Bug #5294: to_f does not convert floating point number the same as Fortran

• Status: Rejected
• Assignee: -
• Target version: -
• ruby -v: 1.9.2p180 (2011-02-18) [i386-mingw32]

[todd.kushner (Todd Kushner)]

We were converting some Fortran programs to Ruby, and found the following floating point number format did not convert as Fortran does on input, i.e., on:

'[+-]nn.E[+-]nn'

the exponent is ignored by to_f, but not by Fortran:

string: '9.E+20 ' float: 9.0000000000000000000000000
string: '+9.E+20 ' float: 9.0000000000000000000000000
string: '-9.E+20 ' float: -9.0000000000000000000000000
string: '9.E20 ' float: 9.0000000000000000000000000
string: '+9.E20 ' float: 9.0000000000000000000000000
string: '-9.E20 ' float: -9.0000000000000000000000000
string: '9.E-20 ' float: 9.0000000000000000000000000
string: '+9.E-20 ' float: 9.0000000000000000000000000
string: '-9.E-20 ' float: -9.0000000000000000000000000

Listed below, we found all other variations we could think of seemed to work properly:

string: '0.9E+21 ' float: 900000000000000000000.0000000000000000000000000
string: '+0.9E+21 ' float: 900000000000000000000.0000000000000000000000000
string: '-0.9E+21 ' float: -900000000000000000000.0000000000000000000000000
string: '.9E+21 ' float: 900000000000000000000.0000000000000000000000000
string: '+.9E+21 ' float: 900000000000000000000.0000000000000000000000000
string: '-.9E+21 ' float: -900000000000000000000.0000000000000000000000000
string: '9E+20 ' float: 900000000000000000000.0000000000000000000000000
string: '+9E+20 ' float: 900000000000000000000.0000000000000000000000000
string: '-9E+20 ' float: -900000000000000000000.0000000000000000000000000
string: '9.0E+20 ' float: 900000000000000000000.0000000000000000000000000
string: '+9.0E+20 ' float: 900000000000000000000.0000000000000000000000000
string: '-9.0E+20 ' float: -900000000000000000000.0000000000000000000000000
string: '0.9E21 ' float: 900000000000000000000.0000000000000000000000000
string: '+0.9E21 ' float: 900000000000000000000.0000000000000000000000000
string: '-0.9E21 ' float: -900000000000000000000.0000000000000000000000000
string: '.9E21 ' float: 900000000000000000000.0000000000000000000000000
string: '+.9E21 ' float: 900000000000000000000.0000000000000000000000000
string: '-.9E21 ' float: -900000000000000000000.0000000000000000000000000
string: '9E20 ' float: 900000000000000000000.0000000000000000000000000
string: '+9E20 ' float: 900000000000000000000.0000000000000000000000000
string: '-9E20 ' float: -900000000000000000000.0000000000000000000000000
string: '9.0E20 ' float: 900000000000000000000.0000000000000000000000000
string: '+9.0E20 ' float: 900000000000000000000.0000000000000000000000000
string: '-9.0E20 ' float: -900000000000000000000.0000000000000000000000000
string: '0.9E-19 ' float: 0.0000000000000000000900000
string: '+0.9E-19 ' float: 0.0000000000000000000900000
string: '-0.9E-19 ' float: -0.0000000000000000000900000
string: '.9E-19 ' float: 0.0000000000000000000900000
string: '+.9E-19 ' float: 0.0000000000000000000900000
string: '-.9E-19 ' float: -0.0000000000000000000900000
string: '9E-20 ' float: 0.0000000000000000000900000
string: '+9E-20 ' float: 0.0000000000000000000900000
string: '-9E-20 ' float: -0.0000000000000000000900000
string: '9.0E-20 ' float: 0.0000000000000000000900000
string: '+9.0E-20 ' float: 0.0000000000000000000900000
string: '-9.0E-20 ' float: -0.0000000000000000000900000

Thanks!

Todd Kushner
Jeppesen
[e-mail]

[todd.kushner (Todd Kushner)]

snippet:

a9e20s = %w{
0.9E+21
+0.9E+21
-0.9E+21
.9E+21
+.9E+21
-.9E+21
9E+20
+9E+20
-9E+20
9.E+20
+9.E+20
-9.E+20
9.0E+20
+9.0E+20
-9.0E+20
(etc...)
}
a9e20s.each { |num|
printf "string: '%-10s' float: %50.25f\n", num, num.to_f
}

[nobu (Nobuyoshi Nakada)]

• Status changed from Open to Rejected

Decimal point without fragment part is invalid in Ruby.

[naruse (Yui NARUSE)]

Nobuyoshi Nakada wrote:

Decimal point without fragment part is invalid in Ruby.

It is because 9.E+20 is parsed as 9.E() + 20.
9: Fixnum
E: a method named "E"
+: a method plus
20: Fixnum