080
http://192.168.0.38:8080
Hit CTRL-C to stop the serverType http://localhost:8080 on the address field of your favorite browser, and you will be given a list of files that you can execute by mouse clicking on its entry.