PM41723: ROUND() REPORTS 'DIGITS < 0' FOR ARGUMENT BETWEEN 0 AND 0.1, AND RETURNS INCORRECT RESULT FOR ARGUMENT BETWEEN 0.1 AND 1.0

8.0.0.1: WebSphere Application Server V8.0 Fix Pack 1
8.0.0.2: WebSphere Application Server V8.0 Fix Pack 2
8.0.0.3: WebSphere Application Server V8.0 Fix Pack 3
8.0.0.4: WebSphere Application Server V8.0 Fix Pack 4
8.0.0.5: WebSphere Application Server V8.0 Fix Pack 5
8.0.0.6: WebSphere Application Server V8.0 Fix Pack 6
8.0.0.7: WebSphere Application Server V8.0 Fix Pack 7
8.0.0.8: WebSphere Application Server V8.0 Fix Pack 8
8.0.0.9: WebSphere Application Server V8.0 Fix Pack 9
8.0.0.10: WebSphere Application Server V8.0 Fix Pack 10
8.0.0.11: WebSphere Application Server V8.0 Fix Pack 11
8.0.0.12: WebSphere Application Server V8.0 Fix Pack 12
8.0.0.13: WebSphere Application Server V8.0 Fix Pack 13
8.0.0.14: WebSphere Application Server V8.0 Fix Pack 14
8.0.0.15: WebSphere Application Server V8.0 Fix Pack 15

APAR status

• Closed as program error.

Error description

• If the argument to the round function is greater than zero,
  but less than 0.1, the processor reports the error message
  "Digits < 0".  If the argument to the function is greater than
  or equal to 0.1 and less than 1, the processor returns the
  value of the argument as the result of the function, rather
  than the rounded result.
  

Local fix

Problem summary

• ****************************************************************
  * USERS AFFECTED:  Users of the XML API for IBM WebSphere      *
  *                  Application Server V8.0                     *
  ****************************************************************
  * PROBLEM DESCRIPTION: For arguments to the fn:round function  *
  *                      in the range 0 to 1, the processor      *
  *                      sometimes reports the error "Digits <   *
  *                      0" and sometimes returns the argument   *
  *                      as the result of the function,          *
  *                      instead of rounding correctly           *
  ****************************************************************
  * RECOMMENDATION:  Install a fix pack that contains this APAR. *
  ****************************************************************
  For an argument of type xs:float, xs:double or xs:decimal, the
  implementation of the fn:round function always represents the
  value as a java.math.BigDecimal object.  It then rounds that
  BigDecimal value using the BigDecimal.round method, supplying
  a java.math.MathContext whose precision is equal to the
  precision of the argument minus the absolute value of its
  scale.
  
  For all values whose absolute value is greater than or equal
  to one, that method of calculating the absolute value is
  correct.
  
  However, for a value that is greater than zero but less than
  0.1, that yields a negative value, and constructing a
  MathContext with a precision less than zero results in a Java
  Exception containing the message, "Digits < 0".  For a value
  that is greater than or equal to 0.1, but less than one, that
  formula yields a MathContext whose precision is zero, which
  indicates to the BigDecimal.round function that no rounding
  should be performed, which results in the argument being
  returned as the result as the fn:round function
  

Problem conclusion

• The problem was resolved by changing the implementations of
  the fn:round function for values of type xs:float or xs:double
  to be a more efficient, correct algorithm that is specific to
  values of those floating point types, rather representing such
  values as Java BigDecimal objects.
  
  In the case of values of type xs:decimal, the implementation
  of fn:round was divided into three cases:  if the value is less
  than 0.5 and greater than or equal to -0.5, the implementation
  returns the result of 0; if the absolute value is less than
  one, and the value is greater than or equal to 0.5 or less
  than -0.5, the implementation returns the result of one, for
  positive values, or negative one, for negative values;
  otherwise, the implementation computes the result using the
  existing technique of using the BigDecimal.round method.
  
  The fix for this APAR is currently targeted for inclusion in
  fix pack 8.0.0.1.  Please refer to the Recommended Updates
  page for delivery information:
  http://www.ibm.com/support/docview.wss?rs=180&uid=swg27004980
  

Temporary fix

Comments

APAR Information

• APAR is sysrouted FROM one or more of the following:

  PM33860

• APAR is sysrouted TO one or more of the following:

Fix information

• Fixed component name

  WEBSPHERE APP S

• Fixed component ID

  5724J0800

Applicable component levels

• R800 PSY