The memory referred to by the reply argument of
property_update_wm_protocols is automatically free'd
by xcb later on, so it is not safe to simply use the
value of reply in our own data structures. If we did
this, future calls to xcb_get_wm_protocols_reply_wipe
free the data which has already been free'd by xcb,
causing a double-free and corrupting the heap. In
addition, it isn't safe to use free'd memory as if
it is still allocated. Instead, duplicate the data
referred to by reply and use the duplicate instead.
It seems to me as if the duplication should actually
be done in xcb_get_wm_protocols_from_reply, but I'm
not really sure. If that is the case, this is simply
a work-around until xcb can be fixed.
Signed-off-by: Ari Entlich <atrigent@ccs.neu.edu>
Signed-off-by: Julien Danjou <julien@danjou.info>
Return the label of the selected menu item as an argument for the
callback function
Signed-off-by: Pierre Mazière <pierre.maziere@gmail.com>
Signed-off-by: Julien Danjou <julien@danjou.info>
Changing the width or height of a (right or bottom positioned)wibox,
after it was initialized, to a value different than used when it was
created would leave the wibox in a wrong position. Position was off by
as many pixels as the difference between the old and the new
value. Now every wibox is repositioned to reflect this new setup.
Signed-off-by: Adrian C. (anrxc) <anrxc@sysphere.org>
Signed-off-by: Julien Danjou <julien@danjou.info>
If c->honor_size_hints is set to false and the client indicates through
its hints that it wants to be fixed size (by setting min and max width
equal), size hints are not ignored due to client_isfixed not taking
honor_size_hints being false into account. This commit fixes that.
Signed-off-by: Gregor Best <gbe@ring0.de>
Signed-off-by: Julien Danjou <julien@danjou.info>
The awful.widget.graph allows to change the maximum value a graph can
handle, thus allows users to use widgets and scripts that don't scale
the values down to 0 - 1 range. The progressbars did not allow this
and worked with a hard-coded value of 1.
Signed-off-by: Adrian C. (anrxc) <anrxc@sysphere.org>
Signed-off-by: Julien Danjou <julien@danjou.info>
fvwm-root is a part of the FVWM window manager, and is one of those
utilities that properly support transparency because it can retain and
publish the Pixmap with which the background has been set. It supports
XBM, XPM, PNG and SVG image formats, but is not very rich in options
as opposed to habak or feh.
Signed-off-by: Adrian C. (anrxc) <anrxc@sysphere.org>
Signed-off-by: Julien Danjou <julien@danjou.info>
Commit 87070ca introduced support for access keys in menus. Chosen
character was "&", an XML entity that has potential to cause problems;
if a user tries to add multiple access keys or just put an ampersand
in the name of the item the menu will break. Breaking happened because
ampersands were not escaped - first one was substituted and the rest
were ignored, but not escaped which triggered a pango markup error.
Signed-off-by: Adrian C. (anrxc) <anrxc@sysphere.org>
Signed-off-by: Julien Danjou <julien@danjou.info>
This fixes problem with Qt 4.6.
Our implementation cannot be standard, and was not standard, so just
stop using this EWMH crap.
Signed-off-by: Julien Danjou <julien@danjou.info>
This commit adds naughty.suspend() and naughy.resume() which allow
suspension of notifications, which is useful e.g. when watching a movie,
where notifications popping up all the time would be disturbing.
While suspended, notifications are collected and displayed after
naughty.resume() is called.
Signed-off-by: Gregor Best <gbe@ring0.de>
Signed-off-by: Julien Danjou <julien@danjou.info>
property_update_wm_hints() pushes the client but never pops it. This is a really
bad thing (tm).
Signed-off-by: Uli Schlachter <psychon@znc.in>
Signed-off-by: Julien Danjou <julien@danjou.info>
There is a push 15 lines before the push that is removed here, but the client
that is pushed there was never popped. Bad.
Fix this by letting luaA_class_emit_signal() do the needed pop for us.
