On Sun, Jun 06, 2004 at 10:00:51AM -0400, Andrew Pimlott wrote: > On Sun, Jun 06, 2004 at 09:09:07AM -0400, Thomas Dickey wrote: > > That's interesting. Comparing with rxvt, I see that rxvt does give > > ButtonPress events as well as the KeyRelease events that xterm gives. > > Perhaps the answer is that xev can only see the events that the application > > is listening to. > > It must be something along these lines (note that xterm also does not > get motion events, which makes more sense), but in what sense is xterm > not listening to button presses? Obviously, it reacts to the mouse > buttons, so I presume it asks to receive these events. Or is it > listening in some non-standard way? I'm guessing that something about > the way it sets up its event filter is confusing xev, but I have no idea > how.
Looking at the code, I think the answer is that in initializing the active-icon logic it modifies the flags there - affecting the regular widget as well. (This is from code dating back to before I started working on xterm). The relevant chunk is /* since only one client is permitted to select for Button * events, we have to let the window manager get 'em... */ values->event_mask &= ~(ButtonPressMask | ButtonReleaseMask); values->border_pixel = term->misc.icon_border_pixel; Now that my attention is on it, I suspect that the intent was to only affect the active icon - not the uniconified xterm. But "values" is the data for the latter. -- Thomas E. Dickey http://invisible-island.net ftp://invisible-island.net
pgpEJxJnjsG5s.pgp
Description: PGP signature