You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@geronimo.apache.org by Alex Blewitt <Al...@ioshq.com> on 2003/09/05 12:33:16 UTC
Re: [javamail] Failing tests - MailDateFormat
On Friday, Sep 5, 2003, at 02:42 Europe/London, Dain Sundstrom wrote:
> I fixed it. In MailDateFormatTest, you are parsing a date into a Date
> object and then loading it into a Calendar to test that the parse was
> correct. The problem is when you create the Date object you lose you
> time zone information, so when you create the Calendar it creates it
> in the machine timezone. I'd bet you are in BST so it worked for you,
> but in any other timezone it fails. I change the test to explicitly
> set the timezone on the new Calendar.
| had a nasty feeling that timezones may be the issue. Don't you just
love unit (and international) testing that catches these glitches
before they go live :-)
Well done.
> I was personally surprised at workings of Calendar. I suggest when
> ever we do Calendar tests we put the timezone somewhere in the middle
> of the Pacific Ocean.
Yeah, good plan. What about -0300 as a time zone -- does that exist
anywhere? Or, alternatively, -0330, because timezones don't always have
to be on the hour boundary ...
Alex.
Re: [javamail] Failing tests - MailDateFormat
Posted by Chris Gerrard <ch...@gerrard.net>.
At 06:33 AM 9/5/2003, you wrote:
>On Friday, Sep 5, 2003, at 02:42 Europe/London, Dain Sundstrom wrote:
>
>>I fixed it. In MailDateFormatTest, you are parsing a date into a Date
>>object and then loading it into a Calendar to test that the parse was
>>correct. The problem is when you create the Date object you lose you
>>time zone information, so when you create the Calendar it creates it in
>>the machine timezone. I'd bet you are in BST so it worked for you, but
>>in any other timezone it fails. I change the test to explicitly set the
>>timezone on the new Calendar.
>
>| had a nasty feeling that timezones may be the issue. Don't you just love
>unit (and international) testing that catches these glitches before they
>go live :-)
>
>Well done.
>
>>I was personally surprised at workings of Calendar. I suggest when ever
>>we do Calendar tests we put the timezone somewhere in the middle of the
>>Pacific Ocean.
>
>Yeah, good plan. What about -0300 as a time zone -- does that exist
>anywhere? Or, alternatively, -0330, because timezones don't always have to
>be on the hour boundary ...
>
>Alex.
If I'm not mistaken,
-0300 is Greenland and most of Brazil, and
-0330 is Newfoundland, Canada's easternmost province,
but
-1100 covers the Pacific Ocean just west of Hawaii and I don't think the
International Date Line comes into play.
-0200 might be useful, just nips the east coast of Brazil and squeezes
between Iceland and Greenland