Re: [gobo-eiffel-develop] Date parsing routines, etc. committed

[Eric B. <er...@go...>]

Colin Paul Adams wrote:
>>>>>>"Eric" == Eric Bezault <er...@go...> writes:
>>>>>
> 
>     Eric> Week days: ----------
> 
>     Eric> In DT_GREGORIAN_CALENDAR, is `is_week' a good name? Wouldn't
>     Eric> it be better to have `is_valid_week'? (I made no
>     Eric> modification about this name in CVS.)
> 
> I'm not sure about the names.
> DT_ISO_DATE_TIME_PARSER also has a routine which in turn calls this
> one (the pre-conditions are different). So I called it is_valid_week
> in there.

What about `is_valid_week_number'? Or `valid_week' (we already
have this scheme: `leap_year' in KL_GREGORIAN_CALENDAR and
`is_leap_year' in DT_DATE)?

>     Eric> Finally, I don't understand why there is a precondition
>     Eric> 'a_year /= 0'.
> 
> That was a hangover from when I was under a mis-aprehension of how
> Gobo was representing BCE dates.

Note that I added the following line in the indexing clause
of class DT_DATE:

   note: "year 0 means 1 BCE, year -1 means 2 BCE, etc."

>     Eric> Speaking about the preconditions of the routines `is_...'
>     Eric> in class DT_DATE_TIME_PARSER, I think that they are wrong.
>     Eric> Indeed, instead of:
> 
>     Eric>     is_date (a_formatted_date: STRING): BOOLEAN is require
>     Eric> formatted_date_not_void: a_formatted_date /= Void zoneless:
>     Eric> not is_zoned_date (a_formatted_date)
> 
>     Eric> I would have expected:
> 
>     Eric>     is_date (a_formatted_date: STRING): BOOLEAN is require
>     Eric> formatted_date_not_void: a_formatted_date /= Void ensure
>     Eric> zoneless: Result implies not is_zoned_date
>     Eric> (a_formatted_date)
> 
> The idea is that is_zoned_date can be implemented by checking for a
> zone at the end of the string, and if it is present and correct,
> passing the string minus the zone to is_date.

I agree with that, and this is how it should be implemented.
But I still think that we should be able to check whether a
string represents a zoneless date without having to call
`is_zoned_date' first. To what extent would the specification
that I suggested affect the implementation?

-- 
Eric Bezault
mailto:er...@go...
http://www.gobosoft.com