Spencer Janssen
On Thu, Aug 28, 2008 at 03:09:05PM -0600, Justin Bogner wrote:
Several people seemed interested in this on IRC, but nobody commented on it on the mailing list. It works for my use cases, but I was hoping others would comment on whether it could be better in any way.
I remember having comments for this patch, but I must have forgotten to send an email. This patch addresses an important problem, but I think it goes about it in the wrong way. There are two sorts of struts programs that we typically use: * override-redirect bars like dzen and xmobar. ICCCM says that we shouldn't manage these windows at all, so lowering them is technically a violation of the ICCCM. Instead, these windows should lower themselves on startup.
The patch as it is didn't really take dzen and xmobar into account (I assumed they were the same as desktop bars)
* desktop bars such as kicker and gnome-panel. We are allowed to manage these, and they even set atoms on their window we can use to stack them. In manageDocks, when a window is detected as a DOCK or DESKTOP window, we should send all DESKTOP windows to the bottom of the stacking order and all DOCK windows to one layer above that.
I'm not convinced one layer above that is the right thing, though the only indication of the behaviour i thought was the standard is in the statement "Possible examples of behavior include keeping dock/panels on top", which is in the EWMH spec for _NET_WM_WINDOW_TYPE. That said, the behaviour I expect is that docks should be at the very top or not visible at all, and never anything in between. Nonetheless, I'm not convinced this patch is perfect, and so prefer it not applied. I would appreciate more discussion on the issue, so that we can write a more satisfactory version.
* "darcs patch: Improvements in documentation (and 3 more)" ** "XMonad.Actions.Plane: removed unneeded hiding" ** "XMonad.Config.Gnome: using XMonad.Actions.Plane" ** "XMonad.Actions.Plane.planeKeys: function to make easier to configure" http://www.haskell.org/pipermail/xmonad/2008-July/006064.html Applicable. No discussion. (Poor Marco! Counting his Xmonad patch, that makes 5 or so patches that just got ignored.)
I think the problem here was probably that there are four patches in that set, and one of them seems dubious (Plane seems like a user preference, not something to default for all Gnome users). The other three patches in that set look okay to me.
Yes, this is why I passed over that patch bundle the first time.
Cheers, Spencer Janssen