C / C++ / MFC

Cool_Dev5-Aug-10 22:48

Cool_Dev

5-Aug-10 22:48

http://support.microsoft.com/kb/238975

Office applications typically registers their the first instance only in the ROT. but they are supposed to register their individual documents in the ROT, so iterating through the ROT using IRunningObjectTable Interface may be the option to get the instance you want, provided you know the document which is opened with that instance.

trioum5-Aug-10 23:54

trioum

5-Aug-10 23:54

can you help me with some source code

Trioum

Cool_Dev6-Aug-10 1:08

Cool_Dev

6-Aug-10 1:08

eventhough multple instances of excel are running, only the first instance would have an entry in ROT. But all the instances registers the documents opened in them in ROT. Its the key to bind to that instance of excel.
say, u have more than one instances of excel apps, and want to get the instance in which "MyExcel.xls" file is opened..

void GetMyExcelInstance()
{
	CApplication oExcelApp; //class wizard generated class for excel app

	CoInitialize(NULL);
	HRESULT hr;

	//Setup ROT
	IBindCtx *piBindCtx = 0; 
	hr = CreateBindCtx( 0, &piBindCtx ); 

	IRunningObjectTable *piROT = 0;
	hr = piBindCtx->GetRunningObjectTable(&piROT);

	//Enumerate registered objects
	IEnumMoniker *piEnum = 0;
	hr = piROT->EnumRunning(&piEnum);

	piEnum->Reset();

	ULONG nFetched;
	IMoniker *piMoniker = 0;

	bool bGot = false;
	while(piEnum->Next(1, &piMoniker, &nFetched) == S_OK && !bGot) 
	{
		// Get Display name, in our case can be the workbook (excel doc) name
		LPOLESTR strName;
		hr = piMoniker->GetDisplayName(piBindCtx, NULL, &strName);
		CString csName(strName);
		CoTaskMemFree(strName);

		if(csName.Find(L"MyExcel.xls") != -1) //or you may compare against full path of MyExcel.xls file
		{
			//Bind to the object. In this case object is excel workbook.
			IDispatch *piDispObj = 0;
			hr = piMoniker->BindToObject(piBindCtx, NULL, IID_IDispatch, (void**)&piDispObj);

			CWorkbook oWorkBook(piDispObj); //class for excel workbook object
			oExcelApp = oWorkBook.get_Application();
			
			bGot = true;

		}
		piMoniker->Release();
	}

	piEnum->Release();
	piROT->Release();
	piBindCtx->Release();

	if(bGot)
	{
		//do here with oExcelApp 

	}
}

do proper error checking.. Thumbs Up | :thumbsup:

trioum6-Aug-10 19:12

trioum

6-Aug-10 19:12

//code is ok up to here
CWorkbook oWorkBook(piDispObj); //class for excel workbook object

// but as I called the following line errro occurs member not found
oExcelApp = oWorkBook.get_Application();

Trioum

omp parallel for reduction (+: sum) computation discrepancies?

Cool_Dev8-Aug-10 0:22

Cool_Dev

8-Aug-10 0:22

did you generate wrapper classs for excel type library? those class members may vary in different visual studio versions. check your class for WorkBook object to see how it returns
Application object. it may be by get_Application() or GetApplication() or by some other means...

Chesnokov Yuriy5-Aug-10 20:22

Re: omp parallel for reduction (+: sum) computation discrepancies?

5-Aug-10 20:22

Using reduction to estimate sum of array elements leads to results differences for more than 1 thread in parallel region

//values in vectors are random between -0.5 ... 0.5
#pragma omp parallel for reduction (+: sum)
        for (i = 0; i < N * N; i++) {
                vec3[i] = vec1[i] * vec2[i];
                sum += vec3[i];
        }

The difference is:

sum = 101.759239; // no open mp or single thread
sum = 101.764313; // more than 1 open mp threads

Чесноков

Peter_in_27805-Aug-10 21:08

Peter_in_2780

5-Aug-10 21:08

Looks like you're using float (not double) and suffering from loss of precision adding big and small numbers.
Simple example:

float sum1 = 0.0, sum2 = 1.0;
for (i = 0; i < 10000000; i++)
  {
  sum1 += 1e-8;
  sum2 += 1e-8;
  }
sum1 += 1.0;
// here, there's no way sum1 == sum2

In a quick test, I got sum1 = 1.092107... and sum2 = 1.0
However, replacing float with double, I get sum1 = 1.09999999998803 and sum2 = 1.09999999939225.
You can get almost any answer you want in float by choosing the order of evaluation. Poke tongue | ;-P

Software rusts. Simon Stephenson, ca 1994.

