Date duration calculation in the BLS

[Former Member]

All,

We are using MII 12.0.6

I am looking at a very peculiar behavior while calcualting the date difference between 2 dates.

For ex. the following expression:

(datediffminutes(datetoxmlformat("2010-3-21 14:06:30 ", "yyyy-MM-dd hh:mm:ss"), datetoxmlformat("2010-05-24 08:14:54","yyyy-MM-dd hh:mm:ss")))/60

yields 1530.13 as the result(this is the correct evaluation) when tested from within the Link editor (clicking the evaluate button).

Now saving this transaction and giving the value of the above expression to an output param followed by calling this trx from an Xacute query yields 1529.13 as the result (I hr difference than above).

I have tested this with a few more dates going as far back as Dec 2009 yielding the same one hour difference.

On the other hand for any start date in May 2010 yields a correct result even when called from an Xacute query.

I tested this in a 12.0.2 system and found this behavior absent as in all was working as expected.

Has anyone come accross this before?Do you know if the latest patch has this fixed?

Just to cross check can anyone giive this a try in their 12.0.6 or higher system and report back?

Thanks a lot for your help.

Kind Regards

Udayan

[jcgood25 · Accepted solution]

12.0.10 Build(12) behaves the same as your present system.

I noticed you were using hh and not HH in your datetime formats - using either 12 or 24 hour formats worked the same as well.

[agentry_src · Accepted solution]

Udayan,

Try dates from Dec to February. Then try to pin down the problem date areas in March and April. It is probably related to Daylight Savings time (somehow).

Regards,

Mike

[Former Member · Accepted solution]

Jeremy,

Are we saying then, this is a bug?Would logging a support tkt help?

Mike,

Seems 28th Mar onwards works well. Anything before that shows the one hr difference.

[Former Member · Accepted solution]

This is what I've observed so far .

For start & end dates in the range 28th Mar to 24th Oct (irrespective of the year) the durations is calculated as expected.

For dates falling outside this range a one hour difference is observed .

It is strange but so far true!

[agentry_src · Accepted solution]

Hi Udayan,

Please open a ticket. It sounds like a bug, although a few years back there was a problem with one of the java packages and Daylight Savings Time. If you search this forum for all time, you will find some of the postings.

Since there have been changes to when Daylight Savings Time start and ends, it is likely related and could possibly be java related instead of MII. I doubt it, but it is possible.

Edit:

A little more research turned up differences in how Daylight Savings Time is calculated in different countries/continents. Coincidentally, the US started on March 14th this year while the European Union started on March 28th (this date sounds familiar). Not sure what timezone, you are in, but you may want to play around with the timezone settings for your computer to see if the error is consistent with US vs European timezones (or wherever you happen to be located).

[Daylight Savings Times around the world|http://www.worldtimezone.org/]

Thanks,

Mike

Edited by: Michael Appleby on May 25, 2010 12:41 PM

[Former Member · Accepted solution]

Mike,

Thanks a lot for your help.

This MII server in question is in CET and you are right the clock changed in Europe on the 28th Mar this year.

So it is very much possible what you are saying is correct.

Now if at all this is a Java related issue , where do I need to look? I mean with 12.x the administration is usually with the BASIS people , where should I tell them to look?

Once again I really appreciate your help in this matter.

Best regards

Udayan

[agentry_src · Accepted solution]

Hi Udayan,

First, please test it with a date of March 13. If it is a Daylight Savings Time issue, it should start returning the correct value. If so, that would mean that there is a problem with the two week difference between US time and CET time.

I think that once that has been proved, we need to have a ticket and let the developers address it. It is a bigger issue than just MII as the EU and US decided to go different ways when the DST changes were implemented in 2007. And it is causing headaches globally.

As far as java is concerned, the developers will have to answer that one.

Wish I had a better solution for you.

Mike

Edited by: Michael Appleby on May 25, 2010 7:02 PM