WARNING: Apple Calculator jacked all to hell

cool!

People I have just upgraded my Mac from 10.3.7 to 10.3.8 and now I cannot start my calculator (Apple 3.0). I have the icon in the dock, I have the Calculator listed in Applications; however, when I click on either - NADA, ZIP, BUGGER-ALL. I went to the Repair Permissions, and reloaded 10.3.8 combo Upgrade - again the same. Any thoughts/fixes? A very new and inexperience convert from PCs. Jeff

Please Apple, bring back the old Calculator from the Classic Mac OS! Since the calculator market share recently has overtaken the Mac OS, we are doomed!! Nahhhh seriously, thi s is a bad looking bug they have there. But you know, a calculator is just one example of things that can be relatively easy to program, but really hard to test for bugs. They can't just try every calculation possible in beta testing...I'm not saying they shouldn't have looked this more or less common 0.1 bug, but I guess they don't teach this one in Math classes... At least it's not like the Windows calc bug which was actually due to a hardware bug in the Pentium II chip and probably required some software work arounds. This is getting common knowledge these days but did you know that Steve Jobs himself actually designed the original Calculator desk accessory layout that remained up through OS 9? That funny story,along with many very very interesting ones about the design of the original Mac/Lisa, can be read at http://www.folklore.org/StoryView.py?project=Macintosh&story=Calculator_Construction_Set.txt

Go to bugreporter.apple.com, and tell them you want the calculator app to use a decimal implementation. If you don't have a bug number, it's nothing but a complaint. Incidentally, the Cocoa framework includes an NSDecimalNumber class, which implements infinite-precision decimal arithmetic. -jcr

Simple fix is to go into "View | Precision..." and set it to "2" if you're working with typical currency calculations, or higher depending on how much accuracy you need. Do you REALLY need 12-digit precision? I think Apple's mistake was simply setting the default precision to "12." The Calculator.app works just like it should.

I believe Apple used double precision in their calculator program. IEEE double precision will start to lose precision after 16 significant digits (or was it 15). As someone suggested before, BCD coding will help, but unless Apple writes its own floating point routine, hardware BCD will still lose the precision (I'm not sure if the newer PPCs still have the option). This is true for PPC or x86. Most calculations including finite element analysis will do just fine with double precision. BTW, TI calculators (at least SR50/51/52) only kept 13 digits, which is 3 digits less than 16 digits, so changing the program to round off to 10 (or 13 like TI did) significant digits is just fine (note: rounding to 2 digits after decimal point may run into problem).

Seems like if you set the precision to something less than 12, then the problem goes away. I agree that setting it down to 2 is not a good idea, but setting to something like 10 or 8 seems very adequate.

10x= 9.9 repeating x = .9 repeating 10x-x = 9.99999etc - .999999etc 9x = 9 x=1 thus .99999 repeating = 1 apple should fix the display

You rarely see these errors on handheld calculators (or in programming languages where being right counts, COBOL for example) because they use Binary Coded Decimal instead of "lazy" floating point. I would say that setting the calculators precision to 2 is a bad idea because: 1. You will eventually accumulate errors at that precision, too. 2. A calculator (or any math program) should not behave this way, period. Apple should fix it.

The problem is common to all floating point math operations on any system -- from mainframes right down to $1.98 handheld calculators. It has to do with the fact that 0.1 is a good decimal number, but a failure as a binary floating point number because it cannot be represented, only approximated. Solution: Set your calculator's precision to 2 (The default is an unnecessarily high value of 12).