variable float problem raspberrypi4(os)

[alexfish]

Hi Peter

can you test these functions

here on pi I get different results(looking for whole numbers)

tail of double

Step : 23.6
counter  dist :24
Step : 24
Step : 24.4
Step : 24.8
counter  dist :25
Step : 25.2
Step : 25.6
counter  dist :26
Step : 26
Step : 26.4
Step : 26.8
counter  dist :27
Step : 27.2
Step : 27.6
counter  dist :28
Step : 28
Step : 28.4
Step : 28.8
counter  dist :29
Step : 29.2
Step : 29.6
counter  dist :30
Step : 30
Step : 30.4
Step : 30.8
counter  dist :31
Step : 31.2
Step : 31.6
counter  dist :32
Step : 32
Step : 32.4
Step : 32.8
counter  dist :33
Step : 33.2
Step : 33.6
counter  dist :34
Step : 34
Step : 34.4
Step : 34.8
counter  dist :35
Step : 35.2
Step : 35.6
counter  dist :36
Step : 36
Step : 36.4
Step : 36.8
counter  dist :37
Step : 37.2
Step : 37.6
counter  dist :38
Step : 38
Step : 38.4
Step : 38.8
counter  dist :39
Step : 39.2
Step : 39.6
counter  dist :40
Step : 40

float
counter  dist :29
Step : 29.2
Step : 29.6
counter  dist :30
Step : 30
Step : 30.4
Step : 30.8
counter  dist :31
Step : 31.2
Step : 31.6
counter  dist :32
Step : 32
Step : 32.4001
Step : 32.8001
Step : 33.2001
Step : 33.6001
Step : 34.0001
Step : 34.4001
Step : 34.8001
Step : 35.2001
Step : 35.6001
Step : 36.0001
Step : 36.4001
Step : 36.8001
Step : 37.2001
Step : 37.6001
Step : 38.0001
Step : 38.4001
Step : 38.8001
Step : 39.2001
Step : 39.6001
Step : 40.0001

BaCon Version Latest(4 days ago)
BaCon version 4.7.1 on Linux aarch64 - (c) Peter van Eerten - MIT License.


BR
Alex

[Pjot]

Hi Alex,

I am not sure which functions you are referring to?

Anyway, in C, the 'float' type and the 'double' type can behave differently. The 'float' type uses less bytes to store information.

PRINT SIZEOF(float)
PRINT SIZEOF(double)


Output on my laptop:

4
8


Because of float using less bytes for storage, the 'float' type is prone to rounding errors.

It sometimes can happen that a float variable containing "0.1" cannot be compared with a calculated float value of "1/10"; the comparison will return FALSE.

If, however, the variable would have been of 'double' type then it might succeed.

In all cases you have to be aware of such rounding errors, they happen in native C as well.

HTH
Peter