>>>>>Yes but your start time in this example is 2007 ... 3 years ago...
>>>>
>>>>Yes, and based on those numbers, it doesn't make any sense that we would have five years and 3 months completed since August 5th.
>>>>
>>>>Do you have any idea why?
>>>
>>>That is due to the nature of a DateTime struct
>>>
>>>with a new DateTime(0), ie 0 ticks, the year = 1, month is 1 and day is one. iow it's a date and not a difference
>>>
>>>You may get away with subtracting one from year, month, and day, but I don't like it, for the following reasons
>>>
>>>(1) Leapyears:
>>>

>>>DateTime(2013, 01, 1, 00, 00, 0) - DateTime(2012, 01, 1, 00, 00, 0)
>>>

>>>gives that extra day. Converting that timespan to a Datetime() yields years = 2, months = 1 and days = 2
>>>
>>>but
>>>

>>>DateTime(2012, 01, 1, 00, 00, 0) - DateTime(2011, 01, 1, 00, 00, 0)
>>>

>>> yields years = 2, months = 1 and days = 1
>>>
>>>
>>>(2) months shorter than 31 days
>>>

>>>DateTime(2011, 02, 1, 00, 00, 0) - DateTime(2011, 01, 1, 00, 00, 0)
>>>

>>>
>>>yields years = 1, months = 2 and days = 1
>>>
>>>
>>>
>>>

>>>DateTime(2011, 03, 1, 00, 00, 0) - DateTime(2011, 02, 1, 00, 00, 0)
>>>

>>>
>>>yields years = 1, months = 1 and days = 29
>>>
>>>In other words, you are adding a timeSpan to DateTime(01, 01, 1, 00, 00, 0)
>>>A timespan is a number of ticks, or say seconds
>>>
>>>The first month of year 0001 has 31 days for a complete months. But the timespan is only a timespan ,ie seconds elapsed
>>>
>>>
>>>
>>>
>>>Also, the number of days in the months of year 0001 is 31, 28, 31, 30, 31 etc
>>>Who says that your timespan is following the same sequence ( say 2011 may 1 - 2011 april 1 is 30 days AND a full month)
>>
>>Doh. Nice little bit of drive by coding (five minutes to spare before interviewing and no check online) and you have to blow it out of the water. I didn't even know that new DateTime(new TimeSpan(0).Ticks) would result in 01/01/01 :-}
>
>Sorry to have killed your idea - but a DateTime() is just a date and a time. It gives a point in time
>
>>But I wonder whether just working with the Timespan would be better. For example an 'accurate' calculation for two dates that were 90 days apart might give a result of 2 months 29 days, 3 months exactly or 3 months and 1 day depending on which three months fell into the span. Calling a timespan of 90 days three different things is, to me, just confusing.....
>
>>But I wonder whether just working with the Timespan would be better.
>Not in this context.

I'm not so sure. As above it may depend on what exactly Michel is trying to achieve?