> I read on the date_time change history about a new function for > calculating ISO 8601 week number. > > I should note that this week number is rather useless without > the corresponding year number. ISO 8601 week-based year is not > always the same as the actual year. For example, 2nd January 1999 > is week 53 of year 1998 (not 1999!). 30th December 1997 is week > 01 of year 1998 (not 1997!). (examples taken from the C99 standard) > > My experience is that people often forget about this important fact > and thus introduce bugs that may be not be catched until such situation > really happens in life. To force library users consider this thing, > I think that gregorian::date::week_number should return a pair: week > number and week-based year number. > > Another option would be to just add a separate function for calculating > week-based year number, however in my opinion it is less desirable > because of reasons mentined above. In any case, providing week number > without providing week-based year number is undesirable.
You make a good point that this can be quite confusing. My thought would be to add an iso_year() function and some documentation instead of changing the current function. Jeff _______________________________________________ Unsubscribe & other changes: http://lists.boost.org/mailman/listinfo.cgi/boost
