|
Stupid Question:
Scenario:
Need to passé Reference data to my IO engine. It opens-Sends-Closes or it Opens-Listens and sends data to User in call back when data is received, then closes. A pointer that can't be global would be perfect, but C# so ref to data would have to suffice. The ref has all the data to run the engine it self. The IOClass has the engine witch intern uses and API wrapper class for win32 IO functions.
Conundrum:
How do I get the managed ref to go through the win32 IO call back so the IO state engine will work?
Note – the Data object must be pointer for win32 and must be able to be read after the IO is complete by the managed C# code. There have been many articles on the IO functions in question, but now one has solved this as far as I have seen.
Class Guy Winslow
{
public:
WINcontrols
H: 925-606-1091
M: 925-922-0222
gdwinslow@wincontrols.com
};
|
|
|
|
|
It's a little difficult to follow your post, but I'll try...
You're saying the callback function exist in win32 then P/Invoke the callback, create a delegate of the property type in managed code, and create the delegate that points to the P/Invoke call. If you already have a delegate and the P/Invoke'd method and the delegate signatures don't quite match up (may you need to do some custom marshaling, although this case is best solved with an ICustomMarshaler implementation), then define a method which has the right signature and calls the P/Invoke'd win32 callback function.
Microsoft MVP, Visual C#
My Articles
|
|
|
|
|
I think i understand what you are talking about. And I thing I am doing this. Sari for mi ignorance new to .NET. This structure my explain better:
[StructLayout(LayoutKind.Sequential)]
public unsafe struct OVERLAPPED
{
public int Internal;
public int InternalHigh;
public int offset;
public int OffsetHigh;
public int hEvent;
public Socket *socket;
}
It’s defined in my API Interface class. This needs to be LocalAlloc, and passed to the ReadFileEx function. This works fine without the Socket pointer, witch is or should be a “ref Socket” to a managed class in C#. I can’t seam to passé it into the OVERLAPPED Structure so the File IO callback can use it to get rest of data and set error info. This socket class instance is created by responses to messages it receives, vary much like Window Sockets. Except it’s our device and not network device. Note - callback must be static. So am I not understanding how you are suggesting to use the P/Invoke or is there something else you guys can suggest.
Class Guy Winslow
{
public:
WINcontrols
H: 925-606-1091
M: 925-922-0222
gdwinslow@wincontrols.com
};
|
|
|
|
|
The Socket class wouldn't work even if it wasn't pinned - it doesn't have a representation in unmanaged code. Instead, you typically declare handles as IntPtrs, and use the Marshal class methods to help with marshaling values. You rarely ever need an unsafe context, either.
In this case, however, you must pin the Socket in memory. There is a chance that the garbage collector (GC) will move the object in managed memory while the unmanaged code is using it. Unmanaged code is just that - unmanaged - so it can't track such changes. Pinning the object in memory will also give you the IntPtr you need. Try something like this, for example:
[StructLayout(LayoutKind.Sequential)]
public struct OverlappedSocket : IDisposable
{
public OverlappedSocket(long internal, long internalHigh, int offset,
int offsetHigh, Socket socket)
{
this.Internal = new IntPtr(internal);
this.InternalHigh = new IntPtr(internalHigh);
this.Offset = offset;
this.OffsetHigh = offsetHigh;
hEvent = GCHandle.Alloc(socket, GCHandleType.Pinned);
}
[MarshalAs(UnmanagedType.SysUInt)] public IntPtr Internal;
[MarshalAs(UnmanagedType.SysUInt)] public IntPtr InternalHigh;
[MarshalAs(UnmanagedType.U4)] public int Offset;
[MarshalAs(UnmanagedType.U4)] public int OffsetHigh;
[MarshalAs(UnmanagedType.SysUInt)] public GCHandle hEvent;
void IDisposable.Dispose()
{
if (hEvent.IsAllocated) hEvent.Free();
}
}
Microsoft MVP, Visual C#
My Articles
|
|
|
|
|
Cool. I was playing with Marshal Attributes last night. That gets me half way there. How do I convert CGHandle back into the Socket class without duplicating it?
