|
C++ based programmers have a completely different mindset than VB programmers.
|
|
|
|
|
I'm curious--what if I said that I have a 100% DbC-compliant implementation working on .NET, and that I can prove it? How much would that be worth to the average .NET developer?
I've been doing a lot of research into DbC in .NET for the past three years and I've managed to come up with an implementation that:
-Language independent. It works on any .NET language (even works on VB.NET, heh  )
-Has full support for Pre/postconditions/invariants
-Completely transparent.
-Supports inheritance for all contract checks (that is, it combines preconditions/postconditions/invariants from parent classes correctly into a single contract check)
-Support for contracts declared on .NET interface members
-Support for attribute-based contracts, similar to XC#
-Assertions are completely debuggable. You can step through them just like any other piece of code that you write.
-Support for parameter preconditions. You can write preconditions and apply them to contracts just as easily as you would in XC#.
-Full diagnostic debugging support. If a precondition/postcondition/invariant fails, we can tell you where it failed, what method called it, and we can even assert on that method that caused it to fail.
-Support for class-based accessibility. This library allows you to write preconditions which actually prevent certain types of callers from accessing your code (similar to Eiffel's class-based visibility). For example, you can use this approach to protect internal parts of your code so that they can only be callable from a particular assembly/type/module. This approach prevents library users from using TypeMember.Invoke() to bypass visibility checks in your code because the code itself can check the type identity of the method caller and throw an exception if an unauthorized attempt is made to call the method. Not bad, eh?
Here's a sample contract definition on a single method called Divide:
<br />
public class MathClass<br />
{<br />
public int Divide(int a, [NonZero] int b)<br />
{<br />
return a / b;<br />
}<br />
}<br />
Where NonZero would be a custom precondition attribute defined as:
<br />
[AttributeUsage(AttributeTargets.Parameter)]<br />
public class NonZeroAttribute : IParameterPrecondition<br />
{<br />
#region IParameterPrecondition Members<br />
<br />
public bool Check(InvocationInfo info, ParameterInfo parameter, object argValue)<br />
{<br />
int value = (int)argValue;<br />
return value > 0;<br />
}<br />
<br />
public void ShowErrorMessage(TextWriter stdOut, InvocationInfo info, ParameterInfo parameter, object argValue)<br />
{<br />
stdOut.WriteLine("Parameter '{0}' cannot be zero", parameter.Name);<br />
}<br />
<br />
#endregion<br />
}<br />
So here's how it works: Every time the method Divide is called, my library (called LinFu, short for Language-INdependent Features Underneath) will check the parameters of your method and throw a preconditionviolationexception if the precondition check fails (much like Kevin McFarlane's approach). If the preconditon check fails, it will call ShowErrorMessage and that will be the message applied to the exception.
(Invariants and postconditions can be applied the same way, but for the sake of brevity, I'll leave it omitted).
Now what makes this interesting is that it doesn't matter whether or not you have access to the original source code--you can modify the compiled binary just as easily as if you had the source code in front of you. In fact, the library can modify the binary both dynamically at runtime AND compile time.
So, call this an unabashed advertisement, if you will
Now, I have to ask the Microsoft research people (and anybody else interested) out there--how much would this be worth to you?
|
|
|
|
|
I would suggest first posting an article on this, preferably with an example that people can try out. Then see what interest it generates. If you eventually want to sell it you probably need to effectively get people beta-testing it for a while.
Also, I assume you are aware that MS Research have Spec#, features of which I assume will make their way into .NET eventually.
Kevin
|
|
|
|
|
Looks interesting. In that example I would prefer NonZeroAttribute to override ToString() to build that message and then the calling code can do what it likes to handle the thrown exception. Also, that Check checks for >0 rather than !=0, but as it's just an example it's OK.
Of greater concern is that Check takes an object (and then tries to cast to int), would it be possible to write a condition that uses generics?
Would you expect users of your code to write their own conditions (which may be just as roughly written as the example)? Or would you include a library of well-written conditions that handle most common situations?