Signed-off-by: Uli Schlachter <psychon@znc.in>
Signed-off-by: Julien Danjou <julien@danjou.info>
Commit a757dda replaced the awful.client.movetotag call with c:tags()
in the tag property handler. Which introduced a regression when more
than one screen is used. The awful.tag.withcurrent function was being
called even when there was a valid tag property. Now the clients
screen is set to the tag screen to ensure the client is moved to the
correct tag and screen.
Signed-off-by: Adrian C. (anrxc) <anrxc@sysphere.org>
Signed-off-by: Julien Danjou <julien@danjou.info>
Changing screen focus to an empty tag leaves focus on a client on the
previous screen, if a tag with clients is focused next the focus
remains on the previous screen. Now awful.autofocus compares the tag
screen and clients screen to learn if the focus should be given to a
client on the focused screen. Client that gets the focus is selected
from history, as usual.
Signed-off-by: Adrian C. (anrxc) <anrxc@sysphere.org>
Signed-off-by: Julien Danjou <julien@danjou.info>
We unregister the default awful.tag.withcurrent 'manage' signal handler
and have awful.rules.apply handle initial client tagging.
Signed-off-by: koniu <gkusnierz@gmail.com>
Signed-off-by: Julien Danjou <julien@danjou.info>
This explicitly registers withcurrent(c) as a manage signal handler so
that it can be removed if necessary.
Signed-off-by: koniu <gkusnierz@gmail.com>
Signed-off-by: Julien Danjou <julien@danjou.info>
Second try:
Turns out I messed up with XCB_EVENT_SENT() and had a "!" too much. The old code
already tried to cope with this, but forgot to actually unmap the window which
it just set to withdrawn state.
This time I tested the patch *again* and now I found even less bugs than on my
last try.
P.S.: I suck.
Signed-off-by: Uli Schlachter <psychon@znc.in>
Signed-off-by: Julien Danjou <julien@danjou.info>
When both height and width arguments are provided to the function
new() there is a check to see if one of them was a screen percentage,
and is expected to be a string. If the user provided absolute pixels,
and argument is of type number awesome will crash. Now tostring is
used to sanitize the arguments during this check.
Signed-off-by: Adrian C. (anrxc) <anrxc@sysphere.org>
Signed-off-by: Julien Danjou <julien@danjou.info>
awful.wibox.align() in 3.4.x gets the available screen area from
capi.screen[screen].geometry.
this can easily result in overlapping wiboxes since the work area
provided does not account for existing wiboxes.
In my configuration I use a vertical wibox positioned at the
right and it half-overlaps my top wibox because it is streched and
hard-coded to align at the middle.
Signed-off-by: Julien Danjou <julien@danjou.info>
Function wibox_update_strut would not take the border width into
account when calculating struts. When a wibox border was in use
clients would overlap the wibox. With a border of 1px we loose 1px of
the wibox, but as the wibox border increases it is "pushed" by the
border nearest the screen edge and clients steal more and more space.
Signed-off-by: Adrian C. (anrxc) <anrxc@sysphere.org>
Signed-off-by: Julien Danjou <julien@danjou.info>
According to the Inter-Client Communication Conventions Manual, if you want to
switch your window to withdrawn state, you unmap it and send a synthetic
UnmapNotify to the root window.
This synthetic event fixes a race condition. When you map and unmap a window
quickly, the map will generate a MapRequest for the WM but won't actually map
the window. Thus, the unmap will be discarded (-> window not yet mapped) and the
window stays map once the WM handles the MapRequest
Before this patch, awesome just ignored the synthetic unmap notify which caused
the bug to appear again. With this patch it doesn't happen anymore.
Signed-off-by: Uli Schlachter <psychon@znc.in>
Signed-off-by: Julien Danjou <julien@danjou.info>
This call is required by the freedesktop specification.
Signed-off-by: Uli Schlachter <psychon@znc.in>
Signed-off-by: Julien Danjou <julien@danjou.info>
In older versions of awesome, the first entry in the layouts table was the
default layout. Currently it's always the floating layout.
