Programming Forums > Application Development > C

sDa · Feb 21st, 2007, 12:57 PM

My problem is this:

I have a float that I want to send over... well never mind that's not important... but the thing is I can only send one byte at the time and as you know a float is 4 bytes so how do I do that w/o loosing precision?

oh and I know this isn't really a C problem but since I'm writing in C I placed it here...

andro · Feb 21st, 2007, 1:27 PM

Send a byte designated to mean a float is incoming, and then send the 4 bytes of the float? Whatever is receiving will have to know how to handle it.

DaWei · Feb 21st, 2007, 1:30 PM

Hook the bytes together on the other end. If the 'other end' is a different platform, you'll have to get the order right. If it can't deal with 4-byte entities, or it uses a different floating-point procedure, your goose is cooked.

sDa · Feb 21st, 2007, 1:33 PM

I think I will have to explain it a bit more... I am sending the float over an I2C bus and what you do to send it is you send 1byte at the time so for instance if I want to send a 16bit int I first send var&0x00ff then (var&ff00)>>8 (I think

)

but how do I do something similar with a float? There is that "," that I really don't know what to do with

DaWei · Feb 21st, 2007, 2:18 PM

A float is just another binary number. It's just interpreted differently. Extend the principle you're currently using.

Game_Ender · Feb 21st, 2007, 7:37 PM

And make sure your target device understands/uses the same floating point format as the other.

On another note the easier way to do this would be something like this (just a demo not real code:

 c Syntax (Toggle Plain Text) 
 
/* bytes is an array of characters the is size long
 * we assume that sizeof(char) == 1, but that is true on almost all platforms
 */
void send_over_i2c(char* bytes, int size)
{
    int i = 0;
    for (i = 0; i < size; ++i)
        i2c_send(char[i]);
}
 
/* Example */
float myval = 10.321;
 
/* Treat the float pointer as the beginning to an array of chars */
send_over_i2c((char*)&myval, sizeof(char));
/* bytes is an array of characters the is size long
 * we assume that sizeof(char) == 1, but that is true on almost all platforms
 */
void send_over_i2c(char* bytes, int size)
{
    int i = 0;
    for (i = 0; i < size; ++i)
        i2c_send(char[i]);
}

/* Example */
float myval = 10.321;

/* Treat the float pointer as the beginning to an array of chars */
send_over_i2c((char*)&myval, sizeof(char));

InfoGeek · Feb 21st, 2007, 10:40 PM

Quote:

 c Syntax (Toggle Plain Text) 
send_over_i2c((char*)&myval, sizeof(char));
send_over_i2c((char*)&myval, sizeof(char));

I think you meant...

 c Syntax (Toggle Plain Text) 
send_over_i2c((char*)&myval, sizeof(float));
send_over_i2c((char*)&myval, sizeof(float));

DaWei · Feb 22nd, 2007, 3:40 AM

sizeof char is always 1. It's defined in the standard.

lectricpharaoh · Feb 22nd, 2007, 12:19 PM

Assuming that floats are the same (size, byte order, binary representation) are the same for sender and receiver, you could use a union.

(Toggle Plain Text)

typedef union
{
  float f;
  char b[4];
} floatBytes;

This way, you stick a float in the union, and grab the bytes (well chars) out. If you're not familiar with unions, they're like structs, only all members start at the same offset (they share the memory, in other words), and the size is the size of the largest member, plus any padding (as opposed to structs, which are the sum of all members and any padding).

jim mcnamara · Feb 22nd, 2007, 3:30 PM

Over tcp the default is big-endian. Even Windows supports
htonl(unsigned long hostlong);
and
ntohl(unsigned long networklong);

Assuming your floats are IEEE (32bit), then you call these to correct endianess. ntohl on recv and htonl on send.

Feb 21st, 2007, 12:57 PM	#1
sDa Newbie Join Date: Apr 2005 Posts: 4 Rep Power: 0	splitting up a float My problem is this: I have a float that I want to send over... well never mind that's not important... but the thing is I can only send one byte at the time and as you know a float is 4 bytes so how do I do that w/o loosing precision? oh and I know this isn't really a C problem but since I'm writing in C I placed it here...

