From: Colin Bartlett <colinb2r@...>
Date: 2009-12-24T19:25:06+09:00
Subject: Re: Date.gregorian_leap? code

On Thu, Dec 24, 2009 at 3:20 AM, Phrogz <phrogz@mac.com> wrote:
> On Dec 23, 2:42 pm, Colin Bartlett <colin...@googlemail.com> wrote:
>> Because of the precedence of && and || this is presumably parsed as
>>      ( y % 4 == 0 && y % 100 != 0 ) || y % 400 == 0
>> and I'm trying to think of reasons why it is coded as it is instead of, say:
>>      y % 4 == 0 && ( y % 100 != 0 || y % 400 == 0 )
>
> I suppose because either one works: ...

It's true that both variations give the same result. But using y = 2003:
  y % 4 == 0 && y % 100 != 0 || y % 400 == 0
evaluates y % 4 == 0 as false and then (as far as I can see)
also evaluates y % 400 == 0 as false,
and what I'm not seeing is why one might do that:
if something isn't exactly divisible by 4 it won't be divisible by 400.
Hence my bracketed suggestion which stops at y% 4 == 0
for years which aren't divisible by 4.

The code in date.rb isn't wrong, because it gives the correct answers,
and won't take much extra processing time to (from my point
of view) "unnecessarily" evaluate y% 400 == 0 even if y % 4 != 0,
so this is admittedly a (very?) minor point, but it's bugging me
that I can't think of a reason why one wouldn't put the brackets in
to avoid evaluating y % 400 == 0 in cases where the calculation
already knows that y% 4 != 0