Confused On References - C# Discussion Boards

19-Oct-05 12:01

I've got an ArrayList of objects. I call up a form to create a new object from one of the objects in the array.
User selects the object to create the new object from.

Now, I need to create a new object from the old object, so I thought this would work:

ArrayList objects = new Arraylist();
objects.Add(obj1);
obects.Add(obj2);
...

MyObject newobject = new MyObject();
{ ... iteratate through arraylist, find correct object }
newobject = correctObjectFromArrayList;

newobject.name = newobjectname;

objects.Add(newobject);

... and now I've modified both objects!
Both objects contain the changes I've made. Why is that?
In C++, if I did this, it would work.
It seems that C# uses references by default? I created a new object, not a pointer to an object.
I'm very confused. Can someone sort me out?

Thanks.

Matt Gerrans19-Oct-05 16:34

Matt Gerrans

19-Oct-05 16:34

Why would that work in c++?

First you initialized newobject with a new MyObject instance, and promptly assigned the reference to an existing object in your collection. The new MyObject instance you created now has no reference to it. You haven't modified "both" objects, you've lost the second one and modified the first.

What exactly are you trying to do? Create the second object, initializing it with values from the first? If so, it sounds like you might want to look at the ICloneable interface. In that case, you'd find the object in your list that you want to clone, create a new object that's a clone of it and add that new object to the list. Unless you are intending something else?

Matt Gerrans

budidharma19-Oct-05 17:32

19-Oct-05 17:32

Thank you. I should have realized.

A MyObject object has quite a bit of dynamic data in it, set by the user. I do not want to be assigned a reference of it to another object.

Like you guessed, I need to be copying all the data from the second object, into the first. Sometimes I want to create blank MyObject objects, and sometimes I want to create MyObject objects initialized with data from another object.

I always add that MyObject back into the ArrayList after it's been created.

Will the ICloneable interface work for that? I've never used an interface, though I know a very little bit about them. Where would I start with this?

Thanks.

Matt Gerrans19-Oct-05 18:07

Matt Gerrans

19-Oct-05 18:07

If you like the Java idiom, then you can use the IClonable interface and its Clone() method. Your object just needs to implement clone and copy the important elements to the other object.

If you like C++ idioms better, you can overload the equal operator and your assignment would work more like you expected; rather than assign the reference to the other object (the default behavior), it could essentially ocpy the object's attributes (to whatever exent you want it to, of course, since you can implement the operator as you see fit).

Matt Gerrans

S. Senthil Kumar19-Oct-05 23:04

S. Senthil Kumar

19-Oct-05 23:04

AFAIK, you can't overload the assignment operator. Even if you did, doesn't that go against the reference semantics that other objects follow?

How about making the OP's class a struct instead?

Regards
Senthil
_____________________________
My Blog | My Articles | WinMacro

budidharma20-Oct-05 3:30

20-Oct-05 3:30

Unfortunetly, my MyObject class makes use of quite a few member functions - so it can't be a class.

S. Senthil Kumar20-Oct-05 4:30

S. Senthil Kumar

20-Oct-05 4:30

budidharma wrote:
so it can't be a class.

Did you mean it can't be a struct? structs can also have member functions, so you should be fine.

Regards
Senthil
_____________________________
My Blog | My Articles | WinMacro

budidharma20-Oct-05 4:51

20-Oct-05 4:51

Oh yes, that's what I meant. I didn't know structs could have member functions.
In that case, what's the difference between a struct and a class?

I thought structs were simply structures of data, no member functions allowed.

S. Senthil Kumar20-Oct-05 5:41

S. Senthil Kumar

20-Oct-05 5:41

This[^] might help you understand structs bettter.

Essentially, structs are passed around by value, classes by reference. For example

struct A
{
   public int x;
}

class B
{
   public int x;
}

static void Main()
{
   A structInstance = new A();
   structInstance.x = 1;
   A structInstance1 = structInstance;
   structInstance1.x = 20;

   B classInstance = new B();
   classInstance.x = 1;
   B classInstance1 = classInstance;
   classInstance1.x = 20;
}

After executing the above code, structInstance.x will be 1 and structInstance1.x will be 20. However, classInstance.x and classInstance1.x will be 20. What it means is that when you say A structInstance1 = structInstance; a new instance of A is created and contents are copied from structInstance. When B classInstance1 = classInstance; executes, classInstance1 and classInstance begin referring to the same instance of B, therefore, changing x using any of the two references gets reflected in both.

Regards
Senthil
_____________________________
My Blog | My Articles | WinMacro

budidharma20-Oct-05 6:07

20-Oct-05 6:07

Well, thank you very much for explaining that to me. That makes perfect sense - and yes, you were correct. I merely need to change my class to a struct and it will work the way intended.

Thanks Again.

Matt Gerrans20-Oct-05 5:34

Matt Gerrans

20-Oct-05 5:34

Right. I don't know what I was thinking. Sleepy | :zzz:

So you're left with Clone().

Matt Gerrans

C# 1byte-1char strings

Dave Shaw19-Oct-05 10:49

Dave Shaw

19-Oct-05 10:49

Hi, Does anyone know an easy way to create a string of characters that are assigned 1 byte for each character, in C#.

I am interfaceing with a rather old API (written in ANSI C, think) from .NET 2005 (beta). The problem is that the normal c# string data type uses unicode encoding, this allocates atleast 8 bytes per char.

I have a Delphi 6 interface for the API that uses PCHAR data type, instead of string, and this works OK.

Also, will I need to declare the API calls with a data type that uses 1 byte per char, instead of just string?

P.S. this is the MSMAPI32.DLL to connect to Micronetics MSM (MUMPS). Incase there are any MUMPSTER's out there. Blush | :O

Thanx!

Dave Shaw
History admires the wise, but elevates the brave. - Edmund Morris

leppie19-Oct-05 11:04

leppie

19-Oct-05 11:04

Dave Shaw wrote:
The problem is that the normal c# string data type uses unicode encoding, this allocates atleast 8 bytes per char.

thats 2 bytes rather...

Use System.Text.Encoding.UTF8.GetBytes() .

^{xacc-ide 0.0.99-preview7 now with C#, C, C++, IL, XML, Nemerle, IronPython, Perl, Caml, SML, Ruby, Flex, Yacc, Java, Javascript, Lua, Prolog and Boo highlighting support!}

Dave Kreskowiak19-Oct-05 14:04

Dave Kreskowiak

19-Oct-05 14:04

Leppie is right, Unicode is 2-bytes, not 8. Unicode makes room for 65,535 characters in a set (255 * 255). 8 Bytes would allow lots and lots and lots of characters, like (255 ^ 8).

RageInTheMachine9532
"...a pungent, ghastly, stinky piece of cheese!" -- The Roaming Gnome

lmoelleb19-Oct-05 20:58

lmoelleb

19-Oct-05 20:58

Unicode makes room for something like a million characters. Unicode does not take up any "bytes". It just define which "character" is at which number. In order to represent Unicode in memory or on file, you need an encoding of the Unicode character set.

The UTF-16 encoding (which is probably what you refer to as this is what C# use internally) will use 2 or 4 bytes per character.

uno freeware19-Oct-05 16:45

uno freeware

19-Oct-05 16:45

well you can always use byte vectors, or pointers to byte in an unsafe part of the code... then you can use normal io.streamwriter for writing and io.streamreader for reading

<- true inside to understand outside ->