How to define the number of decimal places in the settings for a Float?

mcinner1 5 days ago

Hi there!

I'm developing a datafield where the user can set their Efficiency Threshold in the settings as Float with precicion=2.

When - for example - set 0.80 then it's accepted and stored. But when I visit the settings afterwards, the value is shown as float with many decimal places which do not exactly match the given value.

Can I set the precision for settings floats?

Here are the basics in my code:

And here is a screenshot of my Connect IQ mobile app:

jim_m_58 5 days ago

32 bit floats are kind of odd. in that not all values can be saved in 32 bits so some may be displayed a bit oddly

flocsy🤠 4 days ago

either use fixed points (i.e use Number, and save the value *100) or save it as a string. Though I suspect that if you eventually convert it to Float, then you'll probably have the same problem. You could maybe do the whole math in fixed point though.

flowstate 4 days ago in reply to jim_m_58

jim_m_58 said:
32 bit floats are kind of odd. in that not all values can be saved in 32 bits so some may be displayed a bit oddly

This isn't unique to 32-bit floats at all, but it's a characteristic of all floating-point numbers, including 64-bit double-precision floating-point numbers (Double in Monkey C)..

For example, js uses 64-bit floats for its Number data type:

> 10.8/100
0.10800000000000001

It would be fair to say that the problem is even more pronounced with 32-bit floats vs 64-bit floats though.

mcinner1 4 days ago in reply to flowstate

Ok, thank you all.

That means: there is no kind of tag for xml settings like < precision = "2" /> or similar…

flocsy🤠 4 days ago in reply to mcinner1

Precision is a display thing. not a storage thing.

Float f;

f.format("%.2f");

is maybe what you're after.

mcinner1 4 days ago in reply to flocsy🤠

No, it‘s all about displaying! I have no problem with the internal float. Does not matter if exact or with some 0,0000123. Does not matter for my use.

It‘s all about the user not seeing such an odd figure in the settings.

I know - I could use a string. But then I have no control what the user types in and I have to evaluate if their input gives a valid float before using xxxx.toFloat()

ad220 3 days ago in reply to mcinner1

Let them put an integer and set the unit to EF in percent maybe ? Then divide the property by a hundred in the code.

Would that fit your use case ?

mcinner1 3 days ago in reply to ad220

Thank you for your input, this would be possible, but I don‘t want to force users to a workaround.

Maybe I go the string way…