Strange results with StrToFloat

[Gronque]

I had planned to use StrToFloat to convert the data received via UART but it is difficult to find the right format so I did some tests and got the following result:

Float.zip

(122.07 KiB) Downloaded 46 times

What am I doing wrong?

Thanks

[filip]

Hi,

Which version of the compiler are you using ?

Regards,
Filip.

[frank.malik]

Hi Gronque, hi Filip,

I'm not sure if this helps. I just made a quick check using the Necto Studio v.2.3.0 together with the microC for PIC32 and
the Conversions library. I had exactly the same results !

Looking into the code of the str_to_float function, the "E" will not be parsed in any case. Therefore, the last ten or so strings containing
the "E" will not work anyway. Might be an issue with the multiplier variable in the conversion function, but this is beyond my capabilities.

[frank.malik]

Hi,

sorry, my mistake. Both the multiplier and the divider using a int16_t variable. At the same time up to 10 positions before and after the point are
allowed. This means 10^10, this doesn't fit into a int16_t variable and the results are becoming strange.

Basically, this means, everything with a maximum of four digits before and after the point should work. Don't use "E".
If you use a minus value, the number before the point might be reduced to three.

[frank.malik]

Hi,

and again I overlooked something. The multiplier was ok already, and set to uint32_t resp uint64_t. The issue is only the divider.
After setting the divider to uint32_t it's possible to convert float-strings with up to 9 digits after the point correctly (at least as far as I can see).

Code: Select all

  FloatVar = str_to_float("-123.123");              // FloatVar is -1.231230E+002, ok
  FloatVar = str_to_float("0.000000001");           // FloatVar is 1.000000E-009, ok
  FloatVar = str_to_float("123456789.0123456789");  // FloatVar is 0.000000E+000, too many digits after the point
  FloatVar = str_to_float("123456789.123456789");   // FloatVar is 1.234568E+008, ok
  FloatVar = str_to_float("12345.6789");            // FloatVar is 1.234568E+004, ok
  FloatVar = str_to_float("0.0000000001");          // FloatVar is 0.000000E+000, too many digits after the point
  FloatVar = str_to_float("0.0000000012");          // FloatVar is 0.000000E+000, too many digits after the point
  FloatVar = str_to_float("0.000000012");           // FloatVar is 1.200000E-008, ok
  FloatVar = str_to_float("0.00000012");            // FloatVar is 1.200000E-007, ok
  FloatVar = str_to_float("0.0000012");             // FloatVar is 1.200000E-006, ok
  FloatVar = str_to_float("0.000012");              // FloatVar is 1.200000E-005, ok
  FloatVar = str_to_float("0.00012");               // FloatVar is 1.200000E-004, ok
  FloatVar = str_to_float("0.0012");                // FloatVar is 1.200000E-003, ok
  FloatVar = str_to_float("0.00123");               // FloatVar is 1.230000E-003, ok

[Gronque]

Hi,

Which version of the compiler are you using ?

Regards,
Filip.

Hi filip,

I use version 7.6.0

Regards,
Gronque