Read-only properties

[Derek Bartram]

That's the problem though... the amount of people who i've seen doing things like modify data structures in property gets, is pretty high. Why leave elements of the language that can result in really hard to debug code and misconceptions?

[GibbleCH]

But, often, if you don't modify the data of the class, when you set the property, your data may be invalid...and calling functionX at that time will result in invalid results, or errors.

[leppie]

PIEBALDconsult wrote:

I recall the class specifies which members get serialized and deserialized, so this shouldn't be a problem.

Sometimes you want only readonly properties in XML serialization. Unfortunately for de/serialization to work, properties need to have both a getter and a setter.

xacc.ide - now with IronScheme support
IronScheme - 1.0 alpha 2 out now

[PIEBALDconsult]

The set accessor should then be private. However, the base class' contract may not allow that and then you're stuck.

[Derek Bartram]

Yes, but it leads to misconceptions about what the code does. I suppose it all comes down to good code documentation perhaps

[leppie]

PIEBALDconsult wrote:

The set accessor should then be private.

No Xml serialization (I am getting tired typing that!) will choke on that. Just try it!

xacc.ide - now with IronScheme support
IronScheme - 1.0 alpha 2 out now

[PIEBALDconsult]

Oh, I will... I will...

[PIEBALDconsult]

I had no trouble with it.

namespace Template
{
public partial class MyClass : System.Xml.Serialization.IXmlSerializable
{
public MyClass
(
)
{
}

    public MyClass
    (
        string Name
    )
    {
        this.Name = Name ;
        
        return ;
    }

    public string
    Name
    {
        get ;
        
        **private** set ;
    }
    
    public void 
    WriteXml 
    (
        System.Xml.XmlWriter Writer
    )
    {
        System.Xml.XmlDocument doc = new System.Xml.XmlDocument() ;
        
        doc.AppendChild ( doc.CreateElement ( "MyClass" ) ) ;
        
        doc.DocumentElement.InnerText = this.Name ;
        
        doc.WriteTo ( Writer ) ;
        
        Writer.Close() ;
        
        return ;
    }

    public void 
    ReadXml 
    (
        System.Xml.XmlReader Reader 
    )
    {
        System.Xml.XmlDocument doc = new System.Xml.XmlDocument() ;
        
        doc.Load ( Reader ) ;
        
        Reader.Close() ;
        
        **this.Name = doc.DocumentElement.InnerText ;**     
   
        return ;
    }

    public System.Xml.Schema.XmlSchema
    GetSchema
    (
    )
    {
        return ( null ) ;
    }

    public override string
    ToString
    (
    )
    {
        return ( this.Name ) ;
    }
}

public partial class Template
{
    private static System.Xml.Serialization.IXmlSerializable
    Write
    (
        System.Xml.Serialization.IXmlSerializable Subject
    )
    {
        Subject.WriteXml ( System.Xml.XmlWriter.Create ( @"C:\\X.xml" ) ) ;
        
        return ( Subject ) ;
    }

    private static System.Xml.Serialization.IXmlSerializable
    Read
    (
        System.Xml.Serialization.IXmlSerializable Subject
    )
    {
        Subject.ReadXml ( System.Xml.XmlReader.Create ( @"C:\\X.xml" ) ) ;
        
        return ( Subject ) ;
    }

    \[System.STAThreadAttribute()\]
    public static int
    Main
    (
        string\[\] args
    )
    {
        int result = 0 ;

[leppie]

Implementing your own custom IXmlSerializable is cheating ;P For starters you need to use the XmlSerializer. And you code should just represent a basic C# object. Example:

using System;
using System.Xml.Serialization;
using System.IO;

public class Foo
{
public int Bar { get; private set; }
}

class Entrypoint
{
static void Main()
{
XmlSerializer ser = new XmlSerializer(typeof(Foo));

using (Stream s \= File.OpenWrite("foo.xml"))
{
  ser.Serialize(s, new Foo());
}

}
}

Fails!

