|
I have a dialog-based MFC application within which I need to do some serial I/O. Basically, I need to wait for a certain character to come in on the serial port. I'm successfully creating a file for the port in overlapped mode and starting ReadFile with a valid overlapped structure and event handler. Also, I'm using WaitForSingleObject() with a short timeout to check whether the event has been set or I've timed out.
What I want to do is cause my dialog to perform message handling (specifically redrawing) whenever WaitForSingleObject times out. That way the user will see a properly repainting display while we're waiting for the serial characters instead of what looks like a hung application. Any suggestions would be greatly appreciated.
Best Regards,
pHaze426
|
|
|
|
|
Put the serial port stuff in a separate thread and the GUI will remain responsive.
"When I was born I was so surprised that I didn't talk for a year and a half." - Gracie Allen
|
|
|
|
|
I could do that, but I'd still have to have the function launching the thread wait until the serial read completed before continuing with its tasks. Otherwise, I'd have to hook a callback function or something to handle a state machine that's called after each read. I just wondered if there was any other option as far as a function I could call to do processing while I'm waiting.
Best Regards,
Phaze426
|
|
|
|
|
pHaze426 wrote:
I could do that, but I'd still have to have the function launching the thread wait until the serial read completed before continuing with its tasks.
Nothing wrong with that (except for the "wait" part). Once the secondary thread has completed, it would post a message to the primary thread indicating it is done. At that point, the primary thread could do what it needed with the serial port data. No waiting is necessary. Make sense?
"When I was born I was so surprised that I didn't talk for a year and a half." - Gracie Allen
|
|
|
|
|
pHaze426 wrote:
I'm using WaitForSingleObject() with a short timeout to check whether the event has been set or I've timed out.
How are you doing that continuously without multiple threads and without blocking the UI thread? In a mesasge loop?
pHaze426 wrote:
What I want to do is cause my dialog to perform message handling (specifically redrawing) whenever WaitForSingleObject times out
Then use SendMessage() or PostMessage()
"No matter where you go, there your are." - Buckaroo Banzai
-pete
|
|
|
|
|
I guess I am using threads in a roundabout sort of way. When I call CreateFile to open the serial port, I specify the OVERLAPPED I/O flag. Then, when I call ReadFile() I pass it an overlapped structure with an event handle. When it's done, it sets that event handle. (I guess this means that internally, Windows is spawning another thread for ReadFile). Anyhow. WaitForSingleObject just waits until the event handle is set, or its own timeout has expired. So... if I give the call to ReadFile a timeout of 10 seconds and the call to WaitForSingleObject a timeout of 100ms, I will get a chance to do something every 100ms until the serial I/O completes, or readfile times out after 10s.
Along the SendMessage lines, I've used:
this->SendMessage(WM_PAINT, 0, 0); from within this wait loop to no avail. Is there another message I should be sending?
|
|
|
|
|
Well if you don't need to send any data the just call CWnd::Invalidate() that will cause a Paint to occur.
"No matter where you go, there your are." - Buckaroo Banzai
-pete
|
|
|
|
|
Just for everyone's reference, I figured out a way to do this (granted, this way isn't all that clean and is probably more of a 16-bit windows style of doing things, but it works). Rather than using WaitForSingleObject to wait for the I/O to complete, I stuck in this:
while (MsgWaitForMultipleObjects(1,
&ol.hEvent,
FALSE,
10,
QS_ALLINPUT) !=
WAIT_OBJECT_0) {
while ( PeekMessage ( &msg, NULL, 0, 0, PM_REMOVE)) {
DispatchMessage ( &msg);
}
}
So... I just keep dispatching all the messages I received until the serial even becomes signalled. This seems to work well.
