Re: [calsify] Subtraction vs. DST disambiguation

[Michael H Deckers <[e-mail]>]

    On 2020-09-30 21:50, Justin Grant wrote:

> We previously
> agreed that, in order to make subtraction reversible with subsequent
> addition of the same value, we'd reverse the order of operations so with
> subtraction of a positive duration (or adding a negative duration) the time
> portion would be subtracted first, and then the date portion.



     If I understand you correctly, you want to ensure that
          ( Δ - 1 month) + 1 month = Δ
     for any date Δ.

     This is impossible to achieve with the "usual" expectations,
     for instance, that ( Δ - 1 month) is a date in February
     when Δ is a date in March (because you have only 28 or 29
     different dates in February, but 31 dates in March, so that
     such a subtraction of 1 month from dates in March can never
     be injective).

     Of course, there are group operations on the set of dates,
     such as adding and subtracting integral multiples of days
     or weeks.

     And on the set of datetimes (= date plus time of day), there
     are group operations with multiples of formal time units such as
     months so that the subtraction of 1 month maps the set of
     datetimes in the interval
         [2020-03-01T00:00:00..2020-04-01T00:00:00]
     bijectively onto the interval
         [2020-02-01T00:00:00..2020-03-01T00:00:00]
     (eg, as in the subtraction of fractions of months in Oracle SQL);
     but the result of such a subtraction from a date (with time of day
     00:00:00) is in general no longer a date. Do you intend to
     specify such an operation?

     HTH.

     Michael Deckers.