Float.valueOf for a String & round decimal

[Anonymous]

hi all
we've some "strange" rounding behavior with reading a String and convert to a Float value.
from an Xml file, we insert 456193.15 as a String (mysql table).
When I read it & convert it to a Float (with dedicated component or Float.valueOf method) it "rounded" to 456193.16.
Agree for some value like ......,98 or .....,99 !
is it a normal behavior, even if I know that Float doesn't keep "precision"
thanks for advice
regards
laurent

[Anonymous]

hum ..
perhaps the reason is :

Rounding errors: Not every decimal number can be expressed exactly as a floating point number. This can be seen when entering "0.1" and examining its binary representation which is either slightly smaller or larger, depending on the last bit

???
laurent

[Anonymous]

Hi Laurent !
Note that trailing format specifiers, specifiers that determine the type of a floating-point literal (1.0f is a float value; 1.0d is a double value), do not influence the results of this method. In other words, the numerical value of the input string is converted directly to the target floating-point type. In general, the two-step sequence of conversions, string to double followed by double to float, is not equivalent to converting a string directly to float. For example, if first converted to an intermediate double and then to float, the string
"1.00000017881393421514957253748434595763683319091796875001d"
results in the float value 1.0000002f; if the string is converted directly to float, 1.0000001f results.
http://docs.oracle.com/javase/1.4.2/docs/api/java/lang/Float.html#valueOf%28java.lang.String%29
Mohcine

[Anonymous]

hi Cissine
thanks for complement of informations.

Make some other test , and Float type round value as it want
As preconize Big Decimal is the only way to be keep value as it ....
I was suprising about rouding some values (456193.15) that dont' look like to be rounded (oups !!!
But seem to be the normal way .
@+
laurent

[Anonymous]

Hello
During convertation Float.parseFloat("1.123456789") the result is 1.123 float, but it should be 1.123456789.
Please help with that problem.

[_AnonymousUser]

Hi I have also experienced this issue and found it only affected large numbers (100K +) and discovered that using the talend type of double has corrected it.