hi, perhaps some input from the MS Windows event API: They differentiate between key press and release events (WM_KEYDOWN/WM_KEYUP) using "virtual key codes" and "typed character" events (WM_CHAR) which is the UTF-16 code of a typo generated because of some key press.
The virtual key codes include symbols for SHIFT, CTRL and any other button present on the keyboard. For the WM_CHAR events, there is no notion of press/release. regards Ralph Melchior FRANZ schrieb: > Hi, > > * Robert Osfield -- Thursday 04 June 2009: >> I don't think calling it bug clarifies anything, [...] > > I didn't call it a "bug" to clarify anything, but because the current > behaviour is without any doubt broken. Maybe I should have used the > MS euphemism "issue" instead. :-) > > Your examples use the comparatively trivial 'a' and 'A' case. Here clients > can make the assumption that both are on the same physcial key (although > even that is problematic). So, for an aaaAAA sequence you can assume > that with the first 'a' physical key 38 was pressed, and with the third > 'A' the same key was released. > > But take '3' and '#' instead, and a sequence 333###. Here the client > has no way to realize that this was one and the same physical key 12. > It can *not* assume that the '#' is the successor of the '3' on the > same key. Because the two symbols are only on one key on US-keyboards. > On a German keyboard it's '3' and '§'. And clients know nothing about > the layout. > > So there's not a question *if* release events have to be made up, but > *where*. This can be done in OSG, or in client software (given the > necessary, but currently missing information: the keycode). But it looks > like we already agree on that one. :-) > > > >> One also will need to decide if press 'a' and hold down then press >> shift, then press release shift to do aAa should you get the events: > > True. The problem is that OSG events don't actually represent physical > keys, but symbols. (The keycode information is thrown away, after all). So > IMHO we have to think in terms of keySym sequences. And for me aaAAAaa is > a-press ... a-release, A-press...A-release, a-press...a-release. The > shift modifier changes symbol 'a' to 'A', and it's IMHO hard to argue > that 'a' is still considered pressed during all of aaaAAA. That would > mean that for the client there's 'a' and 'A' pressed at the same time > for a while. That's especially bad if both keys are meant to have > antagonistic effects, which is rather common, I guess. > > > >> The proper solution probably lies in have raw keyboard events, and the >> modified virtual events handled in some coherent fashion exactly what >> I can't say yet. > > Agreed. Maybe it's also time for others to add their wisdom, now that > we are already a bit tired and annoyed. (Aren't we? ;-) > > m. > _______________________________________________ > osg-users mailing list > osg-users@lists.openscenegraph.org > http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org _______________________________________________ osg-users mailing list osg-users@lists.openscenegraph.org http://lists.openscenegraph.org/listinfo.cgi/osg-users-openscenegraph.org