To answer the main question; as I doubt it's nothing I couldn't write myself, I wouldn't pay for it, it's far more fun to write it myself... I would then post it free here on CodeProject.
|
|
|
|
|
PIEBALDconsult wrote: Looks interesting. In that example I would prefer NonZeroAttribute to override ToString() to build that message and then the calling code can do what it likes to handle the thrown exception. Also, that Check checks for >0 rather than !=0, but as it's just an example it's OK.
Allowing the user to override ToString() would be the simpler approach, but I figured that if a check fails, it's better to pass the diagnostic information down to the end-user dev so that they can show exactly what went wrong. In this case, it seems a bit overkill for checking if a single parameter is null, but overall, I think it makes everything easier to debug in the longrun.
PIEBALDconsult wrote: Of greater concern is that Check takes an object (and then tries to cast to int), would it be possible to write a condition that uses generics?
It's possible, but at some point LinFu has to package the method arguments into an array of objects, and IMO, that pretty much negates the use of generics because of boxing issues, and whatnot. On the other hand, if you're worried about a potential InvalidCastException, it's easy enough to inspect the signature of the method through the InvocationInfo.TargetMethod
property and check if it's valid.
The only possible benefit I can see here is the syntactic sugar.
PIEBALDconsult wrote: Would you expect users of your code to write their own conditions (which may be just as roughly written as the example)? Or would you include a library of well-written conditions that handle most common situations?
At this point, you'd pretty much have to roll your own. I haven't had much feedback just yet from different users to build a library of common method checks, so other than the canonical NotNullAttribute and NonZeroAttribute, that's just about it.
|
|
|
|
|
Philip Laureano wrote: Allowing the user to override ToString() would be the simpler approach, but I figured that if a check fails, it's better to pass the diagnostic information down to the end-user dev so that they can show exactly what went wrong
I meant rather than the ShowWhatever method that prints to stdout.
Philip Laureano wrote: and check if it's valid.
It just seems poor style, unless you mean to throw InvalidCastException.
I would do something more like:
<br />
if ( Value is int ) <br />
if ( (int) Value == 0 )<br />
throw the normal exception<br />
else <br />
throw something that explains the problem<br />
<br />
return true ;<br />
or
<br />
if ( !( Value is int ) || ( (int) Value == 0 ) )<br />
throw the normal exception<br />
<br />
return true ;<br />
|
|
|
|
|
PIEBALDconsult wrote: Philip Laureano wrote:
and check if it's valid.
It just seems poor style, unless you mean to throw InvalidCastException.
The first few versions of LinFu that I had actually allowed users to throw specific types of exceptions, but to me, it made more sense to split the process into two parts:
1.) Allow users to check the precondition/postcondition/invariant and show whether or not it passes (hence, the boolean part).
2.) And if the check fails in debug builds, the debugger will assert with the message that you write to the stdOut TextWriter. (In release builds, however, it will specifically throw a Pre/Post/InvariantViolationException.)
While it is possible to throw exceptions of your own, I strongly recommend against it.
The whole point here is that exceptions thrown from DbC checks should never be caught. IMHO, they should fail hard so that it forces the programmer to fix whatever violated the precondition/postcondition that caused the check to fail.
However, I can't completely dismiss having users throw specific types of exceptions from checks altogether--if one does throw an exception for a check (whether it is a pre/post/invariant), I could easily catch it and wrap it into the InnerException property of a Pre/post/invariantViolationException.
p.s. Thanks for the interesting feedback!
|
|
|
|
|
This
the code itself can check the type identity of the method caller and throw an exception if an unauthorized attempt is made to call the method.
has me wondering how you do that.
I'm trying to have a method determine who called it and finding limitations. (As posted in this forum.)
|
|
|
|
|
I developed a c# application for my roomate's laptop ,I tried run my exe file on his laptop and I was getting an error message <app.exe has="" ancoutered="" a="" problem="" and="" need="" to="" close=""> so after spending many hours trying to figure out whats going on, I realised that my crystalReport portion of the code was causing the error, so I had take it off the project .
so do you guys know why it wont run if I use CrystalReport ???
ps: the application runs fine in myh laptop I guess cause I have the compiler, my roomate's laptop has only the .framework.
thanks in advance
aPerfectTool
|
|
|
|
|
Is the Crystal Reports redistributables installed on your roommate's computer?
