Xcode NSdata Garbage - ios

Question

I am Working on a Network application ... but before sending it over the network i am testing my packing and unpacking of data ... but i am getting a lot of garbage values
myfield in a ui text field .. my label is a uilabel ... mybutton method is called when a button is pressed
in the rec method the value(garbage) in x varies from -231231223432.... to [phone].....
also if i could like to pack a nsstring with this data how will i pack and unpack it
enum
{
gkMessageSent,
gkMessageNotsent
};
-(void)rec:(NSData *)data
{
const char *incomingPacket = (const char *)[data bytes];
char messageType = incomingPacket[0];
switch (messageType)
{
case gkMessageSent:
{
float x = *(float *)(incomingPacket + 1 );
// value of x are not correct here
NSString *resultString = [[NSString alloc] initWithFormat:#"%f",x];
mylabel.text= resultString;
break;
}
case gkMessageNotsent:
{
mylabel.text=#"2";
break;
}
default:
mylabel.text=#"3";
break;
}
}
-(IBAction)mybutton{
float myvalue=[myfield.text floatValue];
// i check myvalue here and its fine
NSMutableData *data= [NSMutableData dataWithCapacity:1+sizeof(float)];
int myrand=1+rand()%3;
if(myrand==1)
{
char messageType = gkMessageSent;
[data appendBytes:&messageType length:1];
[data appendBytes:&myvalue length:sizeof(float)];
}
else {
char messageType = gkMessageNotsent;
[data appendBytes:&messageType length:1];
[data appendBytes:&myvalue length:sizeof(float) ];
}
[self rec:data];
}
After Some Research I found a way to pack an NSString into NSmutabledata but cant figure out the unpacking
-(IBAction)mybutton {
float myvalue=300;
NSString *resultString = [[NSString alloc] initWithFormat:#"%.2f",myvalue];
NSMutableData *data=nil;
data= [NSMutableData dataWithCapacity:1+([resultString lengthOfBytesUsingEncoding:NSUTF8StringEncoding]) ];
int myrand=1+rand()%3;
if(myrand==1)
{
char messageType = gkMessageSent;
[data appendBytes:&messageType length:1];
[data appendBytes:[resultString UTF8String] length:[resultString lengthOfBytesUsingEncoding:NSUTF8StringEncoding]];
}
else {
char messageType = gkMessageNotsent;
[data appendBytes:&messageType length:1];
[data appendBytes:[resultString UTF8String] length:[resultString lengthOfBytesUsingEncoding:NSUTF8StringEncoding]];
}
[self rec:data];
}
-(void)rec:(NSData *)data{
const char *incomingPacket = (const char *)[data bytes];
char messageType = incomingPacket[0];
switch (messageType)
{
case gkMessageSent:
{
// Have to get the String here
break;
}
case gkMessageNotsent:
{
mylabel.text=#"2";
break;
}
default:
mylabel.text=#"3";
break;
}
}

Answer 1

You should never send "raw" data of unknown size over the network. If you were using integers, you could send the data by not sending it raw, but "correcting" its byte order to "network order" then reversing the process on the other side:
uint32t foo = htonl(myLong);
and to reverse that:
uint32t myLong = ntohl(networkData4bytesLong);
Floats and doubles can be coerced into using the above, but (to my knowledge) there is no standard way to do this. For even the following hack to work, float must be 32bit IEEE format types on both sides (probably true now on most systems).
One way is to create a union with a long and a float in it:
struct foo {
uint32_t lng;
float flt;
};
You should probably add an assert(sizeof(foo) == 4); on both sides too, just to be sure.
You put the float into a union of this type, then use htonl(foo.lng); to convert, and reverse that on the other side.
Perhaps a more reliable (but more difficult way) is to turn the float into a string, send it, then convert back to a floating point type. That said, the above should work for you with the listed assumptions (IEEE, 32bit type);
EDIT: So if you want to use a string, you can do it by appending a null terminated string to the data you send, use a "pascal" type string where the first byte is the length and a NOT null terminated string follows, or you can allow a fixed amount of space that will at least have one terminating NULL. For brevity, I am going to use the 3rd way but the "pascal" way is probably the best to use when you code it.
So, to be 100% sure we have enought space, we'll allocate 20 bytes for the float (way too much):
char floatStr[20];
sprintf(floatStr, "%f", myFloat);
[myData appendBytes:floatStr length:20];
On the other side:
unsigned char *ptr = ...; // set to the start of the string
sscanf(ptr, "%f", &myFloat);
ptr += 20;