[PIG-1781] Piggybank: ISOToDay disregards timezone (should use ISODateTimeFormat instead of DateTime to parse)

• Type: Bug
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 0.8.0

Release Note

in UDFs in org.apache.pig.piggybank.test.evaluation.datetime.truncate, add support for non-UTC timezones in ISO 8601 datetime strings,

• Assignee: Michael Brauwerman
• Reporter: Michael Brauwerman
• Created: 23/Dec/10 20:14
• Updated: 20/Jan/11 18:59
• Resolved: 20/Jan/11 18:59

Description

(Apologies if this is the wrong place to file Piggybank bugs)

Bug in http://svn.apache.org/viewvc/pig/trunk/contrib/piggybank/java/src/main/java/org/apache/pig/piggybank/evaluation/datetime/truncate/ISOToDay.java?view=markup

and other http://svn.apache.org/viewvc/pig/trunk/contrib/piggybank/java/src/main/java/org/apache/pig/piggybank/evaluation/datetime/truncate/ classes that copy-paste the same code.

These classes parse dates like so:
DateTimeZone.setDefault(DateTimeZone.UTC);
DateTime dt = new DateTime((String)input.get(0).toString());

This has two problems:
(1) It messes up JVM static state by changing the DateTimeZone default time zone.
(2) It ignore timezone information in the input string, so times like "2009-12-09T23:59:59-0800" get truncated to "2009-12-10T00:00:00Z", which is the wrong day of year.

Instead, they should use something like this, which respects the input timezone and does not modify any global state:

DateTime dt ISODateTimeFormat.dateTime().withOffsetParsed().parseDateTime(isoDateString);

I have not provided a patch, because I'm not really set up to hack on Piggybank locally.

As a workaround, I am copy-pasting the classes into my own packages, and making the desired change.