# Re: atomic

Date: Thu, 01 Nov 2007 13:23:22 -0300
Message-ID: <4729fd7d\$0\$14864\$9a566e8b_at_news.aliant.net>

paul c wrote:

> Bob Badour wrote:
>

>> David Cressey wrote:
>>
>>> "Bob Badour" <bbadour_at_pei.sympatico.ca> wrote in message
>>> news:4728f759\$0\$14832\$9a566e8b_at_news.aliant.net...
>>>
>>>> It's ironic you should mention base 9. It's a little-known fact
>>>> that, if
>>>> a number is evenly divisible by 8, the sum of the digits of the number
>>>> in base 9 is also evenly divisible by 8. And if one repeats the process
>>>> with that sum until the resulting sum has only one digit and if the
>>>> original number was divisible by 8, the one remaining digit will be 8.
>>>>
>>>> It's like magic. So it's really ironic you should mention base 9 in
>>>> particular.
>>>
>>>
>>> This is a joke, right? For radix n, n-1 has this interesting
>>> property.
>>
>>
>> Shhhhh! You've ruined it! Now nobody will bother converting numbers to
>> base 9 to see if it actually works. Spoilsport!
>>
>>
>>> Likewise, 9 has this fun property in decimals. Bookkeepers in the Bob
>>> Chratchit era used a property much like this one for error detection in
>>> manually managed numbers.
>>>
>>> If this were another person posting, I'd think it was simple oversight.
>>> But you, BB, are likely to know exactly what I've just said.
>>
>>
>> If the number is divisible by any factor of n-1, the sum of the digits
>> will be a multiple of that factor too. Thus, if one repeatedly sums
>> the digits of a hexadecimal number, one can tell very quickly whether
>> the number is divisible by 3, 5 or F.
>>
>> Base 9 is actually a little less useful because it can only tell you
>> if the number is divisible by 2, 4 or 8, and these are often
>> relatively obvious in other bases in any case.
>
>
> Heh, personally, I prefer to think of all the numbers in my db as base
> 10**64 or so with a floating decimal, even if there isn't a dbms that
> supports that.  Helps keep me from getting confused by more intricate
> representations.

If so, shouldn't that be 10**1000000 ? Received on Thu Nov 01 2007 - 17:23:22 CET

Original text of this message