reader fails on some complex numbers

[leppie]

Ikarus Scheme version 0.0.3+ (revision 1538, build 2008-07-15)
Copyright (c) 2006-2008 Abdulaziz Ghuloum

> 8+6.0i
Unhandled exception
 Condition components:
   1. &lexical
   2. &message: "invalid numeric sequence"
   3. &irritants: ("8+6.")
   4. &lexical-position:
       file-name: *stdin*
       character: 3
> 8.0+6i
8.0+6.0i
>

[leppie (leppie)]

-6.0i works though.

[leppie (leppie)]

Some more:

> 8.0@6.0
Unhandled exception
 Condition components:
   1. &lexical
   2. &message: "invalid numeric sequence"
   3. &irritants: ("8.0@")
   4. &lexical-position:
       file-name: *stdin*
       character: 63
> 8@6
Unhandled exception
 Condition components:
   1. &lexical
   2. &message: "invalid numeric sequence"
   3. &irritants: ("8@")
   4. &lexical-position:
       file-name: *stdin*
       character: 69

[leppie (leppie)]

Some more:

> -0i
0
> Unhandled exception
 Condition components:
   1. &assertion
   2. &who: eval
   3. &message: "unbound variable"
   4. &irritants: (i)
>

[leppie (leppie)]

> +1i
+i
> Unhandled exception
 Condition components:
   1. &assertion
   2. &who: eval
   3. &message: "unbound variable"
   4. &irritants: (i)

Sorry im just pasting them all here :)

[leppie (leppie)]

> -0.i
0.0

Returns non-complex number, not sure what it should return however :-)

[leppie (leppie)]

Fails in nan.0 too:

> 8+nan.0i
Unhandled exception
 Condition components:
   1. &lexical
   2. &message: "invalid numeric sequence"
   3. &irritants: ("8+n")
   4. &lexical-position:
       file-name: *stdin*
       character: 149

[leppie (leppie)]

Another few:

> 8+6e20i
Unhandled exception
 Condition components:
   1. &lexical
   2. &message: "invalid numeric sequence"
   3. &irritants: ("8+6e")
   4. &lexical-position:
       file-name: *stdin*
       character: 3
> 8e10+6i
Unhandled exception
 Condition components:
   1. &lexical
   2. &message: "invalid numeric sequence"
   3. &irritants: ("8e10+")
   4. &lexical-position:
       file-name: *stdin*
       character: 12

[leppie (leppie)]

Yet another (sorry for all the individual ones, adding as I find it, hopefully it will provide useful for tests :-) ):

> +nan.0+i
Unhandled exception
 Condition components:
   1. &lexical
   2. &message: "invalid numeric sequence"
   3. &irritants: ("+nan.0+")
   4. &lexical-position:
       file-name: *stdin*
       character: 38

[Abdulaziz Ghuloum (aghuloum)]

On Jul 15, 2008, at 8:35 AM, leppie wrote:

> Yet another (sorry for all the individual ones, adding as I find it,
> hopefully it will provide useful for tests :-) ):

It does. Thanks. Keep them coming. :-)

[leppie (leppie)]

Let's call it the night with this one :-)

> #d-0e-10-0e-0i
Unhandled exception
 Condition components:
   1. &lexical
   2. &message: "invalid numeric sequence"
   3. &irritants: ("#d-0e-10-")
   4. &lexical-position:
       file-name: *stdin*
       character: 8

[leppie (leppie)]

Ok here is the best I can come up with:

#d#e-0.0f-0-.0s-0i

[Abdulaziz Ghuloum (aghuloum)]

On Jul 15, 2008, at 3:26 AM, leppie wrote:

>> -0.i
> 0.0
>
> Returns non-complex number, not sure what it should return
> however :-)

Should be: 0.0-0.0i

[Abdulaziz Ghuloum (aghuloum)]

Fixed most of the problems in 1540. What remains are the 1f2, 1s2, and friends (I only have 1e2 for now) and the angular number syntax 1@2.

[leppie (leppie)]

> Fixed most of the problems in 1540.

Wow that was fast :)

[leppie (leppie)]

> Wow that was fast :)

And thanks to your speediness, I got the inspiration to implement it all for IronScheme too :) (just lacking mantissa width, and all exponent markers default to e).

[leppie (leppie)]

One of the tests are broken:

(test "#d-0e-10-0e-0i" (make-rectangular -0.0 -0.0))

should be

(test "#d-0e-10-0e-0i" (make-rectangular -0e-10 -0.0))

The idea was to test the 'minus' token :)

[leppie (leppie)]

Oops, sorry, my mistake! The test IS correct :)

[leppie (leppie)]

Is this correct?

#x+inf.0+nan.0i

My feeling is that this should not be allowed.

Assume this is valid, then it gets contradicted later in the spec with:

"If the representation of a number object has no exactness prefix, the constant is inexact if it contains a decimal point, an exponent, or a nonempty mantissa width; otherwise it is exact."[1]

But nan's and infinities are always inexact.

So it comes down to whether you count the decimal point inside +nan.0 and +inf.0 ...

Cheers

leppie

[1] http://www.r6rs.org/final/html/r6rs/r6rs-Z-H-2.html#node_toc_node_sec_4.2.8

[leppie (leppie)]

I rather raised this on the R6RS mailing list.

[Abdulaziz Ghuloum (aghuloum)]

I think this was fixed, right?

[leppie (leppie)]

Here is another one :)

> 0@-.0
Unhandled exception
 Condition components:
   1. &lexical
   2. &message: "invalid numeric sequence"
   3. &irritants: ("0@-")
   4. &source-position:
       file-name: *stdin*
       character: 384