Re: [boost] date_time -> ptime

From: Joel de Guzman (djowel_at_[hidden])
Date: 2003-09-05 11:34:33

fred_at_[hidden] wrote:
>> From: Joel de Guzman [mailto:djowel_at_[hidden]]
>>> so I like the current
>>> interface.  I also greatly prefer the syntax of date(1970,1,1) to
> 1/Jan/1970.
> 
>> No problem, you can have either or both.
> 
> There are a large number of date standards out there. The only
> international one that I'm aware of is ISO 8601. (There might be more.
> I'm not a date expert.) ISO 8601 has several advantages over most
> national "standards".
> 
> 1) It's difficult to confuse with conventional American and European
> date formats.
> 
> 2) A lexigraphic sort also sorts the dates.
> 
> 3) It is a standard, so it's slightly easier to justify using than more
> culturally-specific formats.
> 
> So, my question is, why are things like 1/Feb/1970 allowed? I'd be
> happier if either date(19700201) or date(1970,2,1) were permitted.
> 1970/Feb/1 is marginally better than 1/Feb/1970 since it at least
> doesn't look like either the American format or the European format, but
> I don't see what's wrong with simply following ISO 8601 just to say that
> it's been followed.

In my ET implementation (no it's not part of date_time yet, AFAIK), 
I allow:  Y/M/D and M/D/Y only. 

Pardon the confusion, 1/Jan/1970 is indeed an illegal date (asserts)

-- 
Joel de Guzman
http://www.boost-consulting.com
http://spirit.sf.net