<

[leppie]

This is a continuation of the sample: You might say, then use the XmlIgnore attribute, but in that case the property is never emitted, but what if you want to emit it, but you simply dont care about the result after deserializing it?

xacc.ide - now with IronScheme support
IronScheme - 1.0 alpha 2 out now

[PIEBALDconsult]

That's just crazy talk; if the class doesn't implement IXmlSerializable then clearly it can't be serialized to XML, so don't try. This is like trying to use a DataAdapter on a query with a join or a view or something and then complaining that .Update won't work. And as I said, "Not if it's done right"; I done it right, you didn't. I win, neener neener neener! :laugh:

[PIEBALDconsult]

mav.northwind wrote:

Black holes are the places where God divided by 0...

Man: Hey, God, why did you make black holes? God: I like a good BM in the morning, same as the next guy.

[PIEBALDconsult]

Why write a class in a particular (bad) way just to bypass a flaw in some other brain-dead class? Dealing with a class that someone else wrote which doesn't implement the desired serialization is another matter entirely, but when you're writing the class you have control. If you're writing a class intending it to be serialized, then it should be marked with SerializableAttribute and implement ISerializable and/or IXmlSerializable as appropriate. "Use the right tool for the right job." -- Scotty, et al

[leppie]

PIEBALDconsult wrote:

That's just crazy talk; if the class doesn't implement IXmlSerializable then clearly it can't be serialized to XML, so don't try.

The way I showed you is exactly how web services does it. [update] I think, it's 5:30 am, I just woke for a smoke, not thinking really [update]

xacc.ide - now with IronScheme support
IronScheme - 1.0 alpha 2 out now

modified on Saturday, March 22, 2008 11:38 PM

[Super Lloyd]

You missed public virtual Property { get {...} set {..}}

[PIEBALDconsult]

Well that is what I was alluding to when I said: " The set accessor should then be private. However, the base class' contract may not allow that and then you're stuck. " (In a different branch.) However in your example, the derived class should at least call base.Property. If the base class is abstract then you're stuck (but an exception should be thrown or something!). However, I argue that an abstract class should probably not specify that a set is required, which may be a whole new topic.

[Fatbuddha 1]

COOL!!!! That one is outstanding :laugh: Cheers

You have the thought that modern physics just relay on assumptions, that somehow depends on a smile of a cat, which isn’t there.( Albert Einstein)

[Jeff Dickey]

...to a sufficiently motivated/determined idiot. No useful language can enforce a fixed floor for intelligence or sensibility of its (ab)user. Languages that make significant efforts in those directions, like Logo or Wirth's original Pascal, find themselves compartmentalised and marginalised safely out of the way of any Real Programmers™. You can't effectively use language semantics to guard against process failure. The entire reasoning for practices like XP pair programming, test-driven development that requires every bit of code to be tested, and so on is precisely to guard against such ingenious stupidity.

Jeff Dickey Seven Sigma Software and Services Phone/SMS: +65 8333 4403 Yahoo! IM: jeff_dickey MSN IM:    jeff_dickey at hotmail.com ICQ IM:    8053918 Skype:     jeff_dickey

[Derek Bartram]

No useful language is idiot-proof... But languages can be FOR idiots..... *see Java

[Jeff Dickey]

Java, as originally conceived (J2SE, that is) is a perfectly decent language for several different problem domains, including particularly those similar to that for which it was created. The idiocy comes from two things. First, the "Java should do everything including the kitchen sink" mentality that gave us J2ME and (good working definition of "coding horror") J2EE. Secondly, the dishonest marketing commonly associated with the infinite battalions of H1B coders who have perfectly good-sounding (though mass-mimeographed) "qualifications" but couldn't solve a problem in Java (or apparently anything else) if you gave them a flashlight, a map, all the textbooks they could carry and a two-year head start. But they're the fashion, so thousands of lemmings cleverly disguised as "software-using companies" are throwing themselves of the cliff. (It's a little late to realize that you need an exit strategy when you've already been accelerating at 10 m/sec/sec for two or three minutes....)

Jeff Dickey Seven Sigma Software and Services Phone/SMS: +65 8333 4403 Yahoo! IM: jeff_dickey MSN IM:    jeff_dickey at hotmail.com ICQ IM:    8053918 Skype:     jeff_dickey