[xmonad] Getting layout combinators the right way around

[Toby Cubitt]

On Mon, May 03, 2010 at 05:41:51PM -0300, Henrique G. Abreu wrote:
> >
> > I've been trying to produce a layout that works something like
> > TwoPane/DragPane (I don't really understand the difference between these
> > two),
> 
> IIRC on DragPane you can use the mouse to move/resize the layout division.

Ah, I so rarely use the the mouse in xmonad that I'd missed that :) Maybe
it would be worth adding this to the Haddock description of DragPane,
since at present its description is essentially identical to that of
TwoPane.

> > but with tabs for the slave windows. I.e. if there's only one
> > window, it takes up the full screen; if there are two windows, the master
> > window is always in the left pane, the slave on the right; if there are
> > more than two windows, the master is on the left, one slave is displayed
> > on the right, and there are tabs (as in e.g. simpleTabbed) showing the
> > slave windows.
> >
> You've probably missed X.L.Master at contrib, which does just that.
> layout = mastered 0.03 0.5 simpleTabbed

You're quite right, and it does indeed do exactly what I want!

 
> > I've achieved something very close to what I want using:
> >
> >  reflectHoriz $ combineTwo (TwoPane 0.03 0.5) simpleTabbed Full
> >
> > I needed the reflectHoriz because with
> >
> >  combineTwo (TwoPane 0.03 0.5) Full simpleTabbed
> >
> > hidden slave windows ended up in the left pane, rather than appearing as
> > tabs in the right pane. But the (minor) annoyance with using reflectHoriz
> > is that the focusDown/focusUp and Shrink/Expand key bindings go the
> > opposite way to every other workspace (I use PerWorkspace layouts, so
> > simply switching the bindings won't work).
> 
> when using combineTwo that's what you get, the new windows go to the first
> layout by default.
> you could set a manageHook that tests if it's in the right workspace and
> move the window to the second layout.
> It's not very difficult,

...if you do your homework and learn some proper Haskell, which I haven't
had time / have been too lazy to so far...

> but since X.L.Master does exactly what you understood you want...

The X.L.Master solution is perfect. Once again, I'm astounded by the
speed at which posts to this list get a helpful and friendly answer in
return! Thanks very much,

Toby
-- 
Dr T. S. Cubitt
Quantum Information Theory group
Department of Mathematics
University of Bristol
United Kingdom

email: tsc25 at cantab.net
web: www.dr-qubit.org