[teiid-issues] [JBoss JIRA] (TEIID-3938) EdmDate and EdmTimeOfDay output in local timezone

[ https://issues.jboss.org/browse/TEIID-3938?page=com.atlassian.jira.plugin... ] Steven Hawkins commented on TEIID-3938: --------------------------------------- Their own code seems to contradict that, as the assume GMT for parsing. To highlight the issue: TimeZone.setDefault(TimeZone.getTimeZone("GMT+1")); Date date = EdmTimeOfDay.getInstance().valueOfString("11:11:11", true, 4000, 0, 0, true, Date.class); assertEquals("Thu Jan 01 12:11:11 GMT+01:00 1970", date.toString()); String val = EdmTimeOfDay.getInstance().valueToString(date, true, 4000, 0, 0, true); assertEquals("11:11:11", val); The last line will fail because the output will be 12:11:11 instead - which means that the string parse->Java->string cycle is inconsistent when the default timezone is not GMT. This may just be a misunderstanding of Java Date/Time - see EdmDateTimeOffset.createDateTime the case with value instanceOf Date uses Calendar.getInstance(), but has some comment about UTC - which is not meaningful and should not be different than the java.sql.Time handling. They should be using Calendar.getInstance("GMT"). The Calendar case has a similar issue. EdmTimeOfDay and EdmDate only work correctly when represented by either java.sql.Time or Long. So this shouldn't affect us in the TimeOfDay case as the runtime type will be java.sql.Time. However it will be an issue with EdmDate that we either have to workaround by converting our runtime values to longs or get them to fix this. -- This message was sent by Atlassian JIRA (v6.4.11#64026)