Rails Ruby on Rails → 3.0.2

Comments and changes to this ticket

• You flagged this item as spam.
  

  Levin Alexander February 25th, 2009 @ 06:09 AM

  ActiveSupport::TimeWithTimezone behaves as expected:

  
  >> Time.local(2009,3,29,0,0,0).in_time_zone("CET") + 24.hours
  => Mon, 30 Mar 2009 01:00:00 CEST +02:00 # correct
  

• You flagged this item as spam.
  

  Levin Alexander February 25th, 2009 @ 06:36 AM

  There are unit tests in time_ext_test.rb that call for the incorrect behavior:

  
    def test_daylight_savings_time_crossings_forward_start
      with_env_tz 'US/Eastern' do
        # st: US: 2005 April 2nd 7:27pm
        assert_equal Time.local(2005,4,3,19,27,0), Time.local(2005,4,2,19,27,0).since(86400), 'st+1.day=>dt'
        assert_equal Time.local(2005,4,4,19,27,0), Time.local(2005,4,3,19,27,0).since(86400), 'dt+1.day=>dt'
      end
  

  Is Time#since really supposed to behave this way?

• 

  Geoff Buesing February 25th, 2009 @ 03:27 PM

  • State changed from “new” to “open”
  • Milestone cleared.

  I agree, this is wonky behavior. Reason it exists at all is because, in the days before Rails had a proper Duration class (i.e., 2006), this was a way to fake a duration of 1 day and have it cross a DST boundary correctly.

  Now that we have Duration, it's not needed, it's confusing, and needs to go. Adding 24.hours should always advance the time exactly 24 hours, and adding 1.day should advance to the same time on the next day, which will be 23-25 hours later, depending upon if a DST boundary was crossed.

  A patch for this would be welcome.

• You flagged this item as spam.
  

  Michael Curtis March 10th, 2009 @ 02:54 AM

  • Tag changed from “active_support, duration, timezone” to “active_support, duration, patch, timezone”

  This should take care of it. I've updated the unit tests in time_ext_test.rb that were enforcing the incorrect behavior, and added a couple tests to duration_test.rb that test this specific case.

  This is my first patch to rails.. if I did anything wrong please let me know!

• 

  Manfred Stienstra March 10th, 2009 @ 04:40 PM

  Michael, can you remove the commented lines? Also, the f variable doesn't seem necessary.

• 

  Pratik March 10th, 2009 @ 04:55 PM

  • Assigned user set to “Geoff Buesing”
• 

  Geoff Buesing March 10th, 2009 @ 05:02 PM

  Also, no longer a need to go through Numeric#since -- i.e. seconds.since(self) -- we should now be able to call Time#+ directly, i.e. self + seconds.

  Time#since effectively becomes "plus with datetime fallback".

• You flagged this item as spam.
  

  Michael Curtis March 10th, 2009 @ 05:19 PM

  • Assigned user cleared.

  Here's a new patch, reduced to one commit and no longer going through Numeric#since.

  • fix-duration-on-dst.diff 7.4 KB

• You flagged this item as spam.
  

  Michael Curtis March 10th, 2009 @ 05:19 PM

  • Assigned user set to “Geoff Buesing”
• 

  Geoff Buesing March 11th, 2009 @ 02:30 AM

  Revised patch looks good. I'll pull this in after 2.3 final is pushed. Thanks!

• 

  Repository March 29th, 2009 @ 10:47 PM

  • State changed from “open” to “resolved”

  (from [5a8b481f717470b952ac7eb890f260ea98428153]) Time.local instances: Adding 24.hours across the DST boundary adds 24 hours instead of one day [#2066 state:resolved] http://github.com/rails/rails/co...

• 

  Jeremy Kemper October 15th, 2010 @ 11:01 PM

  • Milestone set to “3.0.2”
  • Importance changed from “” to “Low”

  [bulk edit]