(09-02-2014 04:13 PM)Joe Horn Wrote: [ -> ]Thoughts (which you already know but which bear repeating for οἱ πολλοί]): Unlike the other display modes, SCI and ENG modes are intended to show a given number of significant digits, and trailing zeros are significant digits (IF obtained from calculations based on that many significant digits, of course)...
I'll play devil's advocate here, I think the original poster has a valid point.
I always thought of scientific notation as a way to work with large numbers with a mantissa and an exponent in order to avoid showing a lot of digits, never heard of it being restricted to show a fixed amount of digits (when I work numbers by hand, I never write those zeros, does anyone??). Then engineering notation is the same, except the exponent is limited to multiples of 3 to make it more compatible with the use of units.
But the basic idea in my opinion is to avoid showing all the trailing zeros in large numbers.
I understand that you might prefer displaying a fixed amount of digits (not to be confused with significant digits, these are display digits only), but I think removing trailing zeros is a perfectly valid point too, and was actually the intent of the original idea.
Also let's not confuse display digits with significant digits. The former are just for show, while the latter is fixed either by hardware or by specification of the math library you are using.
If you have a number that results from the multiplication of two 8-digit numbers, it will have up to 16 significant digits, even if you choose to display only 6 (assuming the system precision is >=16, of course), so the relevance of those zeros being displayed is questionable. In the original example, (1.200000E3), assuming 16-digits for system precision, you have 14 zeros. Why is showing 5 any better than showing none? Neither case represents the number of significant digits that you have in storage.
As far as the significance of the digits due to engineering uncertainties in the input variables, methods, etc., I don't think displaying a fixed number of digits will help anything. Only the engineer will know how close his result will be in the end.
I think we should let the machine display the numbers in a way that is easy to read for the user, and let the user judge the accuracy of the result.
So count me in as one of the few, I'd rather see 1.2e3 than 1.200000e3, nothing wrong with that.
Claudio