C# Discussion Boards - CodeProject

Colin Angus Mackay15-Jul-08 13:30

Colin Angus Mackay

15-Jul-08 13:30

Have a look in the System.Reflection namespace for Activator (I think)

Recent blog posts:
* Event Organisation (Feedback)
* LINQ to XML (part 4)
* Scottish Developers June Newsletter

My Blog

DaveyM6915-Jul-08 13:45

15-Jul-08 13:45

System.Activator Hmm... a little work to do to figure this one out!

Thanks Colin

Dave

Guffa15-Jul-08 13:57

15-Jul-08 13:57

The answer would be reflection. Whenever that's the answer, you are either doing something pretty complicated, or something wrong.

What is it that you are trying to accomplish, really? Usually there is an object oriented way to solve it.

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

DaveyM6915-Jul-08 14:06

15-Jul-08 14:06

A member variable in my struct is a UInt64. (I'm only using unsigned types here!)

1. Assume that the member variable's value is 8 : The smallest Type that could have that value is a Byte so my method returns Type Byte.

From this Type returned by the struct I want to create a variable of Type Byte.

2. Assume that the member variable's value is 256 : The smallest Type that could have that value is a UInt16 so my method returns Type UInt16.

From this Type returned by the struct I want to create a variable of Type UInt16.

3. UInt32...

4. UInt64...

Dave

Guffa15-Jul-08 14:18

15-Jul-08 14:18

Yes, I got it so far, but what is the point? What do you plan to do with the data?

If you create data of different types, the data has to be boxed in an object. To then use the data, you have to unbox it to the actual type.

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

DaveyM6915-Jul-08 21:47

15-Jul-08 21:47

Depending on the type, the data needs to be handled differently.

I suppose I could box the variable to object inside the method and make it return Object then something like

if (returnedObject is Byte)
{
    Byte x = 0;
}
else if(returnedObject is UInt16)
{
    UInt16 x = 0;
}

but I was ideally wanting to return a Type as the value itself is not needed so there seems no point in returning it - only the type. Also, this way I have to stay within each if block or x is out of scope.

Dave

DaveyM6915-Jul-08 21:59

15-Jul-08 21:59

This works and will suffice if it can't be done by returning Type

object x = new object();
if (returnedObject is Byte)
{
    x = (Byte)returnedObject;
}
else if(returnedObject is UInt16)
{
    x = (UInt16)returnedObject;
}

Dave

Guffa16-Jul-08 1:33

Re: create object from returned Type [modified]

16-Jul-08 1:33

DaveyM69 wrote:
This works and will suffice if it can't be done by returning Type

object x = new object();
if (returnedObject is Byte)
{
x = (Byte)returnedObject;
}
else if(returnedObject is UInt16)
{
x = (UInt16)returnedObject;
}

That code only unboxes the value and boxes it again. You can just copy the reference and get the same result:

object x = returnedObject

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

DaveyM6915-Jul-08 22:13

Re: create object from returned Type [modified]

15-Jul-08 22:13

Got it!

var x = Activator.CreateInstance(returnedType);

It was simple after all Big Grin | :-D

Dave

modified on Wednesday, July 16, 2008 4:21 AM

Guffa16-Jul-08 1:37

16-Jul-08 1:37

That's misuse of the var keyword. It should look like this:

object x = Activator.CreateInstance(returnedType);

It's still pretty useless. You can't really do anything with the data without checking the type and unboxing it. What is it that you are trying to accomplish?

[edit]

Actually, it's totally useless. You have created an instance of a value type without a value, and as it's boxed, you can't set the value of the instance.

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

modified on Wednesday, July 16, 2008 8:52 AM

DaveyM6916-Jul-08 8:29

Creating An Array of Stats from a DataTable

16-Jul-08 8:29

Yeah - I figured that trying to apply it! Returning a new initialized instance of the correct type instead of just the type and checking using is on return is working, I've just had to restructure the code a little. Just seemed a little messy though but problem solved. Thanks for your help.

Dave

Jammer15-Jul-08 10:53

15-Jul-08 10:53

Hi All,

I'm looking to create a set of stats from a DataTable and I'm currently having a bit of trouble working out the best way to do it. There are a few columns in the table that I'm interested in with an unknown number of possible values in each of these columns. I have defined a Statistic type like:

public class Statistic
{
    public string StatNameTag { get; set; }
    public string Description { get; set; }
    public int Value { get; set; }
    public double Percentage { get; set; }

    public Statistic()
    {
    }
}

Which are in turn held in:

public static ObservableCollection<statistic> StatsCollection = new ObservableCollection<statistic>();</statistic></statistic>

Currently I'm thinking that a foreach loop over the table is the way forward. Each time a new value in a given table column is found I'll create a new Statistic object to hold the information in and give it a StatTagName that includes the value in the name which will provide a way to find out if its a new value or not. If its not a new value increment the value on the current Statistic object already in the collection.

Am I going in the right direction with this, or would it make more sense to use the .Compute() method or something else altogether? Or should I skip this altogether and execute a series of ExecuteScalar() commands directly on the database after performing a distinct select on each column i'm interested in?

I'm struggling with the best way to go with this. Everything else seems a bit expensive compared to a foreach loop over an already loaded DataTable object.

