Thank you very much, Jeremy. Ok, maybe I considered the ideal definition too much. (I wrongly thought Janek's case was related to this problem, but no, so forget it.)
So precise documentation on the behavior of current-head will suffice for now, I think. Could you please do it, Jeremy? The new func 'window-head' is not to be done now, since it will more or less involve the precise head location stated in my previous message, but we're not yet sure what is the good solution. (The straightforward one is easy, but then you have to fix all calls to current-head.) We don't hear of any realistic problems related to it, so it won't hurt to put it off until we understand RandR more. If necessary, you can put the link to this thread in the source. Regards, Teika (Teika kazura)
