distance_of_time_in_words method accepts args in both directions

[avakhov]

Method distance_of_time_in_words date_helper.rb#L70 was designed to be symmetrical for arguments swapping. It should also work well for supporting time_ago_in_words date_helper.rb#L132

ae7d0d8 (Take leap years into account more seriously ...) added unsymmetry for long time periods. I propose to fix it.

PS. The condtion with leap years calculation (leap_years = (fyear > tyear) ? 0 : ...) is reasonable. Although from_time will be always less or equal than to_time. fyear can be greater than tyear for example at March 1st.

[sarahhodne]

What is the difference to doing it this way?

The tests pass for ruby 1.9.3-p0.

[avakhov]

It allows to calc leap years correct: date_helper.rb#L98-102

[sarahhodne]

I see. Could you throw in a test to check if leap years are calculated correctly as well, so we don't get this issue again later?

[avakhov]

@dvyjones, I'm sorry explain please that test case did you mean. I tested just argument swapping wich I suggest in the request. Without fix assert R130 fails. distance_of_time_in_words(end_date, start_date) returns 'about 28 years' instead 'almost 28 years' as in previous assert.

[sarahhodne]

@avakhov You said that the code didn't calculate leap years correctly. As far as I can see, we have no test to check if leap years are correctly calculated, so if you could add a test that basically does what you did when you found out that leap years aren't correctly calculated?

[avakhov]

I see. I formulated the idea not quite exact.

Leap years are calculcated correct, but only if from_time > to_time. Method distance_of_time_in_words should work if from_time > to_time and if to_time < from_time (it was true before ae7d0d8 and it's declared in documentation for method). Without my fix the condition (fyear > tyear) is true in case of from_time > to_time and false for back order of args. In the second case leap years discount will be ignored.

I want to return symmetry to this method back.