David's pointing you to the spec is an excellent answer.
A few things I will add to the mix:
* You will need multiple commands ( SECURITY ERASE PREPARE, SECURITY ERASE UNIT, and possibly other SECURITY commands ) from the spec in order to accomplish a secure erase.
* Sending (single-sector) ATA passthru commands from uesr mode applications is possible on Windows (XP SP2 and newer, IIRC), assuming you have Admin privileges on the system. You will need do some Googling on IOCTL_ATA_PASSTHRU_DIRECT in order to find relevant information.
* That being said, you will still be at the mercy of the system BIOS and possibly the O/S. Once a SECURITY FREEZE LOCK is sent to the drive, it is *generally* difficult or impossible to send SECURITY ERASE to the drive. Some motherboard BIOSs send the SECURITY FREEZE LOCK prior to start of boot just to ensure a drive is not accidentally or maliciously erased.
All of this is from my memory, which is several years old at this point.
Have a look at this article HOWTO track a user's idle time[^]; it seems to have a good implementation on how to detect the idle time.
When the idle time (five minutes) has been reached you could show a modal dialog that requires the user to specify his/her user name and password.
I have an application that can also run as a service via srvany. It performs a monitoring task and uses the PlaySound method defined in mmsystem.h to play a .wav notification on certain events. It runs fine under XP as an app or as a service.
Under Win7, the app runs fine. But when running as a service under Win7, I get no sound. The "allow service to interact with desktop" option is available but ignored under Win7 as part of Session 0 isolation. I expect this is somehow related.
I'm looking for verification or alternate approaches.
You are most likely right in your conclusion. I think the session 0 isolation prevents such interactions.
A possible work-around is to have your Service start or interact with a non-service component that runs in the user session.
In a Debug build the compiler performs additional tasks, for instance it zeroes variables you haven't initialized (e.g. WinAPI structs) (in Release build the unitialized data contains garbage).
Have a look at the following CodeProject article: "Surviving the Release Version"[^].
Check your CMyApp::OnFileOpen() code and the functions that are called from there. Comment all lines in that function except the base class CWinApp::OnFileOpen() call and uncomment step by step until the violation occurs. Then you should know which part of your code is responsible.
If you did not call CWinApp::OnFileOpen(), no document is created. So the reason for the crash is probably somewhere in your document creation. Did you call OnFileNew() or use ProcessShellCommands() to create a document?
No, I didn't call CWinApp::OnFileOpen, but I call OpenDocumentFile(...), something like that:
if(IDOK == dlg.DoModal())
POSITION pos = dlg.GetStartPosition();
// the document is loaded, I can see the bitmap in child frame ...
AfxMessageBox("By here, everything it's OK.");
after all, I think that the problem it's in another place ... I really don't know ...
And this is weird, because if I open the same file from MRU, everything is OK ...
It's difficult to find such errors in release versions. Using a message box is a good idea to check where the violation occurs. Because it occurs when not calling OpenDocumentFile(), I assumed that it is somewhere in the document handling (e.g. trying to access non-existing documents).
But when it is OK using ProcessShellCommands() (MRU)
At the moment I have only two ideas:
Call OpenDocumentFile() with a fixed file name to see what happens (no other calls from within OnFileOpen()).
Perform a complete rebuild if you have changed something in your project.
Of course I did ... something weird is happend ... because if I use only CFileDialog dlgFile(TRUE); in CMyApp::OnFileOpen and the application is crashing down .. I wonder if CFileDialog doesn't have some bug ...
Sorry, I did not mean to insult you, but when something really weird is going on, Rebuild All is a good thing to do. I wanted to make sure you had tried that because just last week, I had a weird string related problem that I was trying to chase down. Rebuild All was not my first thought, so I ended up spending at least an hour trying to make sense of it before I finally did the rebuild, which fixed the problem.
"When you don't know what you're doing it's best to do it quickly" - Jase #DuckDynasty