Some people have a memory and an attention span, you should try them out one day. - Jeremy Falcon
|
|
|
|
|
I'm writing a method (MethodA) that needs to know what method of what class
called it (so it can say, "ClassX called me"). Without having to pass the
type down the chain.
To do this, MethodA walks back up the StackTrace and use GetMethod() looking
for the first method it finds with a custom attribute I created.
The problem I'm having is with derived classes which don't override a method
that calls MethodA. It turns out (in this case) that both DeclaringType and
ReflectedType return the base type rather than the derived type.
I boiled this down to:
namespace Template
{
public class ClassA
{
public virtual string
Declaring
{
get
{
return ( (new System.Diagnostics.StackTrace ( 0 , false
)).GetFrames() [ 0 ].GetMethod().DeclaringType.Name ) ;
}
}
public virtual string
Reflected
{
get
{
return ( (new System.Diagnostics.StackTrace ( 0 , false
)).GetFrames() [ 0 ].GetMethod().ReflectedType.Name ) ;
}
}
}
public class ClassB : ClassA
{
}
public class Template
{
[System.STAThread]
public static void
Main
(
string[] args
)
{
ClassA a = new ClassA() ;
ClassB b = new ClassB() ;
System.Console.WriteLine ( a.Declaring ) ;
System.Console.WriteLine ( a.Reflected ) ;
System.Console.WriteLine ( b.Declaring ) ;
System.Console.WriteLine ( b.Reflected ) ; // Shouldn't this
report ClassB?
return ;
}
}
}
The result is:
ClassA
ClassA
ClassA
ClassA
I can certainly understand why DeclaringType return the base type, but I had
hoped that ReflectedType would return the actual type that called it, and
that the result would be:
ClassA
ClassA
ClassA
ClassB <- note
How can I do this?
Am I doing something wrong?
Is there a better way?
Is ReflectedType broken?
(I posted this on MSDN too, and the response was more or less "that can't be done", so I figured I'd get a second opinion here.)
-- modified at 10:56 Monday 27th November, 2006
|
|
|
|
|
The only way I can see that being done is if you can actually get a reference to the actual method caller, and call GetType() on it.
AFAIK, the stack frame only keeps a record of method calls, and not the caller instances, so sorry, we're both out of luck.
|
|
|
|
|
Yeah, that's how it seems to be. But it sure would seem to make more sense the other way. If I am executing Derived.Method, then the frame should have a Derived.Method even though Method is declared in Base and not overridden. I mean come on, is it inherited or not?
It's not affecting me very much, I just need to walk back along the stack a little further to find an actual method that's declared in the derived class in order to find the attribute.
P.S. Did you run into the same thing? Do you have a workaround?
|
|
|
|
|
I don't have a workaround for it since I really didn't need to examine the caller for any embedded attributes. The only only workaround I could see for that is if you were to manually add type information that states which method calls which--and that would be one ugly hack to pull off.
|
|
|
|
|
Could any one please help me to get the real use of Window handlers. And Please give me an example to use it in ASP.Net application.
Thank you.
VB.Net,ASP.Net, VB, ASP, Database
|
|
|
|
|
What do you mean by "Window handler"?? There are Window "Handles", which is just a numeric identifier of a Window for use in Win32 applications. They don't apply to ASP.NET apps. There are also window message handlers, but again, they don't apply to ASP.NET apps.
Dave Kreskowiak
Microsoft MVP - Visual Basic
|
|
|
|
|
Keep the conversation in the forums. Do NOT hit the Email link. I don't care what your personal email is and will not respond to it.
Mr. Dave
I am sorry I put the correct question or not. What I want is here clearly. Thank you if you could give me the reply immediately.
"I am able to open a seperate window using buttin click and few lines of java/vb script code. But I am unable to pass the values of the second or child window browser to the first browser or the parent browser. Can any one you advice me or give me a little much of code to achieve this task.
you can mail me to my personnal email: xxxxxxxx@gmail.com, xxxxxxxxxxxxx@akebonosoft.com"
I put this question because, I can do it using window(event) handlers in Windows application. But in ASP.Net I dont know how to do it.