Re: omp parallel for reduction (+: sum) computation discrepancies?

Chesnokov Yuriy5-Aug-10 21:59

How to change audio output/input devices?

5-Aug-10 21:59

yes, after I replaced with doubles results were equal

Чесноков

odyaho5-Aug-10 17:29

odyaho

5-Aug-10 17:29

Dears
There are two audio output devices :
1. Bluetooth SCO Audio
2. Realtek HD Audio output

And if bluetooth headset connects to PC, the default output device is item 1.
I want to change to item 2. I have tried some methods, but it still unsuccess
I know MCI functions should done it, but I am not sure which functions should use..
Attachment is test code below:

#include <windows.h>
#pragma comment (lib, "winmm.lib")
#include <mmsystem.h>
#include <iostream>
#include <stdio.h>

int main()
{
	int AIDChoise;
	UINT nDevices = waveInGetNumDevs();	// Get No. of devices
	MCI_WAVE_SET_PARMS parms;
	DWORD rc;
	TCHAR ErrMsg[256];
	
	WAVEINCAPS inwcps;
	WAVEOUTCAPS outwcps;

	for (int k=0; k < nDevices; k++)	// Get input audio devices
	{
		waveInGetDevCaps(k,&inwcps,sizeof(inwcps));
		printf(" In devname %d : %s \n",k,inwcps.szPname);
	}
	for (k=0; k < nDevices; k++)	// Get output audio devices
	{
		waveOutGetDevCaps(k,&outwcps,sizeof(outwcps));
		printf(" Out devname %d : %s \n\n",k,outwcps.szPname);

	}
	printf("Please select output device :");
	scanf(" %d",&AIDChoise);
//	getchar();
	  
	switch(AIDChoise)
	{
	case 0:
		printf("0 ~~~~~~~~~");
		parms.wOutput = 0;
		rc = mciSendCommand(0, MCI_SET, MCI_WAVE_OUTPUT, (DWORD)&parms);
		if ( rc != 0 )
			printf("Error Msg : %s \n",mciGetErrorString(rc,ErrMsg,256));
		break;
	case 1:
		printf("1 ~~~~~~~~~");
		parms.wOutput = 1;
		rc = mciSendCommand(1, MCI_SET, MCI_WAVE_OUTPUT, (DWORD)&parms);
		if ( rc != 0 )
			printf("Error Msg : %s \n",mciGetErrorString(rc,ErrMsg,256));
		break;
	default:
		printf("2~~~~~~~~~");
		printf("Invalid command.. \n");
		break;
	}


	return 0;
}

Re: How to change audio output/input devices?

odyaho6-Aug-10 0:34

odyaho

6-Aug-10 0:34

yes, I found out...
use "waveOutMessage" function can change audio device..

IAccessible and tools like AccExplorer / Inspect Objects

Member 38216205-Aug-10 14:53

Member 3821620

5-Aug-10 14:53

Hi,

I would like to use MSAA to access some user interface elements belonging to other applications.
According to tools like AccExplorer or Inspect Objects, the elements I'm interested in are reachable and listed in the tree hierarchy they display.

However, when I use a custom function like WalkTreeWithAccessibleChildren from MSDN (<a href="http://msdn.microsoft.com/en-us/library/dd317975(v=VS.85).aspx">http://msdn.microsoft.com/en-us/library/dd317975(v=VS.85).aspx</a>[<a href="http://msdn.microsoft.com/en-us/library/dd317975(v=VS.85).aspx" target="_blank" title="New Window">^</a>]) I get a different tree, some elements are identical but other have a complete different set of children. The elements I want to access are not available in this tree (I've also tried with methods like get_accChild and accNavigate but it did not work).

Do you know how I can build a tree similar to the one in AccExplorer?

Thank you for your help!

Win32 Scheduled jobs

Jach Mullan5-Aug-10 11:52

Jach Mullan

5-Aug-10 11:52

Hi,
is there any difference between Windows Job and Scheduled Tasks?

I have used the win32_ScheduledJob class for monitoring the scheduled tasks.
But it doesn't return anything even that task in running state.

Please let me know how to create a windows job manually for testing and which win32 class to be used for monitoring?