With this patch the default is again the first entry in the layouts table. At
the same time, floating is moved to be the first entry, so this patch doesn't
change any behavior at all.
Signed-off-by: Uli Schlachter <psychon@znc.in>
Signed-off-by: Julien Danjou <julien@danjou.info>
This makes imlib add images loaded from files to its image cache again. That way
we avoid loading the same image again and again (e.g. if you cycle through the
list layouts all the layout icons are loaded).
To not reintroduce FS#651 (fixed in 2dadce9b), we call
imlib_image_set_changes_on_disk(). This causes two stat() calls if the same file
is loaded again to check if the file's timestamps changed which is still less
expensive than loading the whole image again.
Signed-off-by: Uli Schlachter <psychon@znc.in>
Signed-off-by: Julien Danjou <julien@danjou.info>
Commit 6c5d0ca9 introduced a side effect. If tooltip text isn't changed
place() function is not called. To prevent multiple place() set_geometry
calls that function when tooltip is not visible (before it appears to the
user).
Signed-off-by: Sébastien Gross <seb•ɱɩɲʋʃ•awesome•ɑƬ•chezwam•ɖɵʈ•org>
Signed-off-by: Julien Danjou <julien@danjou.info>
Functions show() and toggle() accept a boolean parameter controlling
keyboard navigation. In case of the clients menu neither is called
directly, instead the clients menu is always re-created in case any
new clients appeard since the last invocation. This patch adds a
keygrabber parameter to the clients() function as well.
Signed-off-by: Adrian C. (anrxc) <anrxc@sysphere.org>
Signed-off-by: Julien Danjou <julien@danjou.info>
This changes wibox_t::mouse_over to a proper reference. That way one can't
remove that widget from underneath us which would lead to an unprotected lua error.
Signed-off-by: Uli Schlachter <psychon@znc.in>
Signed-off-by: Julien Danjou <julien@danjou.info>
You can define a menuitem like before:
awful.menu({ items = { { "&awesome", myawesomemenu,
beautiful.awesome_icon },
{ "open &terminal", terminal },
})
and the letters following the ampersand in itemname can be used as
access key. (at least former fvwm users might find this familiar)
When there are submenus open awesome ascends during the search from the
current menu to the top level menu.
The access key is shown underlined.
There is no warning for multiple used keys, only the first one found
gets the price.
Signed-off-by: Julien Danjou <julien@danjou.info>
This reverts commit 7575f64993.
As seen on ML:
Switching tags with multiple heads causes focus to change
screens with this commit
Signed-off-by: perry <perry.hargrave@gmail.com>
Signed-off-by: Julien Danjou <julien@danjou.info>
We check the wibox border width for inappropiate values now, if we set
border_width to a negative value (invalid) in our theme and draw it, the
actual border_width will be 0 so the menu will be drawn incorrectly.
This fixed this.
Signed-off-by: Hiltjo Posthuma <hiltjo@codemadness.org>
Signed-off-by: Julien Danjou <julien@danjou.info>
There might be multiple screens with visible windows. When closing the
last window of the current screen, try to focus them as well.
If no window has focus and yet some windows are visible (e.g. on another
screen), awesome misbehaves: it uses sloppy keyboard focus (as is the X11
default), but doesn't refresh their border colors or tasklists etc.
Signed-off-by: Tomi Belan <tomi.belan@gmail.com>
Signed-off-by: Julien Danjou <julien@danjou.info>
Ari Entlich
Pierre Mazière
Maciej Blizinski
Adrian C. (anrxc)
Gregor Best
Julien Danjou
Andrei 'Garoth' Thorp
Uli Schlachter
Thomas Arendsen Hein
koniu
Alex Alexander
Gwenhael Le Moine
Pierre Allegraud
Sébastien Gross
Robert Lehmann
Matthias Schröder
perry
Hiltjo Posthuma
Levin Du
Tomi Belan