write-you-a-scheme-v2 / scheme

Write You a Scheme

Write You a Scheme

About this project

Shortcut build:

$ ./build 

for building with stack or

$ ./build-cabal

for building with cabal

Shortcut repl:

$ ./repl

or use ./repl-cabal

Shortcut run binary executable

$ ./run -r                 # Interactive shell
$ ./run -s test/let.scheme # Run file
$ ./run --help

or use run-cabal.

Building the reference implementation.

$ stack exec scheme

Testing the reference implementation. (see test-hs/Spec.hs)

$ stack test

Debugging the reference implementation in GHCI.

$ stack repl 

Building the HTML and MediaWiki files for the tutorial text.

$ stack exec docs

(or substitute cabal for stack in each case).

The Wikipedia files are generated to output/docs/*.wiki. To view the HTML output.

$ firefox output/scheme.html

Contributing

Please feel free to contribute any improvements, corrects or ideas you have! I would love to hear how you are using these ideas, or what you think about this project. Your help will make this project great!

License

This written work is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License. You may reproduce and edit this work with attribution for all non-commercial purposes.

The included source is released under the terms of the MIT License.

scheme's Issues

some minor doc issues

In https://github.com/write-you-a-scheme-v2/scheme/blob/master/docs/00_overview.md :

1. (car '(1 2 3)) should return 1 instead of (1).
2. (cadr '(1 2 3)) should return 2 instead of (2 3) or cdr should be used instead of cadr.
3. https://github.com/write-you-a-scheme-v2/scheme/blob/master/Readme.md is not working. the last part should be changed to README.md.
4. https://pages.lip6.fr/Christian.Queinnec/WWW/LiSP.html is not working.

Incomplete TOC?

Is the tutorial incomplete or the TOC is wrong: https://www.wespiser.com/writings/wyas/00_overview.html

I mean the Roadmap part is missing the links to the IO chapter etc...

Function arguments are evaluated in the wrong scope

Once you fix #23, you will find that the following happens:

(begin
  (define y 42)
  (define foo (lambda (x) x))
  (let (y 12) (foo y)))

This evaluates to 42 instead of the expected 12.

There are two causes:

1. applyLambda takes args and evaluates them in the environment that was in scope when the closure was created (instead of assuming that the arguments passed in are already in evaluated form).
2. The function application case of eval incorrectly passes xs (ie, the unevaluated arguments) to the function definition instead of xVal (the evaluated arguments).

Function application uses incorrect scope

Consider the following example:

(begin
  (define foo (lambda (x) y))
  (let (y 42) (foo 0))

This evaluates to 42 rather than the expected: Unbound Variable: y

Slow response and missing key strokes on REPL

The interactive repl, seen when running./run, is very slow and sometimes does not register key presses. Using Haskeline 0.7.2.2

Build failing on linux mint

Running

stack build --test

in the main directory yields the following output:

src/Parser.hs
  Atom
  Num Negative
  Num Positive
  String
  Bool True
  Bool False
  Nil
  S-Expr: homogenous list
  S-Expr: homogenous list quoted
  S-Expr: heterogenous list
  S-Expr: heterogenous list quoted
  S-Expr: prim call: neg nums
  S-Expr: prim call: atoms
  S-Expr: nested list

Finished in 0.0029 seconds
14 examples, 0 failures

src/Eval.hs
  test/add.scm
  test/if_alt.scm
  test/let.scm
  test/eval_boolean.scm
  test/eval_boolean_ops.scm
  test/eval_lambda.scm
  test/test_quote.scm
  test/test_car.scm
  test/test_cdr.scm
  test/test_cadadr.scm
  test/test_gt.scm
  test/test_scope1.scm
  test/test_args.scm
  test/define.scm
  test/define_order.scm

Finished in 0.2299 seconds
15 examples, 0 failures

eval extra -- sandboxing
  begin/define
  eval args
  test/define_lambda.scm
  eval args
  foldl evals to something 
  foldl call FAILED [1]
  fold FAILED [2]

Failures:

  test-hs/Spec.hs:134: 
  1) eval extra -- sandboxing foldl call
       uncaught exception: LispException (Error Type Mismatch: numeric op Nil)

  test-hs/Spec.hs:134: 
  2) eval extra -- sandboxing fold
       uncaught exception: LispException (Error Not a Function: (quote (quote (+))))

Randomized with seed 1448178989

Finished in 0.0720 seconds
7 examples, 2 failures

Further, scheme -s test/eval_lambda.scm yields:

scheme: Error Not a Function: 5

and the code fails to correctly evaluate recursive structures (!), e.g. a factorial program in scheme will not work.

