> Hello, > > I've been using PyQt for about six months. IMHO, PyQt is *great*, a > perfect > tool for creating first-class portable GUI applications. It is also very > enjoyable to work with Python and Qt, both are elegant and nicely > designed. > > However, I've some issues about the memory usage of PyQt applications, > both > the ones that I wrote and that I use. As an example I've examined the > memory > usage of Eric with K System Guard: > > VmSize VmRss > Initial 49056 35168 > Open Settings Window 51116 37884 > Close & Open Again 52564 39440 > Close & Open Again 54080 40952 > Close & Open Again 55724 42588 > Close & Open Again 57280 44088 > Close & Open Again 58728 45588 > > Both VmSize and VmRss values are constantly increasing. For the test, I > started Eric3, then repeatedly clicked Settings -> Preferences and close > preferences dialog. > > I observe similar memory usage patterns on my own code. As the complexity > of > my dialogs/widgets increase, I see more of incresing memory usage. > > I tried to track down possible leakages on my code by using gc module, > specially gc.get_objects() in Python level. But I still observe increased > memory usage even if gc.get_objects() seems to stay constant. > > I googled and searched various lists about Python, as far as I understand, > Python never releases allocated memory back to the OS, and some objects > like > integers and floats are basically immortal. So it is normal to observe an > "always increasing memory usage" pattern because memory is never released > back. However, once the objects are garbage collected, released memory is > available to be used by new objects. So, for a simple situation like > opening > and closing a dialog, I think the memory usage should remain constant at > *some* point. > > Is this behaviour normal? Can it harm long-running GUI processes? > > Best regards, > Umit OZTOSUN > > System Information: > Intel P4 2.5 GHz, 512 MB RAM > Debian Testing, Python 2.3.4, Qt 3.3.3, PyQt 3.12, SIP 4.0.1, Eric 3.5.0
The current SIP v4 snapshot contains a fix for a memory leak, however it (SIP, not the fix) causes eric problems at the moment. If you can produce a small example that appears to demonstrate a leak then I'll look at it. Phil _______________________________________________ PyKDE mailing list PyKDE@mats.imk.fraunhofer.de http://mats.imk.fraunhofer.de/mailman/listinfo/pykde