Jean-Marc Lasgouttes wrote:
"Peter" == Peter Kümmel <[EMAIL PROTECTED]> writes:Peter> I've reproduced the error and fixed it. The problem was, not Peter> only page up/down keys were dropped. This code does not work Peter> (because of the implicit casts?) Peter> static const int delayed_keys = Qt::Key_PageDown | Peter> Qt::Key_PageUp; if (e->key() & delayed_keys) { I know this stuff is very difficult to get right, but I do not like at all the idea of testing explicitely qt::pageup/down. We should not depend on these hardcoded keys. For example, your patch will fail for people who use the emacs bindings and use Ctrl+v for page down. I am not asking for adding another case for Ctrl+v (!) but I'd prefer a solution which does not take this into account (especially since I do not understand why only PageDown would be a problem (why is paragraph down not a problem?) JMarc
Peter, there's one thing I don't understand regarding this whole issue: doesn't holding the key down generate only *one* "key-down" and *one* "key-up" event? Whereas just pressing a key without holding it geenrates a single "keypress" event for every time it's pressed? (I don't know of these are the correct terms, I don't know Qt at all, but the idea is probably the same). So couldn't we use that to differentiate? In other words, we should process every keypress event, and if that takes longer than it takes the user to generate them -- well, we should still process each one, because otherwise characters get swallowed; and if the keypresses are navigation keys, we should process those too --- every keypress the user presses means something. OTOH, if the user holds a key, we should just scroll or move or whatever as long as it's down, and stop as soon as the key is let go. How much we actually move or scroll or whatever will depend on the machine's speed, but that's fine: the important thing in this case is not to overshoot... I don;t know how mouse events fit in exactly, but I suspect it's similar: clicking multiple times on the scrollbar *should* "overshoot" --- i.e., every click should count. Holding the mouse button there, however, should not overshoot: as soon as the user stops pressing, the scrolling should stop.
Again, I don't really know what I'm talking about, and I haven't been following the details of this issue very closely, but just in case this can be of some help...
