Re: Bitwise Circular Shift in C# - C# Discussion Boards

Ian Uy24-Jul-08 7:26

Ian Uy

24-Jul-08 7:26

Thanks,

For efficiency, can I put:

times%=16;

Since after 16 rotations, its just 0.

I mean, rot 17 is just rot 1.

User 665824-Jul-08 7:31

User 6658

24-Jul-08 7:31

Good point, I'd rather do:

for (int i = 0; i < times % 8; i++)

regards

modified 12-Sep-18 21:01pm.

Luc Pattyn24-Jul-08 7:36

Luc Pattyn

24-Jul-08 7:36

Hi,

if it is a byte, then it is modulo 8 you want.

IMO, assuming you need a lot of those, the fastest implementation for byte data is
using a two-dimensional array, properly initialized.

byteShifter[shiftAmount%8][value];

Of course you could store all kinds of 2-operand functions in such a table,
not just rotates.
For larger data items, this approach is becoming memory costly.

FYI: with C# modulo may yield negative values for negative inputs, you may have
to correct for that.

Smile | :)

Luc Pattyn [Forum Guidelines] [My Articles]

Voting for dummies? No thanks. Dead | X|

PIEBALDconsult24-Jul-08 8:43

PIEBALDconsult

24-Jul-08 8:43

Ew, not loops.

public static byte
RightRotate
(
    byte value
,
    byte times
)
{
    times %= 8 ;
    return ( (byte) ( value >> times | value << 8-times ) ) ;
}

public static byte
LeftRotate
(
    byte value
,
    byte times
)
{
    times %= 8 ;
    return ( (byte) ( value << times | value >> 8-times ) ) ;
}

But as Luc says... table lookup.

Guffa24-Jul-08 23:06

Guffa

24-Jul-08 23:06

PIEBALDconsult wrote:
But as Luc says... table lookup.

Would that really be faster?

It looks to me like validating the indexes and calculating the offset in the array data would be about as much work as just calculating the value.

Despite everything, the person most likely to be fooling you next is yourself.

PIEBALDconsult25-Jul-08 11:03

PIEBALDconsult

25-Jul-08 11:03

Then use unsafe code?

Ennis Ray Lynch, Jr.24-Jul-08 12:15

Ennis Ray Lynch, Jr.

24-Jul-08 12:15

I don't think << is circular but I do believe there is an opcode which is which would mean C++ and not C#

Need a C# Consultant? I'm available.

Happiness in intelligent people is the rarest thing I know. -- Ernest Hemingway

how to "copy local dependencies" in C# !?

Ian Uy25-Jul-08 16:05

Ian Uy

25-Jul-08 16:05

True, << and >> means Left Shift and Right Shift.

But when done in a specific way, it can be "tricked" to do Circular shifts. Smile | :)

Skeletor2324-Jul-08 4:21

Skeletor23

24-Jul-08 4:21

Hi,

Problem:
I have a VS2005 C# Application that uses a third party managed dll which needs a second unmanaged dll.
I added the managed dll to my projects references, and so it is copied to my application path automatically after build. But what about the unmanaged dll?
How to copy it automatically? In C++/CLI you can set "copy local dependencies" and all dependencies are copied automatically.
But in c# I found nothing like that?

Any Ideas? Thanks for any help!

Frank Horn24-Jul-08 4:28

Frank Horn

24-Jul-08 4:28

Don't know if there's an automatism for this in Visual Studio. I usually put the unmanaged dll into the project folder and copy it with the post-build event command line:

copy "$(ProjectDir)unmanaged.dll" "$(TargetDir)\unmanaged.dll"

Skeletor2328-Jul-08 22:32

Skeletor23

28-Jul-08 22:32

that was also my first idea.

but I don't want to copy to the targetdir of the project, but to the output path of the whole solution ... since my project is used in other projects and solution dir's may vary.

so I tried something like that:
copy "$(ProjectDir)unmanaged.dll" "$(SolutionDir)\$(SolutionName)\unmanaged.dll"

Problem is, SolutionDir is not the OutputPath and there's no macro that defines the output path!?
Of course I can use "$(SolutionDir)\$(SolutionName)\$(OutDir) ... but this works only if Solution and the Project are build with same settings.
But why is there no macro for the Solution output path!?