unsafe public static void ReadOperation(System.Int32 dwErrorCode, System.Int32 dwNumberOfBytesTransfered, ref CtrlNet_Api.OverlappedSocket lpOverlapped)
{
ref Socket socket = lpOverlapped.hEvent;
}
With my device I am getting chucks of data. So information in the Socket is being changed as Engine runes. When dune or ERRORs out. I set flag in the Socket class for Application to see status, and Trigger and Event for Calling Thread. Tried to get AddrOfPinnedObject but could not find way to cast it as Socket class.
Class Guy Winslow
{
public:
WINcontrols
H: 925-606-1091
M: 925-922-0222
gdwinslow@wincontrols.com
};
|
|
|
|
|
Again, don't use an unsafe context - you don't need it. The best examples of use for unsafe contexts is when it's faster to use pointers to iterate through characters, pixels, etc.
You don't get the Socket from the GCHandle. All the GCHandle does - the way I said to use it - is pin the Socket in memory so unmanaged code doesn't lose a reference to it in case the GC moves it (which it can do). You can still refer to the Socket in managed code just as you are now. Using the IntPtr cast operator or AddrOfPinnedObject method only gets the fixed pointer to the object for use in P/Invoke'd method calls. It doesn't change it as far as managed code is concerned.
Read about the GCHandle structure in the .NET Framework SDK.
Microsoft MVP, Visual C#
My Articles
|
|
|
|
|
Ok, I got it. I must be getting old. Senility is kicking in. I was looking at the documents and the answer was staring at me. So my code should look like this right:
public static void ReadOperation(System.Int32 dwErrorCode, System.Int32 dwNumberOfBytesTransfered, ref CtrlNet_Api.OverlappedSocket lpOverlapped)
{
GCHandle ghSocket = lpOverlapped.hEvent;
Socket socket = (Socket)ghSocket.Target;
}
I hope this is good for some kind of article or something using C# with Win32 OVERLAPED IO functions (USB).
Addition:
I tryed the solution and got "Additional information: Object contains non-primitive or non-blittable data." this error message. Any ideas?
Class Guy Winslow
{
public:
WINcontrols
H: 925-606-1091
M: 925-922-0222
gdwinslow@wincontrols.com
};
|
|
|
|
|
It really depends on how this is used, but again - you don't need to get the Socket back if you already have it. If you instantiated the Socket in your code (obviously you did), then you already have a reference. You only use the GCHandle for the purpose of calling unmanaged APIs. The original Socket reference that you passed to GCHandle.Alloc is still valid - all you did was pin it into memory (depending on how you used the GCHandle) so that it wouldn't be moved by the GC while being used in unmanaged code.
Microsoft MVP, Visual C#
My Articles
|
|
|
|
|
as i know, if we wanna deal with blob, we may use the image data type.
but my problem is, how can i insert the bytes to the database?
i have tried the following codes but it doesn't work
byte[] photo = GetPhoto(photoFilePath);
.
.
.
SqlCommand addPic = new SqlCommand ("INSERT INTO photoTable Values(" + photo + ")");
.
.
.
public static byte[] GetPhoto(string filePath)
{
FileStream fs = new FileStream(filePath, FileMode.Open, FileAccess.Read);
BinaryReader br = new BinaryReader(fs);
byte[] photo = br.ReadBytes((int)fs.Length);
br.Close();
fs.Close();
return photo;
}
thanks 
|
|
|
|
|
Use parameterized queries, which is better to use anyway that concatentating strings together, which is error prone and can lead to security holes if you don't check user input correctly (parameterized queries have many other benefits over what you're doing, which is the old, insecure way).
