Thread overview
Bits in reverse order in BitArray
Jul 04, 2011
Christian Manning
Jul 04, 2011
Christian Manning
July 04, 2011
Hi, when I populate a BitArray using .init, the bits are in reverse order, although the bytes are in the correct order.
eg:

import std.stdio,std.bitmanip;

void main() {
    ubyte[] arr = [130,56,43,2];

    BitArray ba;
    ba.init(cast(void[])arr,32);
    foreach(b; ba)
       write(cast(ubyte) b); //bits in each byte reversed
                             //01000001000111001101010001000000
    writeln();

    ba.init(cast(void[])arr.reverse,32);
    foreach(b; ba.reverse)
       write(cast(ubyte) b); //bits printed in intended order
                             //10000010001110000010101100000010
    writeln();
}

Why does this happen?

Thanks
July 04, 2011
On Mon, 04 Jul 2011 14:03:35 -0400, Christian Manning <cmanning999@gmail.com> wrote:

> Hi, when I populate a BitArray using .init, the bits are in reverse order, although the bytes are in the correct order.
> eg:
>
> import std.stdio,std.bitmanip;
>
> void main() {
>      ubyte[] arr = [130,56,43,2];
>
>      BitArray ba;
>      ba.init(cast(void[])arr,32);
>      foreach(b; ba)
>         write(cast(ubyte) b); //bits in each byte reversed
>                               //01000001000111001101010001000000
>      writeln();
>
>      ba.init(cast(void[])arr.reverse,32);
>      foreach(b; ba.reverse)
>         write(cast(ubyte) b); //bits printed in intended order
>                               //10000010001110000010101100000010
>      writeln();
> }
>
> Why does this happen?

What exactly are you expecting?  In other words, when using the void[] version of init, you are specifying the exact memory to use, which implies you know how BitArray's internals work.  If you create a BitArray by adding bits to the BitArray, then it's internal storage doesn't matter to you, it can store however it wants.

I'd guess the most logical reason to store the bits in reverse order would be for performance.  Not sure what the merits are, but typically, it's faster to access and manipulate the 0 bit than some arbitrary bit.

-Steve
July 04, 2011
On 04/07/2011 19:34, Steven Schveighoffer wrote:
> On Mon, 04 Jul 2011 14:03:35 -0400, Christian Manning
> <cmanning999@gmail.com> wrote:
>
>> Hi, when I populate a BitArray using .init, the bits are in reverse
>> order, although the bytes are in the correct order.
>> eg:
>>
>> import std.stdio,std.bitmanip;
>>
>> void main() {
>> ubyte[] arr = [130,56,43,2];
>>
>> BitArray ba;
>> ba.init(cast(void[])arr,32);
>> foreach(b; ba)
>> write(cast(ubyte) b); //bits in each byte reversed
>> //01000001000111001101010001000000
>> writeln();
>>
>> ba.init(cast(void[])arr.reverse,32);
>> foreach(b; ba.reverse)
>> write(cast(ubyte) b); //bits printed in intended order
>> //10000010001110000010101100000010
>> writeln();
>> }
>>
>> Why does this happen?
>
> What exactly are you expecting? In other words, when using the void[]
> version of init, you are specifying the exact memory to use, which
> implies you know how BitArray's internals work. If you create a BitArray
> by adding bits to the BitArray, then it's internal storage doesn't
> matter to you, it can store however it wants.
>
> I'd guess the most logical reason to store the bits in reverse order
> would be for performance. Not sure what the merits are, but typically,
> it's faster to access and manipulate the 0 bit than some arbitrary bit.
>
> -Steve

Sorry, I assumed it was BitArray causing this, but that's just what made me aware of it.

It seems dmd does this when accessing pointers directly, so your reasoning is probably right!

Thanks!