Atof

[giskumar]

Hi all,

 

I have a problem while converting the string to real number.

 

my string is like "5644.55688442" when i use (atof "5644.55688442") it returning only 4 points after period like 5644.5568.

 

help me how can i fix this.

 

Thanks,

Kumar.

[lpseifert]

the accuracy is still there, it's just not displayed

 

checked with this

$ (atof "5644.55688442")
5644.56
_$ (rtos (atof "5644.55688442") 2 
"5644.55688442"

[giskumar]

Thanks a lot for quick reply

[Ahankhah]

You can check that atof always returns a real with 6 digits, regardless of what is the precision of returned number:

 

(atof "123456789"); returns 1.2345[color=red]7[/color]e+008
(atof "12345678.9"); returns 1.2345[color=red]7[/color]e+007
(atof "1234567.89"); returns 1.2345[color=red]7[/color]e+006
(atof "123456.789"); returns 12345[color=red]7[/color].0
(atof "12345.6789"); returns 12345.[color=red]7[/color]
(atof "1234.56789"); returns 1234.5[color=red]7[/color]
(atof "123.456789"); returns 123.45[color=red]7[/color]
(atof "12.3456789"); returns 12.3457
(atof "1.23456789"); returns 1.2345[color=red]7[/color]
(atof "0123456789"); returns 1.2345[color=red]7[/color]e+008

 

Howerver, the real precision (accuracy) of real numbers in AutoLISP is 16 digits.

 

to get the number as desired precision, use rtos with atof:

 

(rtos (atof "1.234567890123456789") 2 0); returns "1"
(rtos (atof "1.234567890123456789") 2 1); returns "1.2"
(rtos (atof "1.234567890123456789") 2 2); returns "1.23"
(rtos (atof "1.234567890123456789") 2 3); returns "1.23[color=red]5[/color]"
(rtos (atof "1.234567890123456789") 2 4); returns "1.234[color=red]6[/color]"
(rtos (atof "1.234567890123456789") 2 5); returns "1.2345[color=red]7[/color]"
(rtos (atof "1.234567890123456789") 2 6); returns "1.23456[color=red]8[/color]"
(rtos (atof "1.234567890123456789") 2 7); returns "1.234567[color=red]9[/color]"
(rtos (atof "1.234567890123456789") 2 ; returns "1.23456789"
(rtos (atof "1.234567890123456789") 2 9); returns "1.23456789"
(rtos (atof "1.234567890123456789") 2 10); returns "1.2345678901"
(rtos (atof "1.234567890123456789") 2 11); returns "1.23456789012"
(rtos (atof "1.234567890123456789") 2 12); returns "1.234567890123"
(rtos (atof "1.234567890123456789") 2 13); returns "1.234567890123[color=red]5[/color]"
(rtos (atof "1.234567890123456789") 2 14); returns "1.2345678901234[color=red]6[/color]"
(rtos (atof "1.234567890123456789") 2 15); returns "1.23456789012345[color=red]7[/color]"
(rtos (atof "1.234567890123456789") 2 16); returns [b]"1.23456789012345[color=red]6[/color]" [color=red]<--[/color][/b]
(rtos (atof "1.234567890123456789") 2 17); returns [b]"1.23456789012345[color=red]6[/color]" [color=red]<--[/color][/b]

[Lee Mac]

Also, be sure to note that the precision is 16 digits, not decimal places, as the following code demonstrates:

 

(defun test ( n / i a b ) (setq i -1 a "")
 (while
     (/=
         (strlen (setq b (rtos n 2 (setq i (1+ i)))))
         (strlen a)
     )
     (print (setq a b))
 )
 (princ
     (strcat "\n\nMaximum Precision: "
         (itoa
             (- (strlen b) (cond ( (vl-string-position 46 b) ) ( (1- (strlen b)) )) 1)
         )
         " d.p."
     )
 )
 (princ)
)

 

_$ (test 1.12345678911234567891) ;; 20 d.p, 21 digits

"1" 
"1.1" 
"1.12" 
"1.123" 
"1.1235" 
"1.12346" 
"1.123457" 
"1.1234568" 
"1.12345679" 
"1.123456789" 
"1.1234567891" 
"1.12345678911" 
"1.123456789112" 
"1.1234567891123" 
"1.12345678911235" 
"1.123456789112346" 

Maximum Precision: 15 d.p.

 

_$ (test 123.12345678911234567891) ;; 20 d.p, 23 digits

"123" 
"123.1" 
"123.12" 
"123.123" 
"123.1235" 
"123.12346" 
"123.123457" 
"123.1234568" 
"123.12345679" 
"123.123456789" 
"123.1234567891" 
"123.12345678911" 
"123.123456789112" 
"123.1234567891123" 

Maximum Precision: 13 d.p.

 

_$ (test 1234567891.12345678911234567891) ;; 20 d.p, 30 digits

"1234567891" 
"1234567891.1" 
"1234567891.12" 
"1234567891.123" 
"1234567891.1235" 
"1234567891.12346" 
"1234567891.123457" 

Maximum Precision: 6 d.p.

 

Hence for a Real with integer part of length 16 digits:

 

_$ (test 1234567891123456.12345678911234567891) ;; 20 d.p, 36 digits

"1234567891123456" 

Maximum Precision: 0 d.p.

[eakececioglu]

Hi All,

I have a question about the cad rounding proccess.

For example;

I created a line from command-line but huge coordinate (15digit.15digit ex. Start coor.:(123456789012345.123456789012345,0). 

After that, I use the filter command and explore to "filter.nfl" file, autocad return the start coordinate 123456789012345.156250000000000. I hope rounding value "~.123460" but value is "~.156250"

Why?