Re: a union is always a join!

From: Walter Mitty <wamitty_at_verizon.net>
Date: Sun, 15 Mar 2009 13:52:57 GMT
Message-ID: <ZK7vl.236$6%.106_at_nwrddc01.gnilink.net>


"paul c" <toledobythesea_at_oohay.ac> wrote in message news:pjZsl.15961$Db2.2529_at_edtnps83...
> paul c wrote:
> ...
>> Infinity is not a problem because when we implement on a digital machine,
>> all stored values are finite. ...
>
> Maybe I should have qualified 'values' to be ones that aren't allowed to
> be treated in an approximate way by the operators allowed, such as many
> floating point value representations.

Actually, the floating point arithmetic treats values as producing an exact result under all the operators involved. At least, so it seems to me from looking at the FP operators of several computers. The problem is that the exact result isn't "exactly right" in terms of what we construe exact rightness to mean. In most cases, that is because what we give the arithmetic unit as inputs are only approximations to the real values we would like to give.   Received on Sun Mar 15 2009 - 14:52:57 CET

Original text of this message