- xmonad - Haskell.org

xmonad 0.2 release candidate
by Spencer Janssen 29 May '07

29 May '07

The xmonad team is pleased to announce release candidates for xmonad and X11-extras 0.2: http://xmonad.org/xmonad-0.1.20070528.tar.gz http://xmonad.org/X11-extras-0.1.20070528.tar.gz Over the past month xmonad has gained several features, and therefore, several opportunities for new bugs. We're especially interested in stress testing for: - correct numlock handling (keybindings should work whether numlock is on or off) - randr support - statusbar toggling - the new restart mechanism (remembers window positions) - Xinerama in general If all goes smoothly, these packages will become xmonad and X11-extras 0.2 in a few days time. Cheers, Spencer Janssen

1 1

Re: [Xmonad] patch: wmdock script - manage dockable windows using dmenu
by dons＠cse.unsw.edu.au 29 May '07

29 May '07

k.kosciuszkiewicz: > Donald Bruce Stewart wrote: > > >Got a screenshot? > > Sure. > > > [...] > >Yes, you'd need the trayer to have the override bit set. > > You're probably right, but I've solved it my way - I just mapped a key > to "withFocused unmanage" in xmonad and voila :) On the topic of setting the override-redirect flag, Stefan noted that a few X apps already let us set that via X resources. I added this to the FAQ on the webpage: xmonad let's you use any application as a 'statusbar', as long as it is visible in a given 'gap' on the screen, and has the override-redirect property set to true. Many status bar/dock programs already set this property, for example, dzen or kicker. To set other applications, you can sometimes use normal X resources For example, to use xclock, launch it with xclock -digital -xrm '*overrideRedirect: True' -geometry 1024x30+0+0 And set a gap of (30,0,0,0) in Config.hs. Spencer also uploaded a screenshot of using the KDE kicker as a statusbar (!) http://xmonad.org/images/screen-sjanssen-kde-kicker.png Any other thoughts on useful status bar programs, anyone? -- Don

1 0

patch: wmdock script - manage dockable windows using dmenu
by Krzysztof Kościuszkiewicz 28 May '07

