[NTLUG:Discuss] OT C question

[Johnny Cybermyth]

It would compile, but as you stepped through the code, the array wasn't 
updated by either the function call or the assignment statement.  I'll 
post the actual code so our home audience can follow along!

I was using 2 files for my test since I originally thought it was a 
module linking problem.  Here are the original files:
// main.c------------------------------------
extern unsigned char myarray[];

extern void myfunc(unsigned char *somearray);

void main(void)
{
	myfunc(myarray);
	myarray[0] = '*';
	myarray[1] = '*';
}

// file1.c------------------------------------
unsigned char myarray[5];

void myfunc(unsigned char *somearray)
{
	*somearray = ~(*somearray);
	*(somearray + 1) = ~(*(somearray + 1));
}

// end files ---------------------------------
These files work perfectly as written.  If you modify the the extern 
statement in main.c to:

extern unsigned char *myarray;

the array won't update correctly.  Also, if you try to change the code 
inside of file1.c to use brackets, you won't update the array.

I totally agree with you.  I couldn't believe what I was finding with 
this code.  I'll be the first to admit that I've probably introduced 
some logical error or something, but this is what I've found to be true 
in practice.  I'm using the Microchip c18 compiler, so I decided to 
inspect the generated assembly code and noticed that the two methods 
described above with referring to the external array actually generated 
different assembly code when accessing the array.  I thought that maybe 
the compiler was funny, but I tested the same files using MinGW and got 
the same results, debugging with gdb.  Admittedly, I am not very 
comfortable using gcc and had to learn gdb just for this test, so I may 
have done something wrong in the sanity test.

Let me know if you find differently.  Thanks for looking at this!

--Johnny

Patrick R. Michaud wrote:
> 
> 
> Standard C and C++ have always held that
> 
>     array[n]
>     *(array+n)
> 
> are equivalent, so it shouldn't have made any difference.  So
> there must be something else going on here.
> 
> Pm