Feb 21st, 2007, 1:30 PM	#3
DaWei Resident Grouch Join Date: Jun 2005 Posts: 6,453 Rep Power: 10	Hook the bytes together on the other end. If the 'other end' is a different platform, you'll have to get the order right. If it can't deal with 4-byte entities, or it uses a different floating-point procedure, your goose is cooked. __________________ Abstraction doesn't make it impossible to write bad code; it makes it possible to write superior code. Contributor's Corner: Grumpy on C++ Exceptions DaWei on Pointers

Feb 21st, 2007, 2:18 PM	#5
DaWei Resident Grouch Join Date: Jun 2005 Posts: 6,453 Rep Power: 10	A float is just another binary number. It's just interpreted differently. Extend the principle you're currently using. __________________ Abstraction doesn't make it impossible to write bad code; it makes it possible to write superior code. Contributor's Corner: Grumpy on C++ Exceptions DaWei on Pointers

Feb 21st, 2007, 7:37 PM	#6
Game_Ender Professional Programmer Join Date: May 2006 Location: Maryland, USA Posts: 306 Rep Power: 3	And make sure your target device understands/uses the same floating point format as the other. On another note the easier way to do this would be something like this (just a demo not real code: c Syntax (Toggle Plain Text) /* bytes is an array of characters the is size long * we assume that sizeof(char) == 1, but that is true on almost all platforms / void send_over_i2c(char bytes, int size) { int i = 0; for (i = 0; i < size; ++i) i2c_send(char[i]); } /* Example / float myval = 10.321; / Treat the float pointer as the beginning to an array of chars / send_over_i2c((char)&myval, sizeof(char)); /* bytes is an array of characters the is size long * we assume that sizeof(char) == 1, but that is true on almost all platforms / void send_over_i2c(char bytes, int size) { int i = 0; for (i = 0; i < size; ++i) i2c_send(char[i]); } /* Example / float myval = 10.321; / Treat the float pointer as the beginning to an array of chars / send_over_i2c((char)&myval, sizeof(char)); __________________ Robotics @ Maryland AUV Team - Software Lead

Feb 22nd, 2007, 3:40 AM	#8
DaWei Resident Grouch Join Date: Jun 2005 Posts: 6,453 Rep Power: 10	sizeof char is always 1. It's defined in the standard. __________________ Abstraction doesn't make it impossible to write bad code; it makes it possible to write superior code. Contributor's Corner: Grumpy on C++ Exceptions DaWei on Pointers

Feb 21st, 2007, 1:27 PM	#2
andro Professional Programmer Join Date: Oct 2005 Location: California Posts: 312 Rep Power: 3	Send a byte designated to mean a float is incoming, and then send the 4 bytes of the float? Whatever is receiving will have to know how to handle it.

Feb 21st, 2007, 1:33 PM	#4
sDa Newbie Join Date: Apr 2005 Posts: 4 Rep Power: 0	I think I will have to explain it a bit more... I am sending the float over an I2C bus and what you do to send it is you send 1byte at the time so for instance if I want to send a 16bit int I first send var&0x00ff then (var&ff00)>>8 (I think ) but how do I do something similar with a float? There is that "," that I really don't know what to do with

Feb 22nd, 2007, 12:19 PM	#9
lectricpharaoh Caffeinated Neural Net Join Date: Jun 2005 Location: Dry west coast of Canada Posts: 1,038 Rep Power: 5	Assuming that floats are the same (size, byte order, binary representation) are the same for sender and receiver, you could use a union. (Toggle Plain Text) typedef union { float f; char b[4]; } floatBytes; typedef union { float f; char b[4]; } floatBytes; This way, you stick a float in the union, and grab the bytes (well chars) out. If you're not familiar with unions, they're like structs, only all members start at the same offset (they share the memory, in other words), and the size is the size of the largest member, plus any padding (as opposed to structs, which are the sum of all members and any padding). __________________ And once again, Probability proves itself willing to sneak into a back alley and service Drama as would a copper-piece harlot. - Vaarsuvius, Order of the Stick

Feb 22nd, 2007, 3:30 PM	#10
jim mcnamara Hobbyist Programmer Join Date: Jun 2005 Location: New Mexico Posts: 228 Rep Power: 4	Over tcp the default is big-endian. Even Windows supports htonl(unsigned long hostlong); and ntohl(unsigned long networklong); Assuming your floats are IEEE (32bit), then you call these to correct endianess. ntohl on recv and htonl on send.

Currently Active Users Viewing This Thread: 1 (0 members and 1 guests)

Thread Tools
Show Printable Version Email this Page
Display Modes
Linear Mode Switch to Hybrid Mode Switch to Threaded Mode

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
How do I take a square root?	Fall Back Son	C	26	Oct 23rd, 2006 12:24 PM
converting string to float	beginnerCCC	C	22	Oct 2nd, 2006 11:59 PM
PLEASE HELP Convert FLOAT to STRING	Pedro	C	2	May 23rd, 2005 10:55 AM
Two ints divided into a float.	Dygear	C++	2	Apr 1st, 2005 5:42 PM
float precision and MFC...	Racine_	C++	0	Feb 5th, 2005 8:31 AM

Programming Forums > Application Development > C

splitting up a float