28 May '07

Hello, I'm really enyoing the Xmonad experience so far, but inability to use applications that insist on docking was a drawback to me. So I did my firsts steps in the Xlib programming and here are the results. The hack is really crude; script identifies windows that look like they should be docked and optionally sends mouse events to them (most of such apps will raise the main window after such treatment). I hope someone will find it useful. Cheers, -- Krzysztof Kościuszkiewicz, Gadu: 111851, Jabber: kokr(a)jabberpl.org "Simplicity is the ultimate sophistication" -- Leonardo da Vinci New patches: [Added wmdock application Krzysztof Kosciuszkiewicz <k.kosciuszkiewicz(a)gmail.com>**20070529011349] { adddir ./examples/dmenu-wmdock addfile ./examples/dmenu-wmdock/README hunk ./examples/dmenu-wmdock/README 1 +Use dmenu to "manage" windows that belong to WM-style docks. + +wmdock + A Python script using Xlib bindings to query and send mouse button events + to windows that seem to be dockable. Probably should be rewritten in + haskell ;) + +Config.hs + Add something similar to setup keybindings for wmdock and dmenu: + + , ((modMask, xK_t ), spawn "wmdock -n | dmenu | wmdock send") + + mod-t pops up a menu with all dockable windows. Selecting one will send a + double-button1 click to it (hopefully raising/lowering main app window). + Modify wmdock to suit your taste. + +dmenu is available from + http://www.suckless.org/wiki/tools/xlib + +python Xlib bindings are available from + http://python-xlib.sf.net/ addfile ./examples/dmenu-wmdock/wmdock hunk ./examples/dmenu-wmdock/wmdock 1 +#! /usr/bin/env python + +__version__ = "0.2" +__author__ = "Krzysztof Kosciuszkiewicz <k.kosciuszkiewicz(a)gmail.com>" +__program__ = "wmdock" +__license__ = "BSD" + +import Xlib +import Xlib.X +import Xlib.Xatom +import Xlib.Xutil +import Xlib.display +import Xlib.protocol +import sys +import time + +# globals +disp = Xlib.display.Display() +scrn = disp.screen() +root = scrn.root + +# the atoms that we will need to query for +atom_wm_type = disp.intern_atom("_NET_WM_WINDOW_TYPE") +atom_wm_type_dock = disp.intern_atom("_NET_WM_WINDOW_TYPE_DOCK") +atom_wm_kde_dock = disp.intern_atom("_KDE_NET_WM_SYSTEM_TRAY_WINDOW_FOR") + +def get_property(win, prop, prop_type=Xlib.X.AnyPropertyType): + return win.get_full_property(prop, prop_type) + +def is_dock(win): + """Check if the X window is a dock app""" + # everything in this function is based on code from Ion3 WM + + # test window manager hints + hints = win.get_wm_hints() + if hints: + init_state = Xlib.Xutil.NormalState + if hints.flags & Xlib.Xutil.StateHint: + init_state = hints.initial_state + if init_state == Xlib.Xutil.WithdrawnState and \ + hints.flags & Xlib.Xutil.IconWindowHint and hints.icon_window: + return True + + # check if window has WM_WINDOW_TYPE_DOCK as WM_WINDOW_TYPE property + prop = get_property(win, atom_wm_type) + if prop: + # type should be atom, and value should contain the atom type_dock + if prop.format == Xlib.Xatom.ATOM and \ + atom_wm_type_dock in prop.value: + return True + + # test window manager class + # also all Qt dockable windows I've seen satisfy cls[0].endswith('dock') + cls = win.get_wm_class() + if cls and cls[1] == "DockApp": + # or cls[0].endswith('dock'): + return True + + # test if it's kde-style dockapp + if get_property(win, atom_wm_kde_dock): + return True + + return False + + +def send_mouse_click(window, button=Xlib.X.Button1, double=False, delay=500): + """Send button press event followed by button release event to client""" + # I hope that translation of coordinates is ok here. + coords = root.translate_coords(window, 4, 4) + ev1_dict = dict( + time=0, + root=root, + window=window, + same_screen=1, + child=Xlib.X.NONE, + root_x=coords.x, + root_y=coords.y, + event_x=4, + event_y=4, + state=Xlib.X.NONE, + detail=button) + ev2_dict = ev1_dict.copy() + ev2_dict['state'] = button << 8 # that is ButtonXMask + ev1 = Xlib.protocol.event.ButtonPress(**ev1_dict) + ev2 = Xlib.protocol.event.ButtonRelease(**ev2_dict) + window.send_event(ev1) + disp.sync() + time.sleep(delay/1000) + window.send_event(ev2) + disp.sync() + if double: + time.sleep(delay/1000) + window.send_event(ev1) + disp.sync() + time.sleep(delay/1000) + window.send_event(ev2) + disp.sync() + + +def