srfi 1 test failing on OS X

[saysjonathan]

$ make test-srfi
<snip>
case 13 FAIL: (iota 5 0 -0.1)
   expected (0 -0.1 -0.2 -0.3 -0.4) but got (0 -0.1 -0.2 -0.30000000000000007 -0.4)
<snip>
 FAIL: (iota 5 0 -0.1)
   expected (0 -0.1 -0.2 -0.3 -0.4) but got (0 -0.1 -0.2 -0.30000000000000007 -0.4)

[KeenS]

Verified. It is basically precision problem but there might be something we can do for

[saysjonathan]

Precision issues appear to transcend OS. Verified on Linux:

> (+ 0.1 (+ 0.1 (+ 0.1)))
0.30000000000000007

$ uname -sr
Linux 3.10.0-042stab108.5

[nyuichi]

It seems we cannot fix this issue. The test case should be changed.

[saysjonathan]

What's preventing this from being fixed?

[nyuichi]

Picrin doesn't have internal representation of exact rational number so 0.1 is implicitly converted to double. We cannot avoid this kind of error as long as FPU computes in binary. Meanwhile, I got a similar result running python:

$ python
Python 2.7.10 (default, Oct 23 2015, 18:05:06) 
[GCC 4.2.1 Compatible Apple LLVM 7.0.0 (clang-700.0.59.5)] on darwin
Type "help", "copyright", "credits" or "license" for more information.
>>> 0.1 + 0.1 + 0.1
0.30000000000000004
>>> 

[saysjonathan]

Makes perfect sense.

Same for Ruby:

irb(main):001:0> 0.1 + 0.1 + 0.1
=> 0.30000000000000004

and Haskell:

Prelude> 0.1 + 0.1 + 0.1
0.30000000000000004

👍 for changing the test case.

[saysjonathan]

In case anyone is interested, Chibi and Guile both work:

$ chibi-scheme
> (* 0.1 3)
0.3
> (* 0.1 12)
1.2

$ guile
guile> (* 0.1 3)
0.3
guile> (* 0.1 12)
1.2