Why C Is Statically Typed?

Why we can't just pass the number of bytes we've to read?

The answer is rather simple. C is strictly-typed. It has data types for everything. But the key lies in how those data types are translated at memory level as there exist only raw bytes.

So, how those data types are translated at memory level?

• Lets take an example of storing 42.

• The data type is int, which takes 4-bytes on 64-bit systems.

• But what is 42 at binary level? 00101010

• What if we prefix it with 24 zeroes, like this: 00000000000000000000000000101010 ? It is still 42, isn't it? What has changed is that this is a 4-byte representation of a decimal value 42 and 00101010 was a 1-byte representation of the decimal value 42.

• Suppose 42 is stored at a memory 0x1000. We use the move operation to load it in the accumulator, like this: mov eax, [0x1000] , intel syntax. What does this instruction really mean?

• It means, go on offset 0x1000, read 4 consecutive bytes from there, 0x1000, 0x1001, 0x1002, 0x1003 , sort them from little-endian to big-endian, combine them and put them into the accumulator.

• And what is there on these offsets? Since computers prefer little-endian system, the order of bits would be reversed. 0x1000 = 0x2A (42), 0x1001 = 0x00, 0x1002 = 0x00, 0x1003 = 0x00 .

How does this relate with fread?

• Interpretation is where the problem is.

• Raw bytes are raw because they can be interpreted as the programmer wants. For a char type value, 42 is *. For an integer it is 0d42. In hex system, it is 0x2A. In octal system, it is 0o52. Who will account for these interpretations?

• For char, 0x1000, 0x1001, 0x1002, 0x1003 are logically distinct locations. For int32 , 0x1000, 0x1001, 0x1002, 0x1003 is an array of two values (42 and 0). For int64, it is one-single value, 42.

• This is why fread needs to know which data type we are interpreting (though size of each entry argument) and the number of elements of such data type.