SqlCeCommand addPic = conn.CreateCommand();
addPic.CommandText = "INSERT INTO photoTable (Photo) VALUES(@Photo)";
SqlCeParameter picParam = addPic.Parameters.Add("@Photo", SqlDbType.Image);
picParam.Value = GetPhoto(photoFilePath);This query is much less prone to error and makes binary data possible to work with, unlike string concatenation.
Microsoft MVP, Visual C#
My Articles
|
|
|
|
|
Hi all, I am experiencing a problem and I am not sure how to go about fixing it. I have an ASP.net app that is reading from a database. I have a default database file path hardcoded into my program that the program will use if the user does not specify a database file. When executed, if the program tries to access the default database it is failing. It seems like .net is adding a "C:" to my file path where it should be "\\dir1\\dir2\\DB.mdb" Can anyone help me out so that I can access the default database on the LAN. Thanks again
|
|
|
|
|
NEVER hard code a file path. Put the filepath in the appSettings section at the bottom of the Web.config file and use ConfigurationSettings.AppSettings to get it (defaulting to a hard-coded path if you like). When you do it this way, do not escape back-slashes (that's specific to a programming language - not strings).
If you want to use a file from a share, then use "\\\\SERVER\\SHARE\\Dir1\\DB.mdb" or @"\\SERVER\SHARE\Dir1\DB.mdb" (notice the @, which is a literal character identifier).
Many of the IO members - unless you give them a URL - will assume a local path, which is why "C:" is being prepended to your string (well, the string isn't actually changed but the behavior is the same).
If this file is in your virtual directory structure of your web application, 1) make sure it's protected from user's downloading it but so that the web application can access it, and 2) use Server.MapPath or Page.MapPath to map the virtual path (like /db/db.mdb) to a physical path (like C:\Inetpub\wwwroot\db\db.mdb).
Microsoft MVP, Visual C#
My Articles
|
|
|
|
|
Updating database with changes to dataset on Form Closing
Here is my form closing method:
private void DivFrmClosing(object sender, System.ComponentModel.CancelEventArgs e)
{
int i;
if (dataSet11.HasChanges())
i = dbUpdate(); // method to update table
...
}
Row of data gets updated only when a new row has been added(focus is lost from that row)!!
How do I get around this to accomodate update of a row even when focus is not lost from current row?
|
|
|
|
|
You have to use DataGrid.EndEdit with the appropriate cell information:
DataGridCell cur = dataGrid1.CurrentCell;
DataGridColumnStyle col = dataGrid.TableStyles[0].GridColumnStyles[cur.ColumnNumber];
dataGrid1.EndEdit(col, cur.RowNumber, false); You should add proper error checking and handling, of course. This is only an example. Read the .NET Framework SDK documentation for the classes and members used in the example for more information.
Microsoft MVP, Visual C#
My Articles
|
|
|
|
|
I'm having a problem in VS2K3 whereby a file reference of a project reference isn't being copied to output folder upon compilation.
Here's the solution structure:
ProjectA
References
Assembly1.dll (File reference; CopyLocal = True)
ProjectB
References
ProjectA (Project reference; CopyLocal = True)
After compiling:
ProjectA's output folder contains:
ProjectA.dll
Assembly1.dll
ProjectB's output folder contains:
ProjectA.dll
ProjectB.dll
(but not Assembly1.dll)
Has anyone seen or resolved this?
Craig Boland
|
|
|
|
|
Select the reference under the References node and look at the properties. There is a property to control whether the assembly is copied to the target directory. If the assembly with the same name (name, version, culture, and public key token) is installed into the GAC, you should not copy the assembly to the target folder since the CLR checks the GAC first.
If the property is set to copy the assembly, then check your compiler warnings. Many times a file may be in use and cannot be copied from its source directory (or if the destination is in use, it can't be copied over). The only fix is to restart VS.NET and try again.
Microsoft MVP, Visual C#
My Articles
|
|
|
|
|
Heath,
Thank you for taking time with my question.