walk_tree(win=root, func=None): + """Walk the X window hierarchy and call func on each window""" + if callable(func) and win: + func(win) + if win: + res = win.query_tree() + if res: + for chld in res.children: + walk_tree(chld, func=func) + return func + + +class WindowVisitor(object): + """Visits windows and collects ones that match predicate""" + def __init__(self, condition=None): + self.winlist = [] + self.condition = condition + + def __call__(self, window): + if callable(self.condition) and self.condition(window): + self.winlist.append(window) + + def collected(self): + return self.winlist + + +if __name__ == "__main__": + import getopt + import string + import os + + usage = """Usage: %s [options...] [list|send {ids|names...}] + """ % __program__ + helpmsg = """Options: + -n list only window names (default: False) + -i list only window IDs (default: False) + -b button mouse button number in event to be sent (default: 1) + -d delay delay between mouse press and release events (default: 500ms) + -s send single-click events (default: False) + """ + behaviour = """Operation: + If you tell me to "list", I will find and print IDs and names of dockable + X11 windows on your display. If you tell me to "send", I will send button + click events to each of specified windows (use names or IDs). If you + don't specify any after "send" I will try to read them from standard + input. + + You can combine "list" and "send". By default I just "list". + """ + try: + opts, args = getopt.gnu_getopt(sys.argv[1:], + "b:d:sinh", ["help", "version"]) + except getopt.GetoptError, e: + print >>sys.stderr, e + print >>sys.stderr, usage + sys.exit(0) + + button = Xlib.X.Button1 + delay = 500 + double = True + window_repr = lambda win: "%-10s %s" % (win.get_wm_name(), hex(int(win.id))) + for o, val in opts: + if o == '-b': + button = Xlib.X.Button1 + int(val) - 1 + elif o == '-d': + delay = int(val) + elif o == '-s': + double = False + elif o == '-i': + window_repr = lambda win: hex(int(win.id)) + elif o == '-n': + window_repr = lambda win: win.get_wm_name() + elif o == '--help': + print usage + print helpmsg + print behaviour + sys.exit(0) + elif o == '--version': + print "%s %s" % (__program__, __version__) + sys.exit(0) + + do_print = True + do_send = False + get_ids = lambda: [] + + if not args or args[0] == "list": + args = args[1:] + else: + do_print = False + + if args and args[0] == "send": + args = args[1:] + do_send = True + if args: + get_ids, args = lambda: args, [] + else: + get_ids = lambda: map(string.strip, sys.stdin.read().split()) + + if args: + print >>sys.stderr, usage + sys.exit(0) + + windows = walk_tree(func=WindowVisitor(is_dock)).collected() + + if do_print: + for win in windows: + print window_repr(win) + sys.stdout.flush() + + if do_send: + for word in get_ids(): + try: + winid = int(word, 16) + finder = lambda w: w.id == winid + except: + finder = lambda w: w.get_wm_name() == word + + for win in filter(finder, windows): + send_mouse_click(win, button=button, delay=delay, double=double) + + sys.exit(0) } Context: [Rescreen is in main xmonad now Spencer Janssen <sjanssen(a)cse.unl.edu>**20070528050656] [replace "name" in NamedWindow with a Show instance. David Roundy <droundy(a)darcs.net>**20070526185114] [[Spiral] blend in the scale factor so it doesn't have any effect on the smallest windows joe.thornber(a)gmail.com**20070525032732] [[Spiral] last rect takes all available space joe.thornber(a)gmail.com**20070524120239] [[Spiral] Introduce a simpler Rect data type to remove a lot of the fromIntegrals joe.thornber(a)gmail.com**20070524100423] [[Spiral] divideRects now takes a list of directions to split in joe.thornber(a)gmail.com**20070524090211] [[Spiral] misc tidying joe.thornber(a)gmail.com**20070524085537] [[Spiral] remove old spiral code joe.thornber(a)gmail.com**20070524084805] [[Spiral] add fibonacci spiral joe.thornber(a)gmail.com**20070524084423] [Allow clients of NamedWindows to get at the name. glasser(a)mit.edu**20070523184251] [dzen module (with xinerama support, which requires glasser's Xinerama patch to dzen) glasser(a)mit.edu**20070523184315] [Extract NamedWindow support from Mosaic into its own module glasser(a)mit.edu**20070523155855] [remove