Any input would be great. TIA,

Jammer

Going where everyone here has gone before! Smile | :)

Pete O'Hanlon15-Jul-08 11:03

Pete O'Hanlon

15-Jul-08 11:03

Hmmm - rather than using a DataTable, why not take a look at using Linq to do this? It would look something like:

var p1 = from p in dataContext
 group p by p.Name
 into MyGroup
 select new { MyGroup.Description, MyGroup.Group.Sum(x=>x.Percentage), MyGroup.Group.Sum(y=>y.Value), MyGroup.Description };

I'm typing this blind, so it might not be 100% accurate, but you should get the idea.

Deja View - the feeling that you've seen this post before.

My blog | My articles

Jammer15-Jul-08 11:28

15-Jul-08 11:28

That does look interesting indeed!! I'll have a play around with this method. Thanks Pete.

Jammer

Going where everyone here has gone before! Smile | :)

Jammer15-Jul-08 14:21

15-Jul-08 14:21

I've gotten so far with this:

public void UpdateStats()
{
    var p1 = from p in context.LocalSamples
             group p by p.FileBitRate into MyGroup
             select new {description = MyGroup.Key, percent = MyGroup.Group.Sum(x=>x.Percentage), value = MyGroup.Group.Sum(y=>y.Value) };
}

However, MyGroup doesn't appear to have a .Group method, only a .GroupBy<> but I've seen stacks of examples online using .Group.Sum ... so I'm a bit confused by that. I was also seeing errors saying:

"Invalid anonymous type member declarator"

Adding in description =, percent = solved that one, but I'm still unsure why I'm seeing a problem with the .Group.Sum ...

Tia,

Jammer

Going where everyone here has gone before! Smile | :)

Pete O'Hanlon15-Jul-08 22:23

Pete O'Hanlon

15-Jul-08 22:23

Sorry mate. Change this to:

var p1 = from p in context.LocalSamples                     group p by p.FileBitRate into MyGroup                     select new {description = MyGroup.Key, percent = MyGroup.Sum(x=>x.Percentage), value = MyGroup.Sum(y=>y.Value) };

Deja View - the feeling that you've seen this post before.

My blog | My articles

Jammer16-Jul-08 0:49

16-Jul-08 0:49

Ahhh, thanks fella ... will have a look at this later tonight.

Jammer

Going where everyone here has gone before! Smile | :)

Jammer16-Jul-08 8:38

16-Jul-08 8:38

Hi Again Pete,

On these two elements of the linq code:

percent = MyGroup.Sum(x=>x.Percentage), value = MyGroup.Sum(y=>y.Value

I get the following error:

DataLayer.LocalSample' does not contain a definition for 'Percentage'
DataLayer.LocalSample' does not contain a definition for 'Value'

Should the MyGroup (is this an anonymous type?) actually be an instance of my Statistic class? Or should I be creating the Statistic objects in a foreach loop over the p1 variable once this code has been executed?

the x. / y. is expecting a field from my LocalSample table and Percentage/Value are the things I'm trying to calculate. Bits of this are clearer to me now, but I'm still way off really understanding this.

Thanks chap.

Jammer

Going where everyone here has gone before! Smile | :)

C# Raw IP Address

DwR15-Jul-08 9:19

DwR

15-Jul-08 9:19

Hi,
Can anyone tell me the C# equivalent ofthe following java code?

address = InetAddress.getLocalHost().getAddress();

I've not done anything with networking in C# before but from an admittedly quick inspection it does appear that the java implementation makes more intuitive sense. I'd say it's more feature rich too but I haven't ben able to make complete sense yet of what C# has to offer.

Regards,
Dave

Regards,
Dave

Pete O'Hanlon15-Jul-08 9:27

Pete O'Hanlon

15-Jul-08 9:27

DwR wrote:
Can anyone tell me the C# equivalent ofthe following java code?

address = InetAddress.getLocalHost().getAddress();

IPAddress address = Dns.Resolve("localhost");

Deja View - the feeling that you've seen this post before.

My blog | My articles

DwR15-Jul-08 12:05

DwR

15-Jul-08 12:05

Hi,
Thanks for that. A couple of questions / observations:

1. The C# documentation refers to the resolve method as being obsolete.

2. The java type of "address" is byte[]. The method returns a "raw" ip address. The C# equivalents, including Dns.resolve return an IPAddress[] type which i assume isn't raw. Can C# return raw addresses? I assume the type to use is c# byte[]...

Additionally, I'm unclear as to what the java documentation means by a raw ip address. Can you clarify? I'm expecting that it means that I the IPAddress[] type is not the correct type for this operation.

Regards,
Dave

Colin Angus Mackay15-Jul-08 13:32

Colin Angus Mackay

15-Jul-08 13:32

DwR wrote:
I'm unclear as to what the java documentation means by a raw ip address

One would presume that the raw IP Address is just the bytes that make up the IP Address - as it would normally be with anything described as raw.

DwR wrote:
I'm expecting that it means that I the IPAddress[] type is not the correct type for this operation

What operation?

Recent blog posts:
* Event Organisation (Feedback)
* LINQ to XML (part 4)
* Scottish Developers June Newsletter

My Blog