Business days calculation broken in 0.7.5

[bmulholland]

0.7.4

2.2.3 :004 > Time.now.in_time_zone("America/Los_Angeles")
 => Thu, 07 Apr 2016 12:35:15 PDT -07:00
2.2.3 :005 > 2.business_days.from_now.in_time_zone("America/Los_Angeles")
 => Mon, 11 Apr 2016 12:35:19 PDT -07:00

0.7.5

2.2.3 :004 > Time.now.in_time_zone("America/Los_Angeles")
 => Thu, 07 Apr 2016 12:35:00 PDT -07:00
2.2.3 :005 > 2.business_days.from_now.in_time_zone("America/Los_Angeles")
 => Tue, 12 Apr 2016 02:00:00 PDT -07:00

[bmulholland]

Likewise

0.7.4

2.2.3 :006 > 2.business_days.ago.in_time_zone("America/Los_Angeles")
 => Tue, 05 Apr 2016 12:37:26 PDT -07:00

0.7.5

2.2.3 :006 > 2.business_days.ago.in_time_zone("America/Los_Angeles")
 => Tue, 05 Apr 2016 02:00:00 PDT -07:00

[muzfuz]

From lib/business_time/business_time.rb

      if time.is_a?(Time) || time.is_a?(DateTime)
        unless time.during_business_hours?
          time = Time.beginning_of_workday(Time.roll_backward(time))
        end
      end

Looks to me to be done on purpose.

It seems to be one of the biggest changes is around handling of business days now being defaulted to 9am to 5pm, instead of 12am to 11:59pm.

Make sure you generate your yaml config file (using generator in the README), and change the settings there if you were relying on anything like that.

[bmulholland]

That's a pretty big change to introduce in a point version :(

[muzfuz]

Agreed. I think this should have followed semantic versioning. Lack of a changelog is problematic as well.

[ladelfa]

Ditto. If it's 2:25 pm on Wednesday, I would expect 1.business_day.ago to be 2:25 pm on Tuesday, not the beginning of the business day on Tuesday, whatever that's set to be.

[bokmann]

this was an accident. I'm looking into it now. Accepting multiple pull requests, dealing with monkeypatches on top of ActiveSupport Monkeypatches becomes a lot to deal with. Future versions will have a change log.

[bokmann]

I am not seeing the issue you describe above. Here I am with your test cases:

Time.now
  2016-06-09 10:37:42 -0400
2.business_days.ago
  2016-06-07 10:37:58 -0400
2.business_days.ago.in_time_zone("America/Los_Angeles")
  Tue, 07 Jun 2016 07:38:22 PDT -07:00
Time.now.in_time_zone("America/Los_Angeles")
  Thu, 09 Jun 2016 07:38:51 PDT -07:00
2.business_days.from_now.in_time_zone("America/Los_Angeles")
  Mon, 13 Jun 2016 07:39:03 PDT -07:00

Can you tell me:

• The Activesupport version you are using?
• The version of your TZInfo Gem?
• What version of Ruby you are using?

Or even better, a failing test case added to the tests?

[ladelfa]

This issue may have an interaction with the stub method of rspec 2.x and this way of writing time-deterministic tests :

Time.stub(now: Time.utc(2014, 6, 3, 10, 10))
=> {:now=>2014-06-03 10:10:00 UTC}
1.business_day.ago
=> Mon, 02 Jun 2014 06:10:00 EDT -04:00

Time.stub(now: Time.new(2014, 6, 3, 10, 10))
=> {:now=>2014-06-03 10:10:00 -0700}
1.business_day.ago
=> Mon, 02 Jun 2014 00:00:00 EDT -04:00

Under 0.7.4, stubbing with utc and new give identical responses:

Time.stub(now: Time.utc(2014, 6, 3, 10, 10))
=> {:now=>2014-06-03 10:10:00 UTC}
1.business_day.ago
=> Mon, 02 Jun 2014 06:10:00 EDT -04:00

Time.stub(now: Time.new(2014, 6, 3, 10, 10))
=> {:now=>2014-06-03 10:10:00 -0700}
1.business_day.ago
=> Mon, 02 Jun 2014 13:10:00 EDT -04:00