This problem might be too elaborate to try and get an answer to, but its quite perplexing. In short,my applications screen updates occasionally seem to freeze--I'd swear my application was hung up or blocked on a read or something, because all screen writes from my app stop, both graphics and text. However, I know my application is still running--I am controlling an underwater vehicle, and it continues to perform well--we can drive it about, maintain heading and depth control, etc, but the app's screen updates just quit. All I have to do, however, is move the mouse into my apps window, and graphics updates resume. The freezes are not at a predictable rate--but do seem to get worse over time if I don't restart the computer (not the app) for a while.
Environment: QT4.6, Windows XP SP2, Nvidia graphics card
My application that is doing lots of network IO (UDP sockets), both reading and writing to half a dozen or more separate sockets.
I have at least half a dozen QTimers running, some at relatively high rates, one at 100 hz or so, the rest at around 10 Hz.
There is some evidence that the freezes are related to which of several QTabWidgets have been selected--there is one that displays diagnostics from a tilt unit that I am controlling with a joystick. If the diagnostic tab is showing, then I can get the application to "unfreeze" its graphics by moving the joystick, and hence the tilt unit--and causing update of a QLabel on the tabbed widget. However, if that tab is not showing, the tilt unit seems to not be under control at all--joystick moves do nothing to it. It's the only thing that doesn't work in the application when the screen updates have stopped. I get joystick data via a udp port--however, I know there is nothing wrong with the joystick or UDP port, because the joystick is on a control box that has several other joystick and button devices; they all work fine, and the data from all of them comes in a single UDP datagram. It's one message that is obviously parsing OK, because if they don't all parse, I ignore the message totally.
Note that other parts of my application are continually updating QLabels and qwt dials, etc.
There is some other evidence that if some other tabs are showing, I never get the problem, but that's not conclusive.
Nor is it conclusive that when I run the code on a laptop (using QT 4.5.2) the problem doesn't seem to occur. I can't say it never occurs, I can only say that it didn't
Sorry to be so vague, but the relatively complex nature of the vehicle data presentation and control and the sporadic appearance of the problem mean that I am sure I don't totally understand all of its aspects. Unfortunately, getting the vehicle into the water is a rare opportunity, and testing time is difficult to obtain, so that's the best I can do right now.
My question is whether anybody can provide any insight into what kind of problem will cause QT updates to a window to freeze, and be restarted by the window gaining context--or perhaps by some other combination of QLabel text updates. That might give me some idea of where to try and gather more data--or even to try for a work around.
Thanks for any insight
Jonathan Howland
