Normalise the use of float and double data types throughout the code base

[v-barshaw]

Hi,

Currently float and double data types are used more or less randomly, which leads to unnecessary code bloat -- just search AxxSolder_firmware/Debug/AxxSolder.list file for functions __adddf3, __aeabi_dmul, __aeabi_ddiv, and friends.

Is the use of double data type justified? Does it add anything in terms of accuracy or performance? I very much doubt it.

Also, there is a bit of a mess in dealing with floating-point variables:

• assigning double to float : sensor_values.bus_voltage = Moving_Average_Compute()
• using float instead if integer : float WindowLength; can WindowLength be 6.8762345584903651 ?
• mixing unsigned integers with floats with signed integers : if(WindowPointer < WindowLength - 1)
• and so on

The chosen MCU has hardware floating point unit, so let's use it to the full.

May I suggest a global rework to put this stuff in order? I could volunteer for this.

[AxxAxx]

Hi!

I agree, the code base could need an overhaul when it comes to chosen types, efficiency and code cleanliness in general.
I come from the mechanical side and it would be great to have someone with code-eyes to have a look at the code.

I created a dev branch where I'll merge your current update to and test on the AxxSolder Hardware.

Do you have hardware to test on?

[v-barshaw]

Hi,

No, I don't have the hardware. Can do without it right now, but will need when I start to make the real clean-up.

I have reformatted the code (4 headers and 4 c files) with uncrustify, to get better understanding of the code. Could not commit to dev branch, don't have write access.