Thanks

Re: Win32 Scheduled jobs

Yusuf5-Aug-10 12:15

Yusuf

5-Aug-10 12:15

The Win32_ScheduledJob WMI class represents a job created with the AT command. The Win32_ScheduledJob class does not represent a job created with the Scheduled Task Wizard from the Control Panel. You cannot change a task created by WMI in the Scheduled Tasks UI.[^]

Yusuf
May I help you?

winsock connect() and how to reduce intial timeout value from, 3 seconds to 1 second or less

Alan Kurlansky5-Aug-10 10:29

Alan Kurlansky

5-Aug-10 10:29

If possible what is the easiest way to reduce the connect inital timeout value from 3 seconds to 1 second or less? The excerpt below reduced the retries which is not good enough for my purposes.
Any suggestions would be appreciated.
Thanks

"Now, the initial timeout value - before the first connection attempt - is 3 seconds. After each retry, the amount of time to wait is doubled (3 seconds, 6 seconds etc.). The number of retries can be configured via the 'TcpMaxConnectRetransmissions' parameter in the registry - the default is set to 3 for Windows NT and 2 for Windows 2000. Thus, the first solution to the connection timeout problem is, to adjust this parameter via the Registry. However, changing the registry is a global change, affecting all applications that use Winsock and/or TCP/IP. The following Knowledge Base article (http://support.microsoft.com/default.aspx?scid=kb;en-us;120642) provides detailed information. Nevertheless, one should be careful while playing with these parameters."

Re: winsock connect() and how to reduce intial timeout value from, 3 seconds to 1 second or less

Moak5-Aug-10 12:50

Moak

5-Aug-10 12:50

Alan Kurlansky wrote:
what is the easiest way to reduce the connect inital timeout value from 3 seconds to 1 second or less

There is no 3 second timeout for connect() function. You need to implement your own timeout mechanism or use a socket library that does. It's a short answer, hope it helps. Smile | :)

Chat in Europe Java | [Coffee]

Now with 24% more Twitter

input message injection on windows logon desktop

linuwin5-Aug-10 3:47

linuwin

5-Aug-10 3:47

On a windows(XP) machine i'm trying to inject mouse/keyboard messages (for some remote control; kind of soft-KVM-switch), which works fine as long as a user is logged in. Once the workstation is locked or no user is logged in, the injected messages seem to be ignored. I have tried both mouse_event and SendInput (unsuccessful) as well as to connect the app (runs as nt-service, userinteraction allowed) with the logon-desktop, which (regarding to api-returns) should work, but still no success. Can anyone give me an advice?

OpenMP concurrency issues

Chesnokov Yuriy5-Aug-10 0:53

Re: OpenMP concurrency issues

5-Aug-10 0:53

There is [input data] <---- object ----> [output data] program structure.
Input data is processed in a for loop with specific object (class or struct) which takes chunk of input data performs computation and saves results to output data array.

To convert it to OpenMP should I have as many object references allocated as presumed number of threads will be running to avoid its simultaneous access from multiple threads?

It will not be possible to lock the object for a single thread as the multithreading nature dissapears since only one thread at a time will be allowed to perform computation with that object.

Чесноков

Aescleal5-Aug-10 2:21

Aescleal

5-Aug-10 2:21

It depends on what your object is. If the object doing the processing is a stateless functor (ideally) then just make one per thread and don't worry about locking it. However if it's got state that's relevant across the whole operation of your program you're going to have to share the object and lock it.

Cheers,

Ash

OpenMP performance inconsistency on i7

Chesnokov Yuriy5-Aug-10 0:48

5-Aug-10 0:48

I'm running that code on i7 computer with 4 cores and 8 threads

N = 10000;
...
#pragma omp parallel for
for (i = 0; i < N * N; i++)    
    vec3[i] = vec1[i] * vec2[i];
...

Setting number of threads with omp_set_num_threads to 2, 4 and 8 the performance increase compared to omp_set_num_threads(1) is not consistent: x1.4, x1.55, x1.58

Should be we expect the rates to be: x2, x4, x8 instead?

Чесноков

Sauro Viti5-Aug-10 0:58

Sauro Viti

5-Aug-10 0:58

If you have 4 cores, the highest ratio that you can theoretically expect is 4; using a number of threads higher thatn the number of cores don't produce significant advantages and can reduce the performances because to execute all of them the system has to switch the thread context a lot of times and this will waste time.
The real speed improvement could be less than expected due to the times required to create the threads at the beginning of the parallel construct and destroy them when the parallel construct terminates; if this is the case, by increasing N the rates should became better (e.g. try N = 1000000)

Aescleal5-Aug-10 2:14

Aescleal

5-Aug-10 2:14

Hi Sauro,

the i7 the original poster is using can run 8 threads simultaneously. There are 4 cores and each core is hyperthreaded.

Cheers,

Ash

Sauro Viti5-Aug-10 2:34

Sauro Viti

5-Aug-10 2:34

Ops...

harold aptroot5-Aug-10 13:54

harold aptroot

5-Aug-10 13:54

The HyperThreading performance is not actually double the normal performance though, many parts of the CPU are shared competitively between the two threads running on a single core.