On Tue, Jan 27, 2009 at 12:54:00PM +0100, Karl Hasselström wrote:

On 2009-01-26 11:32:25 -0500, Brent Yorgey wrote:

...

Hmm, that's a tricky one. It should be possible in theory, but you would need to implement something pretty low-level to make it work, maybe even a brand new layout modifier based on the WindowNavigation one. This could make a nice project for someone.

The problem is that there's no way (currently) for the WindowNavigation layout modifier to report what it did (whether it actually moved the focus to a different window or not), so no way to take another action conditioned on that result.

Yeah, I thought that might be the case. A bit over my head at this point, though -- I still haven't understood how the typing works with sendMessage, for example. :-/ And I guess what has to be done is provide a way for sendMessage to return stuff? Hmm, I guess that could be faked with side effects in the X monad, but that doesn't seem pretty.

Actually, I was wrong! Check out the new XMonad.Actions.MessageFeedback, which was just applied in darcs. It lets you send a message and find out whether the message was handled---which, if WindowNavigation works at all in a sane manner, should tell you whether the movement command actually had any effect. You should be able to use this to pretty easily construct an action that first sends a motion command for WindowNavigation, and then conditionally does a nextScreen action or whatever, based on whether WindowNavigation responded to the message. -Brent