[pmwiki-users] Conditional Directive based on time

[Patrick R. Michaud]

On Wed, Oct 12, 2005 at 10:33:26AM -0400, Neil Herber wrote:
> >Keeping this theme, then ("through" means "including" here):
> >
> >    (:if date 200501-200503 :)      2005.01.01 through 2005.03.31
> >    (:if date 20051006-2006 :)      2005.10.06 through 2006.12.31
> >    (:if date 20051006-200601 :)    2006.10.06 through 2006.01.31
> >    (:if date 20051006-20060101 :)  2006.10.06 through 2006.01.01
> >    (:if date 20051006-2005 :)      2005.10.06 through 2005.12.31
> 
> Some truncated dates appear ambiguous.  For example, I could interpret:
> (:if date 20051006-2006 :)
> as either 2005.10.06 to 2006.01.01  -or-   2005.10.06 to 2006.12.31

...or even 2005.10.06 through 2005.12.31, if one thinks of "-2006"
as meaning "strictly before 2006".

> And it somehow looks counter-intuitive that a shorter string like:
> (:if date 20051006-2006 :)  specifies a *longer* period than (:if 
> date 20051006-20060202 :)

Hey, you guys were the ones that wanted the ending date to include
the last day.  :-) :-) In my original formulation, the ending date 
would've always meant "strictly before", thus

    (:if date 20051006-2006:)      2005.10.06 through 2005.12.31
    (:if date 20051006-200601:)    same thing
    (:if date 20051006-20060101:)  same thing
    (:if date 20051006-200602:)    2005.10.06 through 2005.01.31
    (:if date 20051006-20060201:)  same thing

which matches your intuition of "longer strings are longer periods",
but tends to be counter-intuitive.

But even in English, the shorter string often corresponds to the
"longer period".  For example, if I say that "movie XYZ will be
in theaters from now through 2006", people understand this to be
a longer period than if I had said "movie XYZ will be in theaters
from now through August 15, 2006".

> I know somebody can learn the rules, but why not avoid many of the 
> rules by always requiring a complete date?

OTOH, one can always use complete dates and never have to learn the
rules.  Truthfully, I don't expect the mixed-date formats to be all that 
common; i.e., authors would typically limit themselves to
the forms where there's not really a lot of ambiguity, such as

    (:if date 2005-2007:)
    (:if date 200503-200602:)
    (:if date 20050315-20060531:)

In fact, if I hadn't mentioned the mixed formats in my previous
post, I doubt many would've known they were possible.  Ultimately,
it turns out to be a little easier for me to write code that
handles the mixed date formats than it is to try to restrict them,
and I tend to go with the simplest solutions first.

Pm