|
Thank you very much! That seemed to do the trick! Now, just to get it to actually handle the message is another story.
Chalk another one up to good'ol Code Project! Only if you guys know how often this site has saved my @$$. It has been long quest for this old-time Unix developer to learn and develop programs under Windows.
|
|
|
|
|
Hi all,
so im writing a program that install itself as a process, to do this I need the path to process that is running, because my code installs the running process as the service.
Problem:
I cant seem to figure out a way to get the path the running process.
If I call GetCurrentDirectory, it will give me the directory from where the process was run which may not necessarily be the path to the process. furthermore it doesnt include the name of the executable in the path.
Also if I use argv[0], it will only work if I refer to the full path of the executable, which I dont always want to do!
Can someone help me find a way to find the path at which my process is running so that I can use that path to register the running process as a service?
|
|
|
|
|
|
Using MFC:
<br />
CFile file;<br />
file.Open(_T("."),CFile::modeRead);<br />
<br />
MessageBox(file.GetFilePath());<br />
<br />
file.Close();<br />
Ivan Cachicatari
www.latindevelopers.com
|
|
|
|
|
char *dir_buff = (char*)malloc(2048);
GetCurrentDirectory(2048,dir_buff);
MessageBox(0,dir_buff,dir_buff,0);
free(dir_buff);
Ivan Cachicatari
www.latindevelopers.com
|
|
|
|
|
Hi,
My application is keeping track of the coordinates of the cursor through a hook that fires a WM_MOUSEMOVE message and I get the mouse coordinates with GetCursorPos. When I later attempt to reproduce this movement with SetCursorPos, the movement is very jumpy, as very few points are recorded. Just calling GetCursorPos every 1 ms doesn't improve the resolution at all. How can I reproduce a smooth movement of the cursor at a later time?
Thanks,
Aaron
modified 12-Jul-20 21:01pm.
|
|
|
|
|
Hi,
You might want to check and make sure your hook is properly set up to receive all WM_MOUSEMOVE messages from the target. I'm not sure why you mentioned a timer, because when you have a hook properly installed, the message handler should get called automatically whenever the message comes in.
Take a look at this article: http://www.codeproject.com/atl/MouseGestures.asp[^]. It implements a hook to capture mouse messages as well as then performing calculations on the data stored from the mouse messages.
Hope that helps!
Sincerely,
Alexander Wiseman
Est melior esse quam videri
It is better to be than to seem
|
|
|
|
|
The hook seems to be fine, it catches all the other messages. But it appears, and I think I've read, that a mouse move message doesn't get sent for every fraction the mouse is moved. So I'm wondering how to make it smooth.
modified 12-Jul-20 21:01pm.
|
|
|
|
|
Why use GetCursorPos() at all, when the lParam of the WM_MOUSEMOVE message provides the same information?
Since Windows isnt an RTOS, you should not use your current tech. Instead, use the pox passed in WM_MOUSEMOVE, which should be good as long as you get enough of the W_MM messages.
Aaron Stubbendieck wrote:
But it appears, and I think I've read, that a mouse move message doesn't get sent for every fraction the mouse is moved
This may be a possibility, since windows is really designed to be functional, more than "perfect". It may have been one of those design trade-offs done to improve things in general...
Bikram Singh
|
|
|
|
|
I just changed it to use the lParam values, and the performance is identical. Its hard to tell from Spy++, but I'm guessing that there isn't a message sent for each movement. I think it can be done, cause apps like VNC product a smooth movement on the client, and I just want to do that locally. I've looked at the vnc source code and can't find anything magic that they did.
modified 12-Jul-20 21:01pm.
|
|
|
|
|
Jumpy cos no data, possible.
This is interesting, needless to say, i will try this in the next couple of days.
What is VNC? And if you have the sources for it, and it works as you desire, could i see some of it?
Bikram Singh
|
|
|
|
|
That'd be nice if you could take a look at it.
VNC stands for Virtual Network Computing and allows you control and see the screen of a remote computer. You can get the sources here[^]. When a remote user moves the mouse, the movement is smoothly reproduced on the host machine. The code in the server that set the cursor pos doesn't seem any more sophisticated than SetCursorPos.
modified 12-Jul-20 21:01pm.
|
|
|
|
|
I tried to quickly implement a basic dialog-based application which recorded mouse movement for about 2 seconds and then replayed the result. Oddly enough, the movement was smooth and exactly what I did during the recording period. Granted this wasn't through hooking, but I think it proves that one should be able to get every mouse movement, or at least a sufficient number to reproduce a smooth motion.