By the way, the second proposal just works fine.

Frank Horn28-Jul-08 23:25

Frank Horn

28-Jul-08 23:25

By the way, the second proposal just works fine.

Yes, I know. Just didn't think of it cause I always copy things to TargetDir. This is because we often use a kind of plug in technique: application reads from configuration from which dll to load the controls, so we don't have a reference to the controls dll and the application won't find it when started from VS, unless we copy it to the folder where the exe ist started.

Struct / Class strange problem

J a a n s24-Jul-08 4:45

J a a n s

24-Jul-08 4:45

AFAIK there should be a "Copy to Output Dir" property. Select the component in solution explorer and check the properties. There you can set
Build Action -> Component
Copy to output Dir -> Copy Always/Copy if newer

*jaans

lune1224-Jul-08 4:16

lune12

24-Jul-08 4:16

Hi,

I have encountered a very strange problem.

I have a stuct that has a string variable inside.

I'm creating a list of the Struct's type, and try to assign the string variable.
this doesn't work !!!
but If I create a Class instead of a struct this work!!

The Struct:

public struct Struct1
{
    private string s;
    public Struct1(string d)
    {
        s = "";
    }
    public void SetS(string i) { s = i; }
    public string GetS() { return s; }
}

The Class:

class myStr
{
    private string s;
    public myStr(){}
    public void SetS(string i) { s = i; }
    public string GetS() { return s; }
}

The Main:

static void Main(string[] args)
{
    List<Struct1> ls = new List<Struct1>();
    Struct1 str = new Struct1("f");
    ls.Add(str);
    ls[0].SetS("hi");
    string mys = ls[0].GetS();
    Console.WriteLine("mys: " + mys);

    List<myStr> ls2 = new List<myStr>();
    myStr str2 = new myStr();
    ls2.Add(str2);
    ls2[0].SetS("hello");
    string mys2 = ls2[0].GetS();
    Console.WriteLine("mys2: " + mys2);
}

The output is:
mys:
mys2: hello

this seems very strange!
I would be very happy if someone has an explanation.

Thanks.

leppie24-Jul-08 4:31

leppie

24-Jul-08 4:31

lune12 wrote:
this seems very strange!

No, the behaviour is perfectly correct.

lune12 wrote:
I would be very happy if someone has an explanation.

Read up on the differences between value and reference types.

xacc.ide - now with TabsToSpaces support
IronScheme - 1.0 alpha 4a out now (29 May 2008)

User 665824-Jul-08 6:56

User 6658

24-Jul-08 6:56

leppie's is correct.

In case of a struct the operation ls[0] will return a copy of the struct that's inside the list, so assigning a value to it and then discarding it (by letting it get out of scope) will have no effect.

In contrast, in case of a class the operation ls[0] will return a reference to the object in the list, so assigning a different value to it is possible.

regards

modified 12-Sep-18 21:01pm.

Guffa24-Jul-08 11:10

Guffa

24-Jul-08 11:10

In addition to what has already been said:

Mutable structs (ones where you can change the members) should generally be avoided, for just the reason that you have experienced. If you make the struct immutable, it would work, as then you would have to create a new instance to get one with a different value:

public struct Struct1 {
   private string _s;
   public Struct1(string d) {
       _s = d;
   }
   public string S { get { return _s; } }
}

static void Main(string[] args) {
   List<struct1> ls = new List<struct1>();
   Struct1 str = new Struct1("f");
   ls.Add(str);
   // This won't compile:
   // ls[0].S = "hi";
   // You have to:
   ls[0] = new Struct1("hi");
   string mys = ls[0].S;
   Console.WriteLine("mys: " + mys);
}</struct1></struct1>

Structs are tricker to implement correctly than classes, so unless you really need a struct, just stick with classes.

Despite everything, the person most likely to be fooling you next is yourself.