|
|
|
|
|
Ok, I have a CArray container like so:
<br />
CArray <myObject, myObject> objectList;<br />
and I declared a dynamic array in a separate class like so:
<br />
class SelectionBox<br />
{<br />
public:<br />
void AddContent(myObject *);
<br />
protected:<br />
myObject *content;
unsigned long numContents;
}<br />
So now I built my AddContent() function like this:
<br />
void SelectionBox::AddContent(myObject *c)<br />
{<br />
unsigned long counter;<br />
myObject *tempObject;<br />
tempObject = new myObject[numContents];<br />
<br />
tempObject = content;<br />
<br />
content = new myObject[numContents + 1];<br />
<br />
for(counter = 0; counter < numContents; counter++)<br />
{<br />
content[counter] = tempObject[counter];<br />
}<br />
<br />
content[numContents] = c;<br />
<br />
<br />
numContents++;<br />
delete [] tempObject;<br />
} <br />
Finally, in another part of the program, I shoot the address of one of the objects stored in the CArray to the AddContent function.
At first the function appeared to be working.. That is until I later tried to manipulate the original objects in the CArray by manipulating what I thought was the pointer to them in my SelectionBox class. Then I found out that I had copies of data stored and not a pointer to the CArray data. I can't for the life of me figure out how to declare the function and pass the data so that I can store an array of pointers/addresses to the data in the CArray container. Can someone show me an example, or maybe a link to a relevant tutorial?
|
|
|
|
|
|
That worked perfectly!!!! Thanks Pete.. I didn't even know that container existed, .. which will help a quite a few more problems I've been facing. Thanks Again 
|
|
|
|
|
Hello,
I am writing a program that needs to do diffrent stuff depending on which OS it's running.
How can I do this?
Can I do it without GetVersionEx?
Can I do it with #if(WINVER >= #)? If so what values are for which OS?
Thanks.
wWw.KruncherInc.cOm - My cool programs
|
|
|
|
|
KingKruncher wrote:
Can I do it without GetVersionEx?
Why would you want to?
KingKruncher wrote:
Can I do it with #if(WINVER >= #)? If so what values are for which OS?
See here for values.
"When I was born I was so surprised that I didn't talk for a year and a half." - Gracie Allen
|
|
|
|
|
KingKruncher wrote:
Can I do it with #if(WINVER >= #)?
No, that is a compile time directive not a runtime evaluation.
"No matter where you go, there your are." - Buckaroo Banzai
-pete
|
|
|
|
|
You'll eventually need to compile an exe for each version if you'll use #if(WINVER >= # method
It wont be the same exe
Papa
while (TRUE)
Papa.WillLove ( Bebe ) ;
|
|
|
|
|
|
here you can use my code for this
//structure
struct NmpWindowVersion
{
public:
CString WindowName;//name of windows
int Plateform;//0 stand for single user and 1 stand for multi user
CString WindowsVersion;// contain the iformation abbt service pack etc
CString BuildNumber;
};
////function
int MeCheckVersionInfo(NmpWindowVersion &StructVar)
{
OSVERSIONINFO info;
info.dwOSVersionInfoSize=sizeof(OSVERSIONINFO);
GetVersionEx(&info);
int nMajor,nMinor;
CString BuildStr;
//getting Verion nme
switch (info.dwMajorVersion)
{
case 3: StructVar.WindowName="Windows NT 3.51";
nMajor=3;
nMinor=51;
break;
case 4: switch(info.dwMinorVersion)
{
case 0:
if(info.dwPlatformId==VER_PLATFORM_WIN32_WINDOWS)
StructVar.WindowName="Windows 95";
else
StructVar.WindowName="Windows NT 4.0";
nMajor=4;
nMinor=0;
break;
case 10: StructVar.WindowName="Windows 98 ";
nMajor=4;
nMinor=10;
break;
case 90: StructVar.WindowName="Windows Me ";
nMajor=4;
nMinor=90;
break;
}
break;
case 5: switch(info.dwMinorVersion)
{
case 0: StructVar.WindowName="Windows 2000";
nMajor=5;
nMinor=0;
break;
case 1: StructVar.WindowName="Windows XP";
nMajor=5;
nMinor=1;
break;
case 2: StructVar.WindowName="Windows Server 2003 family 2";
nMajor=5;
nMinor=2;
break;
}
break;
}
////getting Plateform ID
switch(info.dwPlatformId)
{
case VER_PLATFORM_WIN32_WINDOWS:StructVar.Plateform=0; break;
case VER_PLATFORM_WIN32_NT :StructVar.Plateform=1; break;
}
///getting Platform version
StructVar.WindowsVersion=(CString)info.szCSDVersion;
////getting the build number
BuildStr.Format("%d.%d.%d",nMajor,nMinor,LOWORD(info.dwBuildNumber));
StructVar.BuildNumber=BuildStr;
return 0;
}
-----------------------------
"I Think It will Work"
-----------------------------
Alok Gupta
visit me at http://www.thisisalok.tk
|
|
|
|
|
I forgot (like I often do) to add help to my app's settings when creating my project. I'm pretty far along in the project now, so I don't want to start over, so what do I need to do to add help now since the project is already created?
If it's broken, I probably did it
bdiamond
|
|
|
|
|
Anytime you forget to do something with AppWizard, it's usually best to create a temp project using the desired settings, and then compare to see what items your project is missing.
"When I was born I was so surprised that I didn't talk for a year and a half." - Gracie Allen
|
|
|
|
|
yeah, I thought about doing that, but I was hoping there was an easier way!! (just lazy  ) thanks!
If it's broken, I probably did it
bdiamond
|
|
|
|
|
what message handlers uses the min,max and close button
|
|
|
|
|
Handle WM_SYSCOMMAND
and these three values of wParam
SC_MINIMIZE
SC_MAXIMIZE
SC_CLOSE
and right before you "return 0;" you should do this:
SetWindowLong(hwnd,DWL_MSGRESULT,0); //For a dialog box
wWw.KruncherInc.com - My cool programs
|
|
|
|
|
Hi,
I am trying to iterate and delete the elements (assuming atleast 1 element) in a CptrList with this:
void *pPt = NULL;
POSITION pos = NULL;
for (pos = m_points.GetHeadPosition(); !m_points.IsEmpty(); m_points.GetNext(pos)) {
// TRACE("pos: %d, %d\n", pos, NULL);
pPt = m_points.GetAt(pos);
m_points.RemoveAt(pos);
delete pPt;
}
TRACE shows that pos is not equal to NULL as it should be after the last element has been seen.
So pos != NULL check does not work contrary to the MSDN Library documentation. THIS IS ONE PROBLEM.
The OTHER is even if there are more than 1 elements the second element returned by GetAt is not
"valid" (0xcdcdcdcd) and RemoveAt asserts.
Any clues?
Regards
Mahendra
|
|
|
|
|
As soon as you remove the element, you invalidate the position. Here's a better way of doing it:
{
void *pPt = NULL;
POSITION pos = NULL;
for (pos = m_points.GetHeadPosition(); pos != NULL; m_points.GetNext(pos))
{
pPt = m_points.GetAt(pos);
delete pPt;
}
m_points.RemoveAll();
}
"Fish and guests stink in three days." - Benjamin Franlkin
|
|
|
|
|
But is this ok - (re)moving from the tail end if a sub-sequence has to be removed and RQUIRES using
RemoveAt. This seems to works:
for (pos = m_points.GetTailPosition(); !m_points.IsEmpty(); m_points.GetPrev(pos)) {
pPt = m_points.GetAt(pos);
delete pPt;
m_points.RemoveAt(pos);
}
pos is still not NULL in the end though!
Or like in the MSDNL RemoveAt help!:
POSITION pos1 = NULL;
POSITION pos2 = NULL;
for (pos1 = m_points.GetHeadPosition(); (pos2 = pos1) != NULL; ) {
m_points.GetNext(pos1);
pPt = m_points.GetAt(pos2);
delete pPt;
m_points.RemoveAt(pos2);
}
Due Regards
Mahendra
|
|
|
|
General 
News 
Suggestion 
Question 
Bug 
Answer 
Joke 
Praise 
Rant 
Admin 
Submit an article or tip
