(equal? + +) raises an error. Found by Daniel Patterson.

On an unrelated note, (eq? + +) returns false because we're rebuilding toplevel function lambda, which is not right.

Ryan Lester's getting this error message.

using toWrittenString as the hashCode for an eq-based hashtable is completely wrong. This issues was found by Tala Huhe.

We need to assign some kind of arbitrary memory location to every object allocated by Moby: that way, we can compute hashCode trivially. Basically, every constructor must do:

this._hashCode = plt.Kernel._hashAcc++

where the Kernel creates a new hashCode number.

From Hari Naraynan:

Hari and Scott found what I think is a weird rendering issue in Moby. Their code, attached, contains some nice and simple js-selects, but, when the user selects something from the select box, extra elements get inserted into the DOM tree. Their draw-html function (called "draw") returns a js-div as the root; the extra are inserted as its siblings. (Found through debugging with Firebug.) Firefox renders them really weirdly; neither Konqueror nor the Androids display anything, and weren't not sure if that's a difference between the Javascript interpretation or if khtml/WebKit just don't render not inside a .

Identified as a bug in jsworld, and hopefully fixed by f2b1fbd

Not sure exactly what the conditions for hitting this are, but two students saw this when using a program with include.

zero-arity append's allowed in advanced.

fixed in 3cc30b3

equal? is breaking on circular data structures.

from Jonathan Sailor:

can we get:

(hash (k v) ...)

as syntactic sugar for:

(let ([h (make-hash)])

 (begin
   (hash-set! k v) ...
   h))

From Jonathan Sailor:

Firstly, check-expect on hashes in Moby seems to be eq?, rather than
equal?. For example, in

(define (mh a b c d)

 (let ([h (make-hasheq)])
   (begin
     (hash-set! h a b)
     (hash-set! h c d)
     h)))

(define hash-1 (mh 'a 'b 'c 'd)) (define hash-2 (mh 'a 'b 'c 'd)) (define hash-3 (mh 'c 'd 'a 'b))

(check-expect hash-1 hash-1) ; works (check-expect hash-1 hash-2) ; not eq, but same created identically (check-expect hash-1 hash-3) ; same mapping set, maybe different order

from Jonathan Sailor:

A student pointed out that, in ASL, make-vector can be used as:

(check-expect (make-vector 4) (vector 0 0 0 0)) (check-expect (make-vector 4 2) (vector 2 2 2 2))

but the two-argument form of make-vector is a compile-time error in Moby.

If Moby is unhappy with multi-arg (but arbitrarly-many arg) functions, we're more than able to just doc it that way. :-)

The program:

http://7.latest.wescheme.appspot.com/view?publicId=cough-smoke-tummy-woozy-crude

causes a bad crash on Firefox 3, with the following debugging information:

[Exception... "Component returned failure code: 0x80040111 (NS_ERROR_NOT_AVAILABLE) [nsIDOMCanvasRenderingContext2D.drawImage]" nsresult: "0x80040111 (NS_ERROR_NOT_AVAILABLE)" location: "JS frame
:: http://localhost:8076/runtime/compressed-runtime.js :: anonymous :: line 1" data: no]

Error was raised around offset=2309 line=45 span=200 id="unsaved-editor892"

make-hash for equality hashes are necessary.

copy and paste bug of the toWrittenString() method. This will be fixed in b830ffb

reported by Emmanuel Schanzer

from the test-repl,

(local [] 1 2 3)

is not dying as expected.

(define (attendance price)
  (+ (* (/ 15 .10) (- 5.00 price))
     120))
(attendance 4.9)

should give us 135, but it's giving us the wrong value. This might be fixed as soon as we incorporate bignums.

(define (f 1) (+ x 10))

says:

MobyTypeError: symbol->string: expects type as 1st argument, given: 1

which is wrong: it needs to say that the thing isn't an identifier.

The expression:

(+ (10) 20)

returns the error message:

TypeError: Result of expression '((plt.types.Rational.makeInstance(10, 1))).apply' [undefined] is not a function.

which is not right: this is leaking the abstraction between Scheme and Javascript.

the image primitives, in particular, and anything dealing with the dom, need to be moved outside the kernel, because it'll interfere in contexts where we're not doing any web-browser stuff.

if a jsworld program with an on-tick errors out, it will continue to error out at the rate of the on-tick. This is unacceptable.

(apply (lambda (x) x) (list 3 4 5)) should raise a runtime error, but it doesn't at the moment. We need procedure values to maintain their arities.

we need to keep an id hash table to avoid looping in toWrittenString, toDomNode, toDisplayedString

the world functions are missing contracts.

complex numbers currently are implmeneted as two floating point numbers, which is wrong.

should be returning complex numbers, but are currently breaking
(integer-sqrt -10) should be 0+3i

require is necessary for modularity.

When reporting a shared value, Moby inserts whatever the initial, non-shared value was, even though upon inspection the shared value is already there.

(define-struct something (thing)) (define a (make-something #f)) (define b (make-something 7)) (set-something-thing! a b) (check-expect a 5)

reports that (make-something #f) is not equal to 5, but replacing the check-expect with (check-expect (something-thing (something-thing a)) 7) succeeds.

complex numbers are missing from the lexer

+inf.0 is giving an error.

I get the following scribble warning from moby 1.1:
setup-plt: WARNING: undefined tag in /dyoo/moby.plt/1/1/manual.scrbl:
setup-plt: ((planet "moby-lang.ss" ("dyoo" "moby.plt" 1 1) "src") recur)
Also, all of the links in the moby documentation on the planet web site are broken. Probably you should use cce/scheme:6:0/planet to fix this.

(from samth on http://planet.plt-scheme.org/trac/ticket/215)

the let and letrec tests fail rarely in firefox on Linux with a complaint about being unable to call optimized closures. I don't know what the exact conditions are to trigger the problem yet.

Fix the lexer to accept inexact literals. Via David van Horn:

i0

Hit an error around: offset=0 line=1 span=3 id=""

js-bidirectional-input is too confusing for people, and it's buggy as well.

TODO: remove split-whitespace from the core and the documentation. Adjust examples to have their own implementation of it in Scheme.

on-tick should be on-tick!, on-tilt should be on-tilt!, etc.

(define f (lambda (x . y) y)) is being treated as a 3-arity function. '.' is being translated to the symbol of the single period.

Evaluating:

'(,this)

at the repl gives a funky error

TypeError: buffer.shift() is undefined