]
Thomas Diesler reopened JBPM-1137:
----------------------------------
Leap Year Date Calculations (v3.2.2)
------------------------------------
Key: JBPM-1137
URL:
http://jira.jboss.com/jira/browse/JBPM-1137
Project: JBoss jBPM
Issue Type: Support Patch
Components: Core Engine
Affects Versions: jBPM jPDL 3.2.2
Reporter: Alejandro Guizar
Assigned To: JBN Patch Team
Attachments: jbpm-jpdl.jar
Based on JBPM-1098 there is concern over date calculations in in jBPM 3.1 and 3.2. There
might be issues around the end of Feb. The problem is characterized as follows. Under the
current model, durations that are not expressed in months or years work correctly.
Consider the following expression.
endDate = new Date(startDate.getTime() + interval);
If startDate is February 28, 2008 and interval is 1 day, then endDate will be February
29, 2008. Every day has the same duration even on leap years. The presentation layer is
responsible of converting the milliseconds in a Date to a suitable representation in the
calendar system. The Java APIs are all aware of leap years, so this is covered as well.
The situation is different for months and years. If the interval above was 1 year, then
endDate would be February 27, 2009. Because of the extra day in 2008, the expected endDate
has shifted backwards one day. To fix this, the Duration class has to preserve the
original unit of time instead of converting to milliseconds up front. Calculation of
endDate can then be delegated to java.util.Calendar as follows.
startCalendar = Calendar.getInstance();
startCalendar.setTime(startDate);
startCalendar.add(unit, interval);
endDate = startCalendar.getTime();
--
This message is automatically generated by JIRA.
-
If you think it was sent incorrectly contact one of the administrators: