If the mouse cursor is hidden (by unclutter, for example), then scrolling
in the window decoration creates an event with a child
(i.e. event->child != XCB_NONE). This causes route_click() to be called
with dest=CLICK_INSIDE, which prevents scrolling through a stacked layout.
To fix this, check if a click is in the window decoration _before_
checking if the event has a child.
Not quite sure why there are so many differences. Perhaps we’ve gotten
out of the habit of running clang-format after every change.
I guess it’d be best to have a travis hook that runs clang-format for us
and reports any problems on pull requests.
input_code is a uint16_t, but xcb_keycode_t is uint8_t, meaning that
only the first byte of input_code is inspected by memmem. On
little-endian platforms, this code would have worked by accident, since
the first byte of input_code represents the 8 least significant bits.
However, on big-endian platforms the first byte is the 8 most
significant bits, which means memmem is scanning bind->translated_to
for the wrong keycode (probably 0).
In order to work correctly on big-endian and little-endian platforms,
simply typecast input_code to an xcb_keycode_t and pass that to memmem.
The observed behaviour associated with this bug is that key bindings
don't work at all. This patch has been tested on an iBook G4 running
OpenBSD -current, and key bindings work properly with this fix applied.
Add support for the _NET_WM_MOVERESIZE client message. This message
enables clients to initiate window moving or resizing. Toolkits like
Gtk3 use this message when the user drags a client-side decorated window
by its title bar. When Gtk detects that the window manager does not
support this client message, it uses a slow fallback implementation.
fixes#1432
Add the `--whole-window` switch for mouse bindings. This switch controls
what part of the container the pointer must be over to trigger a mouse
binding. The default is to only trigger mouse bindings over the
titlebars. With this switch, a mouse binding will be triggered over the
main part of the window as well.
This is a breaking change to the previous behavior, which would trigger
a mouse binding with a modifier over any part of the window.
fixes#1429
Since this commit:
f691a55923
the shell that is used is the system's bourne shell (/bin/sh) and the
env variable SHELL is not considered.
No logic changes.
Send the affected workspace in the "current" property for each workspace
event for any type of workspace event that affects a particular
workspace.
fixes#1411
Rather than just toggling the fullscreen modes, allow to set them
directly with:
fullscreen enable|toggle [global]
fullscreen disable
For compatibility, retain the previous command and its toggling behavior:
fullscreen [global]
fixes#1120
Copy the binding struct before running it and use this copy to emit the
binding event.
This fixes a crash when the command `reload` is used in a binding when
the binding event is emitted.
When dumping a binding, as is done during the binding event, check
symbol for NULL. If it is, dump json null. This prevents a crash when
running a binding that was configured with bindcode.
fixes#1379
The binding event will be triggered when a binding is run as a result of
some a user action. The binding event has the following properties:
change: (str) Currently this will only be "run" but may be expanded in
the future. Included for consistency with other events.
binding: (map) the serialized binding
The "binding" member will have these properties:
input_type: (str) either "keyboard" or "mouse"
input_code: (int) the xcb keycode of the keyboard binding if it was
provided or the mouse button if it is a mouse binding.
symbol: (str) the string representation of the input code
command: (str) the bound command
mods: (list of str) a list of the modifiers that were pressed as string
symbols
fixes#1210
Rectangles passed to function xcb_poly_fill_rectangle are of type
xcb_rectangle_t and defined as:
struct xcb_rectangle_t {
int16_t x;
int16_t y;
uint16_t width;
uint16_t height;
}
The rectangles for the right and lower border had a width and height,
respectively, greater than the actual border width.
Furthermore, offset the bottom border to not overlap with the right one
and, for the top border, use r->width instead of con->rect.width as with
the other borders.
When a named workspace (i.e., a workspace that has a name that does not
begin with text that can be parsed as an integer greater than or equal
to zero) is represented by the ipc as a workspace json object such as
can be queried with `i3-msg -t get_workspaces`, set the num property to
-1 instead of json null.
This is for convenience of ipc consumers using type-constrained
languages such as C which have difficulty cleanly expressing nullable
integers.
fixes#1368
Do not set focused_id to XCB_NONE in route click to force resend focus,
in some cases to an already focused window.
Sending focus again on click is not necessary and may cause problems
with certain wine or mono apps. Resending focus makes combo boxes not
work in Office 2010.
This effectively reverts commit 250577da, so in case this commit causes
any problems with Eclipse to resurface, we’ll revert this commit in
favor of 250577da (the Eclipse-related bug fix).
We're going to call parse_configuration() very early if -C is given on
the command line. Instead of the previous "only_check_config", which has
been a global variable, we now simply pass use_nagbar as false if we're
just validating.
This causes the whole parsing to run without X and of course without
starting nagbar and displaying the errors to standard out/error instead.
The return code of parse_configuration() is now a boolean which
represents whether an error occured during parsing and the programs exit
code is returned accordingly.
Although the config parser still has a lot of side-effects, we now can
parse without the need to have an XCB connection. A nicer implementation
would be to just set the new font and load it just after we're done
parsing, but to ensure we don't break functionality we just load a dummy
FONT_TYPE_NONE if XCB isn't available. The main reason for going this
route is that it's a bit difficult to test fonts in a distribution
agnostic way without bundling fonts with i3 (or Xdummy to be more
exact).
Signed-off-by: aszlig <aszlig@redmoonstudios.org>
In normal border style, two horizontal lines are drawn to outline the
bar decoration. The lower line leaves a gap to the left and right to
align with the border. This gap was hardcoded to 2 pixels. Now it
matches the current border width, if any.
In stacked layout, only the bottommost bar is aligned. In tabbed layout,
no gaps are left in order to close the outline and make the notion of a
tab more clear.
Raise a window when cycling focus between floating windows with `focus
[direction]` command so that newly focused windows are rendered on top
of other windows.
This is done by placing the window last in the floating nodes of the
parent and reordering the stack so the relative order is preserved.
fixes#1322
I think this is not a good feature to have. It shouldn’t be necessary
for a tiling window manager that explicitly discourages the use of
floating windows.
fixes#1324
http://tronche.com/gui/x/icccm/sec-4.html
> The WM_CLASS property (of type STRING without control characters)
> contains two consecutive null-terminated strings. These specify the
> Instance and Class names to be used by both the client and the window
> manager for looking up resources for the application or as identifying
> information.
i3 processes changes to WM_CLASS by updating the cached property and
running assignments. This allows the property to be used in criteria
selection.
fixes#1052
When a window becomes managed, explicitly set the current border width
to the default instead of relying on the default value of -1 to apply
the correct value.
Now that there are two different kinds of default borders, a border
width value of -1 is ambiguous. This can lead to different border widths
effectively being applied when the container changes from tiling to
floating, which is surprising behavior.
This commit extends behavior introduced in this commit to normal
borders:
7afe9cc78b
Explicitly set current border width when BS_PIXEL
fixes#1304
Users can specify a command to run when a button was pressed on i3bar to
override the default behavior. Currently only the mouse wheel buttons
are supported. This is useful for disabling the scroll wheel action or
running scripts that implement custom behavior for these buttons.
Example:
bar {
wheel_up_cmd nop
wheel_down_cmd exec ~/.i3/scripts/custom_wheel_down
}
fixes#1104
Fixes a bug where a normal floating default border is not applied when
the default tiling border is set to a pixel value.
This bug was introduced in this commit:
43b447855d
Consider motif border for floating geometry
Fixes a comment that claimed default floating border could override
motif hints, which was never the case.
fixes#1305
This reverts commit b109b1b20d.
Turns out the change in behavior was unexpected by a number of users, so
let’s revert it and make those users that want this behavior configure
it explicitly.
Sorry for the back-and-forth here.
When a window becomes managed, if the config specifies a pixel border
style, explicitly set the current border width so it does not change
when the container changes from tiling to floating.
fixes#1266
http://tronche.com/gui/x/icccm/sec-4.html#s-4.1.4
> IconicState - The client's top-level window is iconic (whatever that
> means for this window manager). The client can assume that its
> top-level window is not viewable, its icon_window (if any) will be
> viewable and, failing that, its icon_pixmap (if any) or its
> WM_ICON_NAME will be displayed.
For these requests, we just close the window.
fixes#1279
When calculating the geometry of a floating window with motif hints that
specify a border style, take into account that this window will have a
different border style when calculating its position with
floating_enable() when the window becomes managed.
A nice side effect of this is that users can override motif hints with
`new_float` config directives when they are specified other than
`normal`.
fixes#1270
Maintain the _NET_DESKTOP_NAMES property on the root window.
http://standards.freedesktop.org/wm-spec/latest/ar01s03.html#idm140251368131760
> _NET_DESKTOP_NAMES
>
> _NET_DESKTOP_NAMES, UTF8_STRING[]
>
> The names of all virtual desktops. This is a list of NULL-terminated
> strings in UTF-8 encoding [UTF8]. This property MAY be changed by a
> Pager or the Window Manager at any time.
A window may become unmapped on manage when an assignment command unmaps
the window, such as moving it to the scratchpad or killing it.
This can cause i3 focus to be an unmapped window and different than X
focus which can lead to complications
fixes#1283
If a `workspace {N}` or `move to workspace {N}` command is given with N
as a plain number, the workspace of this number is selected for the
context of the command if one exists and there is no workspace with a
name that exactly matches N.
A container selected with criteria should be moved with the `move
[direction]` command, instead of this command always acting on the
focused container.
This removes our last dependency on Xlib! :)
(Okay, an Xlib dependency still comes in through other libraries that we
link against, but it’s not us. Our code is simpler by this change and
uses one less connection to X11.)
A configured mouse binding (for example `bindsym button3 kill`) runs
its command when the mouse button is pressed over parts of a container.
If the binding has no modifer, it will only run when the button is
clicked on the window titlebar.
Otherwise if the binding has a modifier, it will run over the titlebar
or any part of the contained window.
fixes#558
http://tronche.com/gui/x/icccm/sec-4.html#s-4.1.7
> Clients using the Globally Active model can only use a SetInputFocus request
> to acquire the input focus when they do not already have it on receipt of one
> of the following events:
> * ButtonPress
> * ButtonRelease
> * Passive-grabbed KeyPress
> * Passive-grabbed KeyRelease
Since managing a window happens on a MapNotify (which is absent from this
list), the window cannot accept input focus, so we should not try to focus
the window at all.
Fixes an issue with xfce4-notifyd which (correctly) declines focus when
we send WM_TAKE_FOCUS, which puts i3 in a state where i3 focus and X
focus are different when a notification appears.
_NET_NUMBER_OF_DESKTOPS:
> This property SHOULD be set and updated by the Window Manager to
> indicate the number of virtual desktops.
We interpret this property as the number of noninternal workspaces.
This should be the last commit that formats a big bunch of files. From
here on, whenever I merge patches, I’ll run clang-format like described
in the title.
This has multiple effects:
1) The i3 codebase is now consistently formatted. clang-format uncovered
plenty of places where inconsistent code made it into our code base.
2) When writing code, you don’t need to think or worry about our coding
style. Write it in yours, then run clang-format-3.5
3) When submitting patches, we don’t need to argue about coding style.
The basic idea is that we don’t want to care about _how_ we write the
code, but _what_ it does :). The coding style that we use is defined in
the .clang-format config file and is based on the google style, but
adapted in such a way that the number of modifications to the i3 code
base is minimal.
I consider this behavior broken and not respecting the standard, but it
happens in real life, and it’s better for i3 to not busy-loop in such a
situation :).
fixes#1259
When `focus_follows_mouse` option is on, prevent an uneeded render on
pointer enter when the con is already focused.
This pointer enter might be caused by entering a window decoration of an
already-focused container.
This extra render can cause concurrency issues when focus is set
asynchronously with WM_TAKE_FOCUS.
Add run_binding function to bindings.h.
> Runs the given binding and handles parse errors. Returns a
> CommandResult for running the binding's command. Caller should render
> tree if needs_tree_render is true. Free with command_result_free().
_NET_ACTIVE_WINDOW:
> The window ID of the currently active window or None if no window has
> the focus.
This fixes a bug that would not update _NET_ACTIVE_WINDOW when focus
changed to an i3 container without a window such as a branch or
workspace content container.
parse_command returns a struct that contains useful information about
the result of a command as a whole (instead of the intermediate
representation used during parsing).
parse_command now requires the caller to allocate the yajl_gen used for
generating a json reply. This is passed as the second parameter to
parse_command. If NULL is passed, no json reply will be generated.
This patch adds a new configuration option "mouse_warping [output|none]".
When mouse warping is disabled, mouse cursor does not jump to middle of current
screen when changing workspaces between multiple outputs. This introduces a
"special" cursor state, where focus is in one window and cursor on another.
Useful for eg. scrolling a web page with mouse wheel while typing into another
window on keyboard.
Workspace assignments with bare numbers assign all workspaces with that
number to the specified output.
Workspace assignment by number is overridden by workspace assignment by
name.
Change the name of structs CommandResult and ConfigResult to
CommandResultIR and ConfigResultIR to show they are an intermediate
representation used during parsing.
Floating windows already had their own border style, but the width was
the same for all windows.
The configuration directives 'new_window' and 'new_float' can now be
used simultaneously to have different border widths for floating and
tiled windows.
fixes#1244
Implement the configuration option within the bar config directive for
custom workspace numbers with the directive `strip_workspace_numbers
yes`.
This directive strips the workspace name of the number prefix and
delimiter. When the workspace name consists only of the number, it will
default to show the number.
For example:
* "2:5" -> "5"
* "4:$" -> "$"
* "8" -> "8"
This allows customization of i3bar for alternate ordering of workspaces
which has a legitimate use for alternate keyboard layouts such as
Dvorak.
fixes#1131
Allow client requests of type _NET_ACTIVE_WINDOW to switch workspaces if
they indicate they are a pager. Otherwise, set the urgency hint on that
con to indicate that something happened.
This allows task switchers like skippy-xd to work properly.
http://standards.freedesktop.org/wm-spec/latest/ar01s03.html#idm140251368127856
Change the primary binding accessor to `get_binding_from_xcb_event`.
This function gets a binding from a generic xcb event of type KeyPress,
KeyRelease, ButtonPress, or ButtonRelease by determining the input type
(keyboard or mouse), the modifiers pressed from the filtered event
`state`, managing the proper fall back in case mode switch is enabled,
and finally querying the bindings for a binding that matches the event.
The logic of querying keyboard bindings is not intended to be altered by
this change.
The general accessor has been slightly modified to work with mouse
bindings and made private because it is only used in bindings.c
http://standards.freedesktop.org/wm-spec/latest/ar01s03.html#idm140251368149456
The _NET_CLIENT_LIST property of the root window:
> These arrays contain all X Windows managed by the Window Manager.
Dock clients are not managed windows, so they should not be included in
_NET_CLIENT_LIST or _NET_CLIENT_LIST_STACKING.
Workspaces won't work properly if they start with "__", so reserve that
namespace altogether. Disallow renaming workspaces to reserved namespace
and using reserved namespace in configuration.
Fixes#1209.
The fullscreen_mode event is a window with the "change" property set to
"fullscreen_mode". This event should be emitted whenever a window enters
or exits fullscreen mode.
This event can be used to turn off dpms off when a window is fullscreen
or display the fullscreen container name in the status line for
instance.
Send all the options in the bar block on the barconfig_update event.
This will eventually allow for dynamically updating bar colors with the
`reload` command.
