Fixed-Point (FXP) configuration needs to show more decimal digits

[altenbach]

When configuring FXP datatype, the range, max, and delta are shown with a resolution of only four decimal digits. This is not enough!

(For example, look at a type with 12 fractional bits. While the delta is 0.000244141..., it shows up as 0.0002, which is off by over 20% and could give the false impression that we have a nice round decimal delta)

Silly enough, the indicator field is 60% blank. There would be plenty of room to show more information and the indicator could even be enlarged and still fit inside the box.

The insert in the figure shows how it would look like with a display format of 6 significant digits. Seems to be much more realistic!

IDEA: Show more significant digits for the range indicators of FXP configuration!

In addition, the current "desired delta" indicator should be renamed to "Delta".

It is an indicator, so we cannot enter anything we desire anyway!

The delta is a result of the binary limitation of the configured bits. It is what it is! The word "desired" might raise wrong expectations! 🙂

[SteveChandler]

There needs to always be some indication that not everything is shown in an indicator. It can lead to all kinds of confusion.

[X.]

Not only that, but I would argue that there should be a way to modify these settings AT RUNTIME (using a contextual menu) for any type of numeric control... You can do that for graph axes, so I don't think the (LabVIEW) Bible prevents it.

This being said, I never used FXP so why not give a kudo for Christmas...

[JLewis]

I agree with your formatting suggestion, but am curious as to how you use the information in that indicator? I have never found the "delta" information very useful (beyond order of magnitude) because I'm typically more interested in the number of fractional bits. I had been toying with the idea of suggesting that we change the context help format to make that information more explicit by adding a binary format display, something like 2.44E-4 (2^-12). Would you miss the extra 3 significant digits in that scenario?

The "desired delta" is an artifact from early plans to make those fields editable, as an alternate way of specifying the type. That came with a whole set of complications and tends to encourage inefficient hardware implementations so there are no plans to implement that functionality at the moment. I created CAR 330636 to rename the indicator as you suggested.

[altenbach]

I actually like the delta display, because it is informative. I am typcaily interested about the smallest step size between numbers.

The problem is that the format is a fixed number of decimal digits, so in the described scenario we only get a single significant digit! A display of  2.44E-4 (3 significant digits) is definitely much better than 0.0002 (1 significant digit).

For even smaller deltas (>14 fractional bits) we actually suddenly get E format with 6 significant digits. The problem is most serious for 13 fractional bits and gest better again for smaller differences.

Thanks for the CAR#. 🙂

[G-Money]

Available in LabVIEW 2012