All references, as stated in the problem definition, are set to CopyLocal=True, yet indirect dependencies (references of references) are not copied, at least not file references.
Looking at the various projects in my solution, I see the Project references are linked to the 'obj' folder of the referenced project. Now I wonder when dependencies are copied and how they are determined, because this process seems to miss the indirect file references. Interestingly, however, the Deployment projects do a more exhaustive search along the dependency hierarchy because it finds many related files. My solution uses MS Commerce Server and some of the projects have two dozen or so dependencies, most of which don't need to be deployed because Commerce Server is a product pre-installed on the target machines.
While writing this reply, I created a small sample project to illustrate my point. In the sample, App1.exe has a project reference to ProjRef1.dll. ProjRef1 has a file reference to FileRef1.dll (which is not part of the solution) and, naturally, it works correctly. Maybe there's a problem with the size of my solution (30 projects).
|
|
|
|
|
Doubtful - I managed a 60+ project solution for builds, many of which have indirect dependencies.
Again, check your build output window for warnings about files being in use. This is most often the problem.
Microsoft MVP, Visual C#
My Articles
|
|
|
|
|
Hi,
Does anyone know what happens with errors and threading?
I have a front-end app that spawns a new thread to call a method in a dll.
This all worked well, job got done, thread closes, everything happy.
OK, so now i've got live data: The app starts, appears to work, cpu drops to zero, but no output, an event that the dll should fire doesn't happen and the thread exited with code 0 message doesn't appear in the output window.
This leads me to believe that an error is occuring in the thread and the thread is vanishing for some reason. I have got error handling in the dll, but probably not enough.
Is there anything special i have to do to get errors passed back out of the thread, there seem to be a lot of web pages talking about threading, but none seem to involve error handling.
Thanks in advance if you can help, point me in the right direction
Russ
|
|
|
|
|
See the Application.ThreadException event documentation in the .NET Framework SDK.
Microsoft MVP, Visual C#
My Articles
|
|
|
|
|
I have a dataset (theDataSet) that contains three tables. One parent (table1), a child(table2), and a child of the child (table3). I am using dataGrid2.SetDataBinding(theDataSet, "table1" ) to have the tables displayed. This all works well. The problem is, I do not want to allow the user to delete or add rows in table1 and table2. Before I did the SetDataBinding and relation stuff I used a dataView with allowDelete and allowNew set to false. (Thanks for that one Heath ) How can I incorporate a dataView with what I have now.
Thanks in advance
Jenny
|
|
|
|
|
The same as before. Create a DataView over whichever tables you want to change the behavior of and bind those by setting the DataGrid.DataSource property to the DataViews.
If those tables are all related by specifing the relationship between tables as the DataSource, then you'll have to get the DataTable from the DataRelation, get the DefaultView, and then set those properties.
Microsoft MVP, Visual C#
My Articles
|
|
|
|
|
I tried this and it doesn't work. What am I doing wrong? 
dr = new DataRelation("Class Limits",theDataSet.Tables["table1"].Columns["NoOf"],
theDataSet.Tables["table1"].Columns["NoOf"]);
dataGrid2.SetDataBinding(theDataSet, "table1");
DataTable pt = dr.ParentTable;
pt.DefaultView.AllowDelete = false;
pt.DefaultView.AllowNew = false;
|
|
|
|
|
Actually, I said if you have related tables being displayed in a DataGrid that you assign the DataRelation as the DataGrid.DataSource. You've created a DataRelation, but you haven't done anything relevant with it.
Microsoft MVP, Visual C#
My Articles
|
|
|
|
|
I'm sorry for my stupidity in this matter, but I obviously just don't get it. Can you give me an example of what you are talking about?
|
|
|
|
|
General 
News 
Suggestion 
Question 
Bug 
Answer 
Joke 
Praise 
Rant 
Admin 
Submit an article or tip
