On 11/02/2010, at 5:27 PM, Chuck Hill wrote:

> On Feb 10, 2010, at 10:14 PM, Lachlan Deck wrote:
> 
>> On 11/02/2010, at 5:10 PM, Chuck Hill wrote:
>> 
>>> On Feb 10, 2010, at 9:49 PM, Lachlan Deck wrote:
>>> 
>>>> I've been debugging a problem we've got where the fetched timestamp is 1 
>>>> hour out for times that fall after (or close after) the date when day 
>>>> light savings comes off.
>>> 
>>> That seems pretty normal for Java where a day is 24 * 60 * 60 * 1000 
>>> milliseconds, not a day like we usually think of it.  If I don't care about 
>>> the time, I just set it to 12:00 noon.  So when Java adjusts the time, it 
>>> is 11AM or 1PM, but at least on the correct day.  Is that what you are 
>>> seeing?  Or am I misunderstanding the problem?
>> 
>> Yes. This is a timestamp / datetime (not a date only value).
>> 
>> The correct value should be 4pm Sydney time. But for this last one it's 5pm.
> 
> Are they all consistently on one side or the other of the DST boundary?  Is 
> this last one different?  Does it matter which one is the last?  Is the last 
> always wrong regardless?

It's not random. The last one happens to be just after the time when daylight 
savings ends.

>> I'll try again :)
>> A list of records in the database with a field called 'endTimestamp' all 
>> have a value where their time-part == 4pm Sydney time. Once fetched, one of 
>> them is wrong.
> 
> Got it that time.  Unless that one is in a different DST vs non-DST that is 
> just plain crazy!

I've finally found the problem. It's the toy again[1] :) 

So these connection params make it work:
...&zeroDateTimeBehavior=convertToNull&useGmtMillisForDatetimes=false&useTimezone=true&serverTimezone=Australia/Sydney&useLegacyDatetimeCode=false

The main difference is whether or not to useGmtMillisForDatetimes. With a 
previous jconnector version there was a bug where it used the local timezone 
regardless and this was necessary... now that this has been /fixed/ this new 
issue came up. Fun. At least this time it should be good for some time to come.

Ugh. Why they don't just have a parameter called doTheRightThingAlready=true is 
beyond me :)

Thanks.

with regards,
--

Lachlan Deck

 _______________________________________________
Do not post admin requests to the list. They will be ignored.
Webobjects-dev mailing list      ([email protected])
Help/Unsubscribe/Update your Subscription:
http://lists.apple.com/mailman/options/webobjects-dev/archive%40mail-archive.com

This email sent to [email protected]

Reply via email to