Float, Double, to bytes?

[Florian]

I am currently using this way to cast a double or float to an array of bytes (uint8_t) and back:

To bytes:

double data = 5.5;

std::queue<uint8_t> ndata;
uint8_t* tmp = (uint8_t*)&data;
    
for (int i = 0; i < sizeof(double); i++)
    ndata.push(tmp[i]);
    
return ndata;

And back:

double data = 0.0;

std::queue<uint8_t> ndata;
uint8_t* tmp = (uint8_t*)&data;

for (int i = 0; i < sizeof(double); i++)
    tmp[i] = ndata.front(), ndata.pop();

return ndata;

This is butt ugly code though, is there any other way to do this properly in a C++ way (or with Boost)? The resulting bytes will be eventually sent over a network connection (loopback interface actually).

I am working with GCC on Linux x64.

[Elysia]

Do you need to store it in a queue, though?
Isn't a pointer enough?

[cyberfish]

The resulting bytes will be eventually sent over a network connection

Careful.

Endianness - Wikipedia, the free encyclopedia

[Florian]

Do you need to store it in a queue, though?
Isn't a pointer enough?

Various other kinds of object and types will be stored in the queue, and then together send across the network.

Careful.

Endianness - Wikipedia, the free encyclopedia

I am aware of this. i have solved this problem with integers, but i do not know how to solve this with floats or doubles, since you can't use bit shifting on them.

The transmitted data will stay on the same machine, but i don't think that's a valid reason to forget endianness.

[Salem]

Careful.
Not everyone uses IEEE754
Survey of Floating-Point Formats at MROB

[SyntaxError]

One thing you have to consider is some machines are little-endian and other's are big-endian. I think that even goes for floats and doubles too. This means that if you move your bite array from one machine to another and reconstruct it, the number will be wrong because the bytes will be reversed. That's not to mention the fact that not all machines use the same floating point format. However that's getting to be less of a problem these days with IEEE standards and such. If I recall there used to be a few machines that didn't even use two's compliment for integers, but this isn't something to worry about these days.