Add support for mutually recursive definitions

Once #23 is fixed, you'll find that the following will not work:

(begin
  (define (even x) (if (== x 0) #t (odd (- x 1)))
  (define (odd x) (if (== x 0) #f (even (- x 1)))
  (even 9))

Returns Unbound variable: odd rather than the expected #f

Update: Actually once #23 and #25 are fixed, you'll find that even singly-recursive functions (such as foldl from the stdlib) will not work. Just trying to get factorial working is worth thinking about.

Parsing Issue w/ (+ 1 2)

from the project dir, run

./run

and enter (+ 1 2), there is a parser error.
The desired output is "List [Atom \"+\", Number 1, Number 2]"
./src/Parse.hs should be updated...

This should be a good chance to write the docs on Parsec and parsing...

weirdness around begin fn

Repl> (begin (let (x 1 y 20 z 300)(+ x y z) ) )
Error Not a Function
Repl> (begin (define zzz 2)(let (x 1 y 20 z 300)(+ x y z) ) )
321

Repl doesn't persist definitions

The Repl doesn't remember definitions:

Repl> (define myid (lambda (x) x))
myid
Repl> (myid 1)
Error Unbound Variable: myid

This works:

Repl> (begin (define myid (lambda (x) x)) (myid 1))
1

Possible suggestion on qualified import

scheme/docs/01_introduction.md

Lines 43 to 44 in e09e25c

	{-# LANGUAGE OverloadedStrings #-}
	import Data.Text as T

While working through the tutorial, I noticed that this would probably make more sense as a qualified import. I can definitely make the change but looking for guidance 😄

Rename repository

I think the repository name of scheme is not descriptive enough and would like to propose changing it to something like write-you-a-scheme-v2 (like the organization). I know I can change the directory to which the project is cloned, but it's more user-friendly if one does not have to do that. :)

Link to "Monad Transformer Step by Step" broken

In the introduction the Evaluation Monad section mentions a guide about using monad transformers but its link - https://www.wespiser.com/writings/sources/Transformers.pdf - is broken. I couldn't find that file anywhere on your website, is it still available online?

parse3.hs

issue with new parser, parse3.hs

[1 of 1] Compiling Main             ( Parse3.hs, Parse3.o )

Parse3.hs:18:9:
    Couldn't match type ‘[Char]’ with ‘T.Text’
    Expected type: Lang.GenLanguageDef T.Text () Identity
      Actual type: Lang.LanguageDef ()
    In the expression: Lang.emptyDef
    In the expression:
      Lang.emptyDef
        {Tok.commentStart = "{-", Tok.commentEnd = "-}",
         Tok.identStart = letter <|> oneOf "+-/*",
         Tok.identLetter = alphaNum <|> oneOf "_'",
         Tok.reservedOpNames = ["'", "\""],
         Tok.reservedNames = ["true", "false", "let", ....]}

A pdf version of wyas-v2?

Hi @adamwespiser , first of all thanks a lot for providing a nice upgrade for Write Yourself a Scheme! As a beginner, it helped me grok Haskell and create projects out of it 👍

I'm just wondering if there would be any .pdf version of this. It may be helpful for someone like me who is not always connected to the internet / prefers reading on a paper print-out.

Thank you so much!

Old version of the tutorial posted at wespiser.com

I believe https://wespiser.com/writings/wyas/home.html does not show the current version of the tutorial. For example https://wespiser.com/writings/wyas/00_overview.html does not include 39f2edf.

If the wespiser.com version of the text is not meant to be updated, then perhaps it would be a good idea to state this explicitly at the home page or in the overview, and recommend using the version from the repo?

Is this project abandoned?

From what I see, 5 years have passed but the last 3 chapters (Standard Library, IO, Conclusions) still have not been addressed (of these, IO being the most important probably). Should the project be considered abandoned, then?

Parsec w/ T.Text

@sdiehl, having problems getting the parser to work creating LispVal values of strings

gce-work:scheme$ stack build && stack exec scheme
Repl> '(a)
List [Atom "a"]

Repl> '("a")
Parser "<stdin>" (line 1, column 4):
unexpected "a"
expecting end of "\"" or literal string

Repl> '(\"a\")
Parser "<stdin>" (line 1, column 4):
unexpected '\\'
expecting end of "("

It looks like the Parsec standard lisp parser is taking a different approach: https://hackage.haskell.org/package/lispparser-0.3.1/docs/src/Text-ParserCombinators-Parsec-Lisp.html

any insight on parsing in double quoted strings from the repl into LispVal?