Aleem.
Passing values between browser windows has nothing to do with the .NET Framework. It's done entirely in Java script, which is something outside of my realm of experience. You might want to post this question in the ASP.NET Forum.
Dave Kreskowiak
Microsoft MVP - Visual Basic
|
|
|
|
|
Hi,
Can someone tell me if there is a simple way of finding all descendents of a given class in an application?
Or, is it best to use reflection to run through all classes in the currently loaded assemblies and check if they inherit from the class that I'm interested in?
In essence, what I want to do is this:
I have a base class : MyBaseClass
I inherit 'N' classes from it : MyNewClass1, MyNewClass2 ... MyNewClassN
A 'plugin' assembly inherits some more : TheOtherClass1, TheOtherClass2
I'd then like to be able to get a list of all the loaded classes that inherit from MyBaseClass ...
Cheers
Shaun
|
|
|
|
|
A class contains no information about it's descendents. You have to check each class for inheritance.
---
b { font-weight: normal; }
|
|
|
|
|
That's kind of what I thought. I ended up going through the classes in each assembly and checking that they were descended from the correct class.
Thanks
Shaun
|
|
|
|
|
hi all
is there any idea about jabber messanger,,
i want to create my own jabber chat server and client messanger ,, whre to statrt ,,, what is required ? can any one explain me the right way to follow ,, any help plz
thanks in advance
hello
|
|
|
|
|
The right way is to scope out the project from a "blue sky" point of view and work your way down, getting more detail on every level.
|
|
|
|
|
We see more and more text boxes that contain a faded text to help the user enter the right information. A good example you could have a text box wuith a label "Email" and faded in the text box we could have something like "YourAddress@yourdomain.com" to help the user. Normally as soon as data in entered , this information disapears and is never seen again. I think it is called shadowing. .NET Windows forms doesn't seem to support this, or did I miss something.
|
|
|
|
|
A website that has a control like what you are looking for is: XP Common Controls (XPCC)[^]
This site has a control that can give the shadowed text called a XPCueBannerExtender. The description on the site: The CueBannerExtender allows you to display a grayed text inside a textbox or group box that will vanish when the control gets the focus. Because this is an extender control you can simply place it onto your form and add the CueBannerText to display to each textbox and groupbox you wish.
However if you do use other controls on this site such as the XPLetterBox and XPLoginEntry. Please consider using my XP Logon Control[^] that is based off of both of the controls and adds some other user interface enhancements.
Regards,
Thomas Stockwell
Programming today is a race between software engineers striving to build bigger and better idiot-proof programs, and the Universe trying to produce bigger and better idiots. So far, the Universe is winning.
Visit my homepage Oracle Studios[ ^]
|
|
|
|
|
This effect can be fairly easily done.
Use the designer to add some sample text like email address and change the coloer to your light grey color. The designer will look like something lke:
Me.TextBox1.ForeColor = System.Drawing.Color.Silver
Me.TextBox1.Text = "Sample text"
As soon as the user starts to type in the text box you want the text to clear and the color to change. So handle a couple possible events like this:
Private m_bFirst As Boolean = True
Private Sub TextBox1_TextChanged(ByVal sender As System.Object, _
ByVal e As System.EventArgs) Handles TextBox1.TextChanged
If m_bFirst Then
'clear the text
TextBox1.Text = ""
TextBox1.ForeColor = Color.Black
End If
End Sub
Private Sub TextBox1_MouseDown(ByVal sender As System.Object, _
ByVal e As System.Windows.Forms.MouseEventArgs) _
Handles TextBox1.MouseDown
If m_bFirst Then
'clear the text
TextBox1.Text = ""
TextBox1.ForeColor = Color.Black
End If
End Sub
I have just made these changes to the form where I was using the text box. If you wanted to use this style more then once you could make a custom control.
|
|
|
|
|
General 
News 
Suggestion 
Question 
Bug 
Answer 
Joke 
Praise 
Rant 
Admin 
Submit an article or tip
