Thanks guys for those explanations. Now I understand what is going on
Cheers, Greg On Thu, Dec 16, 2010 at 7:25 AM, <[email protected]> wrote: > This is why it depends on exactly which memory you’re measuring – linking in > a library DOES use memory, because of the way virtual tables work in C++: > just “knowing about” the type – i.e. the metaobject information, even with > no instances, means the vtable must be loaded, and that of course has > references to all the virtual functions (often much of the class), in > addition to all signals and slots and property accessors in Qt’s metaobject > info. > > > > Importing a module means making types available, and that means loading and > processing. While in theory QML could delay some of that until it comes > across a type it doesn’t know about yet, that seems a pointless > optimization. > > > > So if the memory is mainly read-only shared code pages, there is nothing > unexpected happening. > > > > As for documentation, this is internal detail that is not really relevant to > QML authors, and could in theory change at any time, so not something anyone > should be relying on one way or another. > > > > -- > > Warwick > > > > > > I'm sharing those numbers not because I think they are somehow too high, or > too low, but because I am surprised to see memory usage *grow by a > significant amount* even when I'm *not instantiating* anything. > > > > Put it another way: the fact that a single "import WebKit 1.0" line adds 10 > MB of memory usage, even though I'm NOT instantiating any WebView element is > an unexpected behaviour to me, and may indicate either a bug with the way > QML deals with imports, or a normal but unexpected behavior that should be > explained / documented. > > _______________________________________________ Qt-qml mailing list [email protected] http://lists.qt.nokia.com/mailman/listinfo/qt-qml
