In my C program, I write the file bit by bit(for example :1010011111010000),but when I read them out bit by bit, the result is not which I have written(it become 1111010000101001). And the results are all right before it when reading.
What is wrong?????

my system is redhat 9.0
I use the fopen , fread, fwrite as so on,just make them bit by bit.

Thank you, sincerely

It sounds like Endianness.

Are those exact value examples? My first thought was endianness too, but if you gave exact examples, that doesn't quite appear to be the case. (Although, it could still be partly to blame.)

How exactly are you writing "bit by bit?" The smallest unit you can read/write with fread/fwrite is a byte. Maybe some code examples are in order.

I do not know if it is a endianness, and my example is exactly my program's result.

I pack the binary as follows.

int file_pack:ack(int size, long int value)
{
int i;
int tempint;
for (i=0; i<size; ++i, ++ptr, value = value>>1, sum = sum<<1) {
if (value & 1) sum |= 1;

if (ptr == 8) {
tempbuffer[bufferindex] = (unsigned char)sum;
bufferindex ++;
if(bufferindex == 1024){
if(fwrite(tempbuffer, 1, bufferindex, poutput) < bufferindex)
fatal_pack_unpack("\nPack error.");
/* printf("\nPack a block of data %d." , bufferindex);*/
bufferindex = 0;
}
/* printf(" %x ", tempch);*/
++num_of_packed_bytes;
++temp_num_of_packed_bytes;
sum=0;
ptr=0;
}
}
return i;
}

class file_pack
{
public:
int ptr ;/* how many bits are packed in sum so far */
int sum ; /* packed bits */
unsigned char tempbuffer[1024];
unsigned char tempbuffer[1024];
unsigned int bufferindex;

........
}



class file_unpack
{
public:
int ptr ;/* how many bits are packed in sum so far */
int sum ; /* unpacked bits */
unsigned char tempbuffer[1024];
int countbuffer;
int buffersize;
....
}

file_unpack::file_unpack(FILE *fptr)
{
ptr = 1;
sum = 0;
pinput = fptr;

num_of_unpacked_bytes = 0;
countbuffer = 0;
/*initialize the unpack process*/

if((buffersize = fread(tempbuffer, 1,1024, pinput) )< 1)
fatal_pack_unpack("\nUnpack error.");
/*printf(" %x ", tempch);*/
printf("\nUnpack 2 buffer %d char", buffersize);
sum = (int) tempbuffer[countbuffer];
countbuffer ++;
sum <<= 1;
num_of_unpacked_bytes ++;
}

long file_unpack::unpack(int size)
{
int i;

unsigned long int value = 0;
/* size == -2 means we initialize things */
if (size == -2) {
return((long)0);
}

/* size == -1 means we want to peek at the next bit without */
/* advancing the pointer */
if (size == -1)
return((long)((sum&256)>>8));
for (i=0; i<size; ++i, ++ptr, sum <<= 1) {
if (sum & 256) value |=(unsigned long) (1<<i);

if (ptr == 8) {
if(countbuffer == 1024 || countbuffer == buffersize){
if((buffersize = fread(tempbuffer, 1,1024, pinput) ) < 1)
if(feof(pinput)){
printf("\nRead out all the files.");
exit(1);
}
else
fatal_pack_unpack("\nRead input file error.");
countbuffer = 0;
/* printf("\nUnpack 2 buffer %d char", buffersize);*/
}
if(countbuffer < buffersize)
sum = (int) tempbuffer[countbuffer];
countbuffer ++;
num_of_unpacked_bytes ++;

/*printf(" %x ", tempch);*/

ptr=0;
}
}

return((long)value);
}


Before my error, I have pack some other data longer than 8 bits, they are not wrong, just from this one, error happens.
Why?
Maybe my fault, but I can not find out it.
Thanks

I have forgot that I execute the pack & unpack on the same computer(Dell optiplex).

Have you open file with "ios::binary" flag?

I open the file by fopen which uses "rb+", and write the file which is "wb+"

I saw no benifit from your pack && unpack. Why dont you fwrite(value) to file and then fread from file later?

because i have to record non-8-bit data to file,so may be i can not align to byte.

I don't know if it's the cause of your problem, but you'd be well advised to zero your sum and ptr variables at the start of the file_pack method, not at the end - they may contain random values first time through.

I have set them zero in the construct method. Thanks.

Just a thought,
GH

Thank you for all your help!

But finally I give up ,just pack the bytes by bytes, and I have solved the problem .