Do not handle click events on dock clients because they are not managed
windows. Dock clients are responsible for sending the message to i3 to
focus a workspace if that is appropriate. i3bar now sends the message to
focus the correct workspace when that is appropriate.
Otherwise, it could interfere with the dock clients own click handling,
which could be an action to focus a different workspace than i3 had
assumed, such as would be the case with a workspace widget.
The log message at the top of ipc_send_window_event would segfault when
no xcb window is present, such as for a split container.
The log now shows the con id which is more relevant anyway.
If a `bindsym` config directive specifies a symbol beginning with
"button", the binding will be given the type B_MOUSE for the indicated
button number.
Example:
bindsym $mod+button2 exec echo 'button two'
This will be interpreted as having input code (now `keycode`) 2 and type
B_MOUSE.
The mechanism to find and run mouse bindings on mouse events is not
implemented.
When checking the window type for a resize command, first check to see
if the window property is null before checking whether or not it is a
dock window. The window may be null in the case it is a branch
container.
fixes#1220
This commit:
commit fb6d117c42
Author: Axel Wagner <mail@merovius.de>
Date: Thu Dec 30 21:09:32 2010 +0100
Port sighandler to tree-branch
blocks SIGPIPE. Unfortunatelly blocked signal set is preserved accross
execve()s, so any program executed by i3 inherited blocked SIGPIPE signal.
This leads to courious effects when pipe writer does not terminate after
exiting pipe reader.
Simple reproducer is to spawn a new xterm by a key shortcut and then run this
command there:
$ perl -e 'print $SIG{PIPE}, qq{\n}'
IGNORE
fixes#1237
If input focus is set by the window manager, it is not necessary to send
WM_TAKE_FOCUS because it has already taken focus.
http://tronche.com/gui/x/icccm/sec-4.html#s-4.1.7
> The goal is to support window managers that want to assign the input
> focus to a top-level window in such a way that the top-level window
> either can assign it to one of its subwindows or can decline the offer
> of the focus. For example, a clock or a text editor with no currently
> open frames might not want to take focus even though the window
> manager generally believes that clients should take the input focus
> after being deiconified or raised.
Both setting input focus and sending WM_TAKE_FOCUS is effectively
setting focus on the window twice which is certainly against the spirit
of the spec, if not the letter.
fixes#1167
Add and update the _NET_CLIENT_LIST property on the root window to
better comply with ewmh standards.
Information on this property can be found here:
http://standards.freedesktop.org/wm-spec/latest/ar01s03.html
> These arrays contain all X Windows managed by the Window Manager.
> _NET_CLIENT_LIST has initial mapping order, starting with the oldest window.
fixes#1099
Otherwise, we read random garbage. Found by the testsuite due to
t/185-scratchpad.t failing because a window was incorrectly recognized
as a floating window.
According to 4.1.7 of the iccm spec
http://tronche.com/gui/x/icccm/sec-4.html#s-4.1.7
> Windows with the atom WM_TAKE_FOCUS in their WM_PROTOCOLS property may
> receive a ClientMessage event from the window manager (as described in
> section 4.2.8) with WM_TAKE_FOCUS in its data[0] field and a valid
> timestamp (i.e. not CurrentTime ) in its data[1] field.
Adds the timestamp parameter to send_take_focus to avoid the dangerous
use of a global variable.
This memory leak is related to matching on con_id or matching when the
container has a null window. In that case, windows that do not match
would leak the owindow struct.
This memory leak is associated with matching by a criteria that uses a
regular expression. Without freeing a regex before calling match_init,
it will definitely be lost.
cmd_resize does not implement a method to resize dock clients.
A command like `[instance=".*"] resize grow width 160 px or 16 ppt`
when a dock client matched would crash i3.
fixes#1201
Windows that match the following criteria are floated by default:
- dialog, utility, toolbar, or splash windows,
- modal windows, or
- windows that have specified equal minimum and maximum size.
closes#1182
The function ