Hi Robert,

Friday, June 8, 2007, 7:47:17 PM, you wrote:

> On Fri, 2007-06-08 at 19:35 +0100, Richard Davey wrote:
>> Hi Robert,
>> Friday, June 8, 2007, 7:21:39 PM, you wrote:
>> > Precisely defined limits are not the same as precisely defined values. I
>> > might precisely define the amount of entropy on a random value as being
>> > some formula based on the current temperature of my CPU. The formula is
>> > quite precise, but the value is only precisely known after I take the
>> > temperature... and even then it probably doesn't serve me to know it
>> > explicitly unless I'm verifying results. If I then take that random
>> > value and use it to make other decisions then the end result isn't
>> > exactly precise.
>> The end result isn't "precise", but it will always exist between some
>> very real and very price possible values, those controlled by the min
>> and max temp of your CPU.

> ... and the resolution at which the temperature can be measured. But we
> digress, this is merely a specific example, there are many more metrics
> that can be sampled to give an extremely large range of values.

And I'd wager those values too have precise minimums and maximums. You
can pluck an "approximate" value from between them however you like,
but I'm still waiting for evidence that that value will never lie
between some very tangible limits, and still be of any kind of
real-world value in your code. Due to the very nature and way in which
CPUs work, and the effect this passes up through to any language that
cares sit on it, at the end of the day if the architecture itself
cannot deal with 'approximate' values, anything that happens on the
higher level will always be reduced to a value that exists between two
very solidly set limits. If you don't impose those in your code, what
happens beneath will impose it for you regardless. I'd be happy to
read some papers or code that you feel demonstrates what you're trying
to get across if you have any links.


Zend Certified Engineer

"Never trust a computer you can't throw out of a window"

PHP General Mailing List (http://www.php.net/)
To unsubscribe, visit: http://www.php.net/unsub.php

Reply via email to