DTV 1.0b2 FTF Avatar
  1. OMG Issue

DTV — Confusing text for Gregorian calendar

  • Key: DTV-39
  • Legacy Issue Number: 17427
  • Status: closed  
  • Source: Thematix Partners LLC ( Mr. Edward J. Barkmeyer)
  • Summary:

    It is not clear what list the text bullets following the diagram in 12.3.1 are members of, and they are not sentences. Further, they have the pattern: "Gregorian year coordinate composed of a Gregorian year, for example 2010", but coordinates are defined to 'indicate' time points. They are not "composed of" time points in any sense. On the other hand, a 'G. year month coordinate' is composed of two time coordinates, but not two time points.

    The definition of 'Gregorian year coordinate' is: "absolute atomic time coordinate that indicates a Gregorian year that has the index equal to the index of the Gregorian year coordinate". The term 'Gregorian year coordinate' is being used in its own definition. The definition should read: "absolute atomic time coordinate that indicates a Gregorian year". (There are no other time coordinates that indicate Gregorian year time points.) There is a related Necessity: Each Gregorian year coordinate indicates the Gregorian year that has an index that is equal to the index of the Gregorian year coordinate. This pattern also applies to G. day of month, G. day of year coordinates, and hour, minute, second coordinates. It applies to numeric 'Gregorian month coordinates', but the time coordinate "January" does not have an index, per se. "January" is a term for the time point, but not an index (integer). The UML model (Figure 12.3) makes 'atomic time coordinate has index' a derived relationship, but that is false, given the intent: the index of the time coordinate is used to identify the time point, not taken from the identified time point. And in that case, the non-derived 'index' property is 0..1.

    The definition of 'Gregorian year month coordinate' misuses the verb concept 'compound time coordinate combines atomic time coordinate': "Definition: absolute compound time coordinate that combines the set of

    {a Gregorian year coordinate, a Gregorian month coordinate}

    and indicates the Gregorian month that..." A 'set' is not an 'atomic time coordinate' and cannot play that role. What is intended is:
    "Definition: absolute compound time coordinate that combines a Gregorian year coordinate and that combines a Gregorian month coordinate and that indicates a Gregorian month." That is sufficiently delimiting. The structural rule that determines which month it indicates can be stated as a separate Necessity. This "combines the set of" pattern is used in every compound time coordinate definition in 12.3.

    Note also that the formal statement of these definitions/necessities suffers from the lack of a basic arithmetic vocabulary (in Annex D?). Elsewhere DTV just says the arithmetic expressions are described in English or mathematical notations. If the arithmetic expressions are removed from the definitions, it becomes easier to do that. The mathematical formulations can be stated in CLIF and OCL.

  • Reported: DTV 1.0b1 — Thu, 14 Jun 2012 04:00 GMT
  • Disposition: Resolved — DTV 1.0b2
  • Disposition Summary:

    Clauses references in this second are to the beta-2 specification.

    The lists of Gregorian calendar, week calendar, and time of day time coordinates are revised to make them clearer and more formally linked to the glossary entries. The glossary entries themselves are revised along the lines suggested in the summary.

    ‘January’ is a term for a time point on the Gregorian year of months scale. That time point does have an index on that scale. Hence the construction ‘index of January’ is valid.

    The use of ordinary arithmetic, unstyled, is described in clause 5.1 and continues with this resolution.

  • Updated: Fri, 6 Mar 2015 20:58 GMT