SwapFocus (which is no longer possible) David Roundy <droundy(a)darcs.net>**20070523153841 This module depended on the focus stack. ] [Fix Spiral's module name Spencer Janssen <sjanssen(a)cse.unl.edu>**20070522170909] [[SPIRAL] add spiral tiling layout joe.thornber(a)gmail.com**20070522062537] [Make RotView compile. Miikka Koskinen <arcatan(a)kapsi.fi>**20070522075338 As I'm not a Xinerama user, I'm not sure if rotView should consider only hidden workspaces or also visible but not focused workspaces. I thought hidden workspaces only would be more logical. ] [bug fix in DwmPromote. whoops. Miikka Koskinen <arcatan(a)kapsi.fi>**20070522062118] [make FindEmptyWorkspace compile Miikka Koskinen <arcatan(a)kapsi.fi>**20070521123239] [make DwmPromote compile Miikka Koskinen <arcatan(a)kapsi.fi>**20070521123140] [updated Dmenu.hs to work with zipper StackSet Jason Creighton <jcreigh(a)gmail.com>**20070521233947] [Add GreedyView Spencer Janssen <sjanssen(a)cse.unl.edu>**20070521220048] [Rescreen: collects new screen information Spencer Janssen <sjanssen(a)cse.unl.edu>**20070521164808] [Fixes for windowset -> workspace rename Spencer Janssen <sjanssen(a)cse.unl.edu>**20070521042118] [TwoPane: hide windows that aren't in view Spencer Janssen <sjanssen(a)cse.unl.edu>**20070518224240] [make Mosaic even less picky by default. David Roundy <droundy(a)darcs.net>**20070516175554] [add clear window message in Mosaic. David Roundy <droundy(a)darcs.net>**20070516175518] [Comment only Spencer Janssen <sjanssen(a)cse.unl.edu>**20070517211003] [Add instructions for TwoPane Spencer Janssen <sjanssen(a)cse.unl.edu>**20070517210206] [Add TwoPane Spencer Janssen <sjanssen(a)cse.unl.edu>**20070517195618] [throttle the exponential expense when many windows are present. David Roundy <droundy(a)darcs.net>**20070516022123] [make mosaic configure windows by name rather than by Window. David Roundy <droundy(a)darcs.net>**20070512215644 Note that this is still pretty flawed. Often window names change, and the layout then stagnates a bit. Gimp, for example, opens most its windows with the same name before renaming them, so you have to hit mod-return or something to force a doLayout. Also, gimp still overrides xmonad regarding the size of its main window. :( ] [XMonadContrib.FindEmptyWorkspace Miikka Koskinen <arcatan(a)kapsi.fi>**20070513184338 With this module you can find empty workspaces, view them and tag windows to them. ] [make DwmPromote compile Miikka Koskinen <arcatan(a)kapsi.fi>**20070513184254] [make DwmPromote compile again Miikka Koskinen <arcatan(a)kapsi.fi>**20070510154158] [make DwmPromote compile Miikka Koskinen <arcatan(a)kapsi.fi>**20070503105236] [add SwapFocus. David Roundy <droundy(a)darcs.net>**20070512191315] [make rotView only consider non-visible workspaces (Xinerama) Jason Creighton <jcreigh(a)gmail.com>**20070510012059] [fix commend in RotView. David Roundy <droundy(a)darcs.net>**20070505185654] [switch to Message type for layout messages Don Stewart <dons(a)cse.unsw.edu.au>**20070505014332] [Fix instructions in Mosaic. Chris Mears <chris(a)cmears.id.au>**20070503222345] [add Mosaic layout. David Roundy <droundy(a)darcs.net>**20070503151024] [-Wall police Spencer Janssen <sjanssen(a)cse.unl.edu>**20070503211700] [Make RotView build, and add a brief description. Chris Mears <chris(a)cmears.id.au>**20070503104234] [comment: Gave URL to xinerama-enabled dmenu patch Jason Creighton <jcreigh(a)gmail.com>**20070503053133] [Put dmenu in X too Spencer Janssen <sjanssen(a)cse.unl.edu>**20070503053727] [Add dmenu (thanks jcreigh) Spencer Janssen <sjanssen(a)cse.unl.edu>**20070503052225] [add RotView module. David Roundy <droundy(a)darcs.net>**20070421233838] [XMonadContrib.DwmPromote: dwm-like promote Miikka Koskinen <arcatan(a)kapsi.fi>**20070501082031 I like the way dwm's equivalent to xmonad's promote works, so I implemented dwmpromote. ] [add simple date example Don Stewart <dons(a)cse.unsw.edu.au>**20070429064013] [more details Don Stewart <dons(a)cse.unsw.edu.au>**20070429061426] [add readme Don Stewart <dons(a)cse.unsw.edu.au>**20070429061329] [Initial import of xmonad contributions Don Stewart <dons(a)cse.unsw.edu.au>**20070429061150] Patch bundle hash: 8c693213500612db37f8411c2498d77d9268dcd9

