Loop from 0x0000 to 0xFFFF

Question

I'd like a loop that uses a UInt16 (ushort) to loop through all of its values. However, this doesn't do it:

for( ushort i = 0; i < UInt16.MaxValue; i++ )
{
    // do something
}

The problem is that the loop will quit when i == 0xFFFF and not "do something". If I change the 'for' statement to "for(ushort i = 0; i <= UInt16.MaxValue; i++ )", then it becomes an infinite loop because i never gets to 0x10000 because ushorts only go to 0xFFFF.

I could make 'i' an int and cast it or assign it to a ushort variable in the loop.

Any suggestions?

you have the same issue counting down – Ferruccio Oct 23 '08 at 20:23 — Ferruccio, Oct 23 '08 at 20:23

score 17 · Accepted Answer · edited May 23 '17 at 10:33

17

Use a do...while loop

ushort i = 0;
do
{
    // do something
} while(i++ < UInt16.MaxValue);

There is an interesting discussion of testing loops at the top vs. the bottom here.

edited May 23 '17 at 10:33

Community

1
1

answered Oct 23 '08 at 19:27

Chris Marasti-Georg

34,091
15
92
137

Much nicer than my answer. :) – Jeff Yates Oct 23 '08 at 19:31
That's brilliant, yet disturbing somehow ;-) – Ferruccio Oct 23 '08 at 19:34
extremely disturbing; i'm about to remove my answwer as I made a goof, but counting down would work too, yes? like for(ushort i=UInt16.MaxValue; i>=0; i--) – warren Oct 23 '08 at 19:39
Nope, sorry warren. Since i is unsigned, it would never be less than 0. You'd still want a do...while. – Chris Marasti-Georg Oct 23 '08 at 19:48
That's why I didn't say "less than" i picked "greater than or equal" – warren Oct 23 '08 at 19:49
though you may still be right - I haven't actually tried it, and my brain's about fried on logic and counting right now :P – warren Oct 23 '08 at 19:50
warren - if i can't be negative, it will always be greater-than-or-equal-to 0. – Chris Marasti-Georg Oct 23 '08 at 19:55
Oddly enough, most of the arithmetic and comparison for ushort is being done with 4 byte ints (take the above code and compile it and load it up in ildasm or reflector)! Personally, I'd be inclined to use an int as the loop variable for this particular case. – plinth Oct 23 '08 at 19:57
Normally I would use an int, but there are calls where I need 'i' to be a ushort to call the correct method (overloading). I'd rather not have to create another variable just for the cast to ushort. – Robert Deml Oct 23 '08 at 20:09
You don't have to create another variable, just cast it explicitly to ushort. – Ferruccio Oct 23 '08 at 20:26
It could be a ushort and not be computed with 4-byte ints if it were not a PC architecture. I used to work on an embedded cpu where this question came up every day. – Mnebuerquo Oct 23 '08 at 20:34

score 5 · Answer 2 · answered Oct 23 '08 at 19:47

5

UInt16.MaxValue evaluates to 0xffff, not 0x10000. I think you can do this with a do/while loop, as a variation on burkhard1979's answer.

ushort i = 0;
do {
   ...
} while (++i != 0);

answered Oct 23 '08 at 19:47

spoulson

21,335
15
77
102

1

I prefer this answer. Much more fun! – Paul Hargreaves Oct 23 '08 at 20:36
The != 0 is redundant. But this method is the sort of thing I use. Integer overflow ftw! – Artelius Nov 09 '08 at 23:33
True. I'm a little used to C#, which doesn't implicitly convert int to bool like C/C++ does. – spoulson Nov 10 '08 at 13:15
Since it is tagged C# I think it would not work without the !=0 – Burkhard Apr 20 '10 at 14:44

score 4 · Answer 3 · answered Oct 23 '08 at 19:28

4

You could simply replace the for by a do-while loop.

ushort i = 0;
do
{
i++;
...
} while(i!=UInt16.MaxValue);

answered Oct 23 '08 at 19:28

Burkhard

14,596
22
87
108

score 1 · Answer 4 · answered Oct 23 '08 at 20:02

1

does it have to be a short? why not just

for(int i = 0;i<=0xFFFF;i++)
{
  //do whatever
}

answered Oct 23 '08 at 20:02

Kevin

7,162
11
46
70

score 1 · Answer 5 · answered Oct 23 '08 at 20:20

Assuming that your code suffers from an off by one error (the current code, stops just before having the final value evaluated. Then the following might answer you.

Very simple, as your counter is a 16 bit unsigned integer, it can not have values bigger than 0xffff, as that value is still valid, you need to have some value that extends beyond that as the guard. However adding 1 to 0xffff in 16 bits just wraps around to 0. As suggested, either use a do while loop (that does not need a guard value), or use a larger value to contain your counter.

ps. Using 16 bit variables on modern machines is actually less efficient than using 32 bit variables as no overflow code needs to be generated.

Loop from 0x0000 to 0xFFFF

5 Answers5