Compiler/MSP430FR5994: What use union?

update

if union write as

union long_as_array{
	unsigned char a[4];
	unsigned long l;
};

and load to l=0x01020304
in a array [0x00,0x04,0x00,0x00]

strange...

thank for your ansver/

I write test clear code

Variant 1

#include <msp430.h>

union long_as_array{
	unsigned char a[4];
	unsigned long l;
};

unsigned char pos;
unsigned char BUFF[20]={0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19};
unsigned int exp;
unsigned char neg;

void IEEE2Long(unsigned char* buff, unsigned char pos){
	union long_as_array IEEE754;
	IEEE754.l=*(unsigned long*)&buff[pos];
	neg=IEEE754.a[0] & BIT7?1:0;
	exp=(IEEE754.a[0]<<1)+(IEEE754.a[1] & BIT7?1:0);
}

int main(void) {
    WDTCTL = WDTPW | WDTHOLD;	// Stop watchdog timer
    IEEE2Long(BUFF,12);
	while(1);
}

Strange result in view

I move union variablle to global

#include <msp430.h>

union long_as_array{
	unsigned char a[4];
	unsigned long l;
};

unsigned char pos;
unsigned char BUFF[20]={0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19};
unsigned int exp;
unsigned char neg;
union long_as_array IEEE754;

void IEEE2Long(unsigned char* buff, unsigned char pos){
	IEEE754.l=*(unsigned long*)&buff[pos];
	neg=IEEE754.a[0] & BIT7?1:0;
	exp=(IEEE754.a[0]<<1)+(IEEE754.a[1] & BIT7?1:0);
}

int main(void) {
    WDTCTL = WDTPW | WDTHOLD;	// Stop watchdog timer
    IEEE2Long(BUFF,12);
	while(1);
}

and it work!

Thanks, found this in

_defs.h

math_intr_access.h

math_private.h

but I do not need the full support of the whole range float. I operation on each byte in send me float from sensor

My sensor out data in percentages and temperature in range -100 to 100

If load standart float lib - lost many memory and very slow operation.

I invented a trick :) in float range 128 to -128 with accuracy 10E-5 (real use 10E-4)

use only bit shift and 1 divider at the end calculate and transfer float to long for operation

for example

98.7654321 = 9876543

0.001234=123

further operate with long

Okay, then you might want to consider using fixed-point arithmetic (e.g. Q16). This lets you use the integral operations but still have fractional values.

thanks

I understood

Union work only in global scope!