Here is the main code for recording and replaying ('points' is a CArray of CPoints and 'pt' is a CPoint, both defined in the class definition):
<br />
void CMouseMoveDlg::OnMouseMove(UINT nFlags, CPoint point) <br />
{<br />
if(bStarted)<br />
{<br />
GetCursorPos(&pt);<br />
points.Add(pt);<br />
}<br />
<br />
CDialog::OnMouseMove(nFlags, point);<br />
}<br />
<br />
void CMouseMoveDlg::OnReplay() <br />
{<br />
for(int h = 0; h < points.GetSize(); h++)<br />
{<br />
pt = points.GetAt(h);<br />
SetCursorPos(pt.x, pt.y);<br />
Sleep(10);<br />
}<br />
}<br />
If you would like I could have a look at your code. If you wish to send it to me (or just the relevant portions) my email address is Alexander@wisemanweb.com. I will try to implement this logic inside a hook as soon as I have a chance.
Sincerely,
Alexander Wiseman
Est melior esse quam videri
It is better to be than to seem
|
|
|
|
|
Okay wow. I was just sitting here, reading MY OWN CODE, and something hit me. Did you actually check the data stored in your data structure to see whether all of it was there, or were you just going by what it looked like? If you look at my post immediately before this, I added a call to the Sleep function in order to give a little bit of spacing between the SetCursorPos() calls. It could be that all of the data is there but you don't see smooth mouse movement because you are calling SetCursorPos too quickly. Try adding something like Sleep(10) and see if the motion is smooth. If that doesn't work, check your data structure (print it to the screen or something) and check the results against a Spy++ on the Window.
If all else fails, I can look at your code if you would care to send it to me.
Sincerely,
Alexander Wiseman
Est melior esse quam videri
It is better to be than to seem
|
|
|
|
|
Wow! that was it! Just adding a Sleep(1) call after every SetCursorPos call, made it work just fine. I guess all it needed was a moment to draw it.
Thanks!!
modified 12-Jul-20 21:01pm.
|
|
|
|
|
Cool! Glad I could help.
Sincerely,
Alexander Wiseman
Est melior esse quam videri
It is better to be than to seem
|
|
|
|
|
try using some sleeps
or threads
gabby
|
|
|
|
|
I believe its jumpy because there isn't enough data.
modified 12-Jul-20 21:01pm.
|
|
|
|
|
mmm
i think you're right
gabby
|
|
|
|
|
Hello,
This time the question if about threading... hope you'll appreciate it.
Well then, I have a worker thread being created at some point in my App. It happens that this [new] thread creates about 5 modeless dialogs which continuously output data. This thread was derived from CWinThread ; Everything fine so far...
Now, this is where it gets tricky: inside the InitInstance I initialise a class, which has no links whatsoever to a message pump nor to the main frame window. Moreover, it has to execute in a kind of a loop. Here's an example:
BOOL CClass::InitInstance()
{
CG1 theG1;
theG1.Run();
return FALSE;
}
This was my first (frustrated) attempt and I quickly learned that whenever the instruction execution reached inside theG1.Run , there were no Windows Messages being received/processed.
So I made some changes. Instead of having it run inside InitInstance , I implemented a solution with OnIdle . I don't really like it, but let's see how I did it:
CClass::OnIdle(long nCount)
{
theG1.ExecuteNext();
return TRUE;
}
Now my question:
Is is possible to make the class run in a real multi-threading environment rather than in OnIdle ? I know I would have to somehow make a connection with the thread's Message Pump... but how?
Any suggestions would be greatly appreciated,
David
PS: In case you got confused with the methods I'm using above (Run ; ExecuteNext ) , I am working on an academic project aimed at emulating a processor on a x86 machine.
|
|
|
|
|
dNimrod#X wrote:
I have a worker thread being created at some point in my App. It happens that this [new] thread creates about 5 modeless dialogs...
This raises a lot of red flags, namely that worker threads should not create UI components. If you must create a UI component in a thread other than the primary thread, it should be a UI thread (i.e., thread with a message pump). See here for more.
"When I was born I was so surprised that I didn't talk for a year and a half." - Gracie Allen
|
|
|
|
|
Hi,
I've been reading on the subject and decided to follow your cue. You're absolutely right -- there is the need to have a message pump thread.
Thanks very much for the reply. You have been very helpful indeed.
David
|
|
|
|
|
Hi,
I'm developing a database application using CDatabase & CRecordSet classes. My SQL Server database
is sitting on other machine and it's effecting application performance when I tried to insert large number of records
and retrieving the records for generating some reports(i.e. to fetch 1000000 records at a time for creating a report). Application is getting too slow.
Need some suggestions in this regard.
Thanks,
|
|
|
|
|
Hi,
probably instead of using mysql_store_result you should using mysql_use_result. It will give you a boost, but also a lot of drawbacks. See the C-API documentation for that.
GSte
|
|
|
|
|
Hi,
I recommend to segment this query or install more physics memory.
navisoft
www.latindevelopers.com
|
|
|
|