Java, Excel, and floating point:
Ok, Mr. Big One, I'll make my usual futile attempt to take this from the beginning. Two questions for the assembled multitudes:
1. Does Excel in general use floating point math? I thought financial things in general were done in fixed point. When last I looked at Excel, I couldn't find any obvious word one way or the other.
2. Is there anything different on copy protection in Windows 98, compared to previous Microsoft products? You have to have a key, but up through Office 97 (lastest Microsoft product I have), the keys were all interchangeable, I always used the first one I found. This was for Office 95, 97, OSR2 and NT4. Anything new and different?
Now, back to the start of my current little, uh, tiff.
www2.techstocks.com
Maybe it is not just me and my spreadsheet math???
techweb.com;
The floating point article is actually quite good, as near as I can tell. Talks about real issues in a clear manner. Directly as possible:
But the most complex problem, and the one Java Grande considers the most significant, concerns how the Java language specification mandates the handling of floating-point calculations. "The way the Java language standard is written today, Java says something to the effect that you have to get the same results on all platforms," Dongarra said. "That's going to have rather serious consequences if things are left that way."
Anybody who knows anything about floating point would agree with that. In general, programmers are told early on that you shouldn't expect exact results from floating point arithmetic, and that a comparison for equality between two floats is always a bad idea, unless you want the result false.
But, that brings us back to Reggie's snide little lead-in:
Maybe it is not just me and my spreadsheet math???
I can't explain the correct contextual meaning of that without having to "search throught the 30,000 some odd posts to pull two or three of [Reggie's] posts out of context", so you'll have to take my word for it that the main point of the Java/floating point article is about 180 decrees around from what was at issue in the sad historical discussion of fraudulent Java "high end financial math". That is, Java couldn't exactly reproduce Reggie's Excel results, so Java was a fraud. Oh well.
As to #2, that's a repeat of the question I asked in the note Reggie chose to respond to, www2.techstocks.com , 10394 here. Just a simple question, I don't know. The response didn't seem to have any relation to the question I could perceive. Maybe I was reading it out of context. Apologies in advance for polluting this august forum with these personal matters..
Cheers, Dan. |