2 3

Re: [Xmonad] Config.hs: version Config.hs.template instead?
by dons＠cse.unsw.edu.au 28 May '07

28 May '07

wagnerd: > Donald Bruce Stewart wrote: > >glasser: > >>Can we delete Config.hs from the repository and version a > >>Config.hs.template instead? (Part of the build process could be to > > > >Integrating this with Cabal would be the main problem. > > Could the integration be as simple as changing Setup.lhs to read > something like the one below? > ~d > > #!/usr/bin/env runhaskell > > import System.Directory > > import Control.Monad > > import Distribution.Simple > > makeCopy = copyFile "Config.hs.template" "Config.hs" > > main = doesFileExist "Config.hs" >>= flip when makeCopy >> defaultMain Oh, this is a nice idea (I think). We'll look at this after the release. -- Don

1 0

darcs patch: Give link to bugtracker in "BUGS" section of manpage
by Jason Creighton 28 May '07

28 May '07

Mon May 28 19:58:51 MDT 2007 Jason Creighton <jcreigh(a)gmail.com> * Give link to bugtracker in "BUGS" section of manpage

2 1

Issue 1 in project xmonad
by codesite-noreply＠google.com 28 May '07

28 May '07

Issue 1: Stable numlock handling. http://code.google.com/p/xmonad/issues/detail?id=1 Comment #3 by SpencerJanssen: The current handling of numlock seems sufficient. Issue attribute updates: Status: Fixed -- You received this message because you are listed in the owner or CC fields of this issue, or because you starred this issue. You may adjust your issue notification preferences at: http://code.google.com/hosting/settings

1 0

Issue 5 in project xmonad
by codesite-noreply＠google.com 28 May '07

28 May '07

Issue 5: Not able to cycle through windows when in fullscreen layout mode http://code.google.com/p/xmonad/issues/detail?id=5 Comment #4 by SpencerJanssen: It's been nearly a month without additional information on this bug. Closing for now. Issue attribute updates: Status: Invalid -- You received this message because you are listed in the owner or CC fields of this issue, or because you starred this issue. You may adjust your issue notification preferences at: http://code.google.com/hosting/settings

1 0

Issue 9 in project xmonad
by codesite-noreply＠google.com 28 May '07

28 May '07

Issue 9: 'amarok' (KDE music program) causes reproducible failures from xmonad. http://code.google.com/p/xmonad/issues/detail?id=9 Comment #3 by SpencerJanssen: We can't reproduce this issue, and the initial reporter has not given sufficient detail. Closing for now. Issue attribute updates: Status: Invalid -- You received this message because you are listed in the owner or CC fields of this issue, or because you starred this issue. You may adjust your issue notification preferences at: http://code.google.com/hosting/settings

1 0

Config.hs: version Config.hs.template instead?
by David Glasser 28 May '07

28 May '07

Can we delete Config.hs from the repository and version a Config.hs.template instead? (Part of the build process could be to copy the template onto the real file if it doesn't exist yet.) That way it's easy to hack on xmonad and customize your own experience with the same checkout. (And you can even version Config.hs in your own repository and just never push any patches to Config.hs, if I understand darcs correctly.) --dave -- David Glasser | glasser(a)mit.edu | http://www.davidglasser.net/

2 1

Darcs best practice: Resolving conflicts
by Jason Creighton 28 May '07

28 May '07

Hi guys, Dave Glasser and I recently both ran into an issue with darcs trying to merge patches with changes made locally. (Mine were unrecord; I think Dave's were recorded but unpushed patches.) Appearently, this may be due in part to conflicts in the main repo. It turns out that if you resolve conflicts the wrong way, you can end up with "merger" patches in the main repo that can cause problems later. I don't really understand that bit at all. But anyway, I talked to Heffalump on #darcs about this, and what I took away from the conversation was this: Whenever you have conflicts between two recorded patches, you should use amend-record to fix the conflict instead of making a separate "fix conflicts" patch. You can't use amend-record on any patch that has been released into the wild: only patches that have never left your local repo. If the conflicts are between a patch and unrecorded local changes, my understanding is that it's fine to just fix the conflicts in the files and then record. Disclaimer from Heffalump: 11:43 <jcreigh> Heffalump++ Thanks for all your help. 11:44 <jcreigh> I'm going to kick an email out to the xmonad ML warning people about the proper way to resolve conflicts, because at least myself was unaware, so possibly others. 11:44 <Heffalump> I don't think it's widely known. 11:45 <Heffalump> you should probably qualify it as not being advice from someone who is core to darcs development Disclaimer from me: The above is my understanding of the situation. I could be wrong. Hopefully David Roundy will be kind enough to chime in if that is the case. Jason Creighton

1 0