From: Yitzhak Sapir (yitzhaks_at_[hidden])
Date: 2003-01-22 05:22:28
> From: Jeff Garland [mailto:jeff_at_[hidden]]
>
> > Hmmm while I can see your point, I still think a default
> constructor should
> > be provided. And as I, along with all of the users in the
> messages you cited
> > seemed to have expected that default constructed dates
> would be set to
> > 'not_a_date_time', I'd suggest that this would be the most
> sensible default
> > value. I don't see any point in *not* providing a default value.
>
> Keeping the interface to a minimum, preventing
> accidental/surprising values,
> avoiding the controversy of discussing what an appropriate value for
> the default constructor is. Well, 2 out of 3 anyway :-)
>
> But seriously I'm willing to add it, but I don't think I've heard
> a compelling use case yet...
I wrote a date formatting function. This function can take a datetime,
date, or time duration, and format it to text. The core however, is
implemented using datetime_t. The function implements the format
as described in Microsoft's documentation:
http://msdn.microsoft.com/library/default.asp?url=/library/en-us/vbaac10/html/acproFormatDate.asp
However, to implement format "General Date" (format letter "c"), I
need to know if the given input contains a date, a time, or a
datetime. Since the functions pretty much perform the same
thing, I have a core that implements functionality for datetime and
then I use this in the function that implements date/time separately.
However, to implement the no time I have to query whether the time
is 0:0:0.000. This is obviously problematic, since this is a
perfectly valid time. For the date, I don't even have a "start date" to
use. So I pass the core function a bool that tells it whether to ignore
the date, and pass in an arbitrary date when I want just the time.
It would be nicer, though, if I could just pass "not_a_date" or
"not_a_time". I could even expect the datetime function to throw
an exception if the requested format is for "month" and the
"not_a_date" was passed.
There is also a problem in serialization. A lot of serialization
systems first construct an empty struct, and then read in the
values. Doing it this way prevents serialization in those
systems of dates.