C / C++ / MFC

18-Dec-09 4:38

Hello,

1)How can I Defrag Individual Files ?
2)How can I Defrag Registry Hives ?

Thanks,

Richard MacCutchan18-Dec-09 9:18

18-Dec-09 9:18

Adnan562 wrote:
1)How can I Defrag Individual Files ?

If you mean the registry you can't; if you mean the disk then just right click on the drive in Windows Explorer and select Properties, and then Tools.

Adnan562 wrote:
2)How can I Defrag Registry Hives ?

You cannot do this unless you understand what each of the keys is for. Most software adds keys to the registry for its own purpose and any changes made by a user has the potential for creating complete chaos on your computer. The only sensible way to clean the registry completely is to reinstall Windows followed by all your application software. A rather extreme option I agree, but the only safe one.

Adnan56218-Dec-09 9:36

18-Dec-09 9:36

I mean to say, to defrag via code.

Windows API I can use.

I want to develop something like PageDefrag from SysInternals
http://technet.microsoft.com/en-us/sysinternals/bb897426.aspx

Can any provide link to article or API info ?

Thanks,

Richard MacCutchan18-Dec-09 9:52

18-Dec-09 9:52

Adnan562 wrote:
I mean to say, to defrag via code.

What do you mean by defrag? Unless you understand the contents of the registry completely, and all the links and cross references between the different keys, you could end up in a total mess.

The Sysinternals people are experts with detailed knowledge of the internal systems of Windows, much of which is not published material. I assume you do not have such experience.

Adnan562 wrote:
Can any provide link to article or API info ?

You have already provided the link; take a look at Mark Russinovich's article and the links he provides.

Adnan56218-Dec-09 13:04

18-Dec-09 13:04

Registry Hives are in files, I know where it is.
Open PageDefrag,you will see the files.

The article link is dead.

Richard MacCutchan18-Dec-09 23:29

18-Dec-09 23:29

I found this MSDN page[^] which may help you.

Adnan56219-Dec-09 7:39

winsock select don´t react on incoming message

19-Dec-09 7:39

Thank you.
I will read.

CrazyDogg18-Dec-09 4:31

CrazyDogg

18-Dec-09 4:31

Hi @ all,´
i have the problem, i´m trying to send a message via UPD to a client but the select method reacts if the message is incoming on the client.
The client is running as a windows service.
With wireshark I can see the incoming message.

The client is waiting on messages following

while(g_bRun)

{

FD_ZERO(&fdSet);

FD_SET(g_pSipSocket->getSocket(), &fdSet);

rv = select(maxfd + 1, &fdSet, NULL, NULL, &timeVal);

AddDBS(1, "fdSet: %d\n", fdSet.fd_count);

if(rv > 0)

{

rv = recvfrom(g_pSipSocket->getSocket(), buf, sizeof(buf), 0, (SOCKADDR*)&sa, &salen);



if(rv != -1)

{



delay(100);

buf[rv] = '\0';

writeLog(buf);

}

}

}

the special clue is, if I´m sending a message to the client directly after starting my service then it works. If I´m waiting then or before ca. 2 - 3 minutes then the problem manner.

Sorry for my bad english

Best regards
Crazy

JNI CallVoidMethod leads to Runtime Error

TheInfernalCrow18-Dec-09 4:22

18-Dec-09 4:22

Premises:
1) jobject_instance is a reference to the correct object (another non-void method call worked just as expected), Type: jObject
2) j_method_id is the correct id for the called method, Type: jmethodID
3) Method expects one int as argument and is (of course) void
4) Code is compiled to DLL and used by external application

m_jni_env->CallVoidMethod(jobject_instance, j_method_id,(jint)n);

This results in: Runtime Error R6025, pure virtual function call. Any ideas?

Regards

Richard MacCutchan18-Dec-09 9:21

18-Dec-09 9:21

Can you show the Java code for j_method_id and how you obtain its pointer?

TheInfernalCrow18-Dec-09 23:53

18-Dec-09 23:53

Hi,

sure:

Getting the method id:

j_method_id = m_jni_env->GetMethodID(m_class_ref, "testMethod", "(I)V");

m_class_ref is the correct jclass reference, contructor call and another method using this reference are working as expected

and the Java code of the method:

public void testMethod(int n){
	try {
		this.debugOutWriter.write("testMethod called, n = " + Integer.toString(n) );
		this.debugOutWriter.write("----------------------");
		this.debugOutWriter.newLine();
		this.debugOutWriter.flush();
	} catch (IOException e) {
		// Nop
	}
}

debugOutWriter is a BufferedWriter just used to test the entire jni stuff before actually implement the real functionality.

Richard MacCutchan19-Dec-09 3:16

19-Dec-09 3:16

I've just compared this with my own JVM test system and all the calls and parameters look fine. I can only surmise that the external app or the DLL are somehow not in sync. If you call the Java function direct, rather than through the dll does that work? Also, are you sure the runtime error is on the call to the java method, rather than to some class method in the DLL?

TheInfernalCrow19-Dec-09 4:08

19-Dec-09 4:08

Yes, using the class in a plain java application works just fine.

Directly in front and behind the JNI call for the void method are the following debug outputs ...

fprintf(m_debug, "calling java void method\n" );
fflush(m_debug);

m_jni_env->CallVoidMethod(jobject_instance, j_method_id,(jint)n);

fprintf(m_debug, "called java void method\n" );
fflush(m_debug);

... and the last entry in debug text file is "calling java void method".

But while playing with both, C++ and Java code to get closer to the reason of this error, I discovered that removing

this.debugOutWriter.write("testMethod called, n = " + Integer.toString(n) );
this.debugOutWriter.write("----------------------");

from the Java code solves the problem. Funny thing is, the file access does not seem to be the problem because the lines

this.debugOutWriter.newLine();
this.debugOutWriter.flush();

remain as they were and the "CR/LF"-newline is written into the file correctly. But using the write method is causing the error.

Even if this is only a debug output that won't be required in the final version, I'm still curious about it's origin.
File writing operations won't be required for real functionality, but file read operations and the Java2D framework, and I doubt that this is just a special attitude of the BufferedWriter class..

Richard MacCutchan19-Dec-09 7:27

19-Dec-09 7:27

Well, I'm afraid I cannot figure out what is wrong with any of your code, it all looks fine to me. Sorry, but I'm unable to offer any further suggestions.

TheInfernalCrow21-Dec-09 3:00

21-Dec-09 3:00

Thanks for your efforts anyway Smile | :)

I think I just figured out the problem:

The application I'm developping the DLL for seems to be multi-threaded and up to now I didn't attach/detach the current thread to the JVM.
Since the job to be done is linear and there's no gui I didn't thought of that (JVM is created in constructor, method calls are in the interface-method).

I have no idea why the void method without the write operations as well as the object returning method that's called earlier worked just fine without attaching the current thread (and this manner is reproducable!), but it's all working now Smile | :)

Once again, thanks for your effort!

Richard MacCutchan21-Dec-09 5:57