WString: add toDouble

[Ivan-Perez]

String::toFloat internally converts the string into a double (using atof) and then truncates into a
float, so why not to add a method to return the double?

[sandeepmistry]

I've been trying this out today. One observation, on AVR:

sizeof(float) = 4
sizeof(double) = 4

So both floats and doubles are 32-bits. So, looks good to merge on AVR.

Will test non-AVR cores (SAM, SAMD) shortly, since we would want to port the new API to those as well.

[sandeepmistry]

@Ivan-Perez were there some particular test sketches or values you used to determine that String::toDouble() was needed?

[Ivan-Perez]

I didn't know that both float and double types were the same size. I supposed that double might be 64-bit, but I've just checked it and as you say they are the same type (reference also confirms this).

Then maybe this PR is not useful at all and can be closed without merging.

[Chris--A]

@Ivan-Perez

Then maybe this PR is not useful at all and can be closed without merging.

Well it'll be useful for the 32-bit platforms that support a double that is larger than a float.

[sandeepmistry]

@Ivan-Perez thank you!

@akash73 could you please add the new toDouble function to the String Object reference.

[sandeepmistry]

This change has been ported to the SAMD core in arduino/ArduinoCore-samd@c7c6f70, and I've also opened a pull request for the 101 core: arduino/ArduinoCore-arc32#293.

[Ivan-Perez]

Reviewing the code once again, I think that in case of failure both functions (toFloat and toDouble) should return NaN instead of 0.

In the next days I'll create a new PR to address it.

[q2dg]

Well...data types in Arduino deserve a coffe break...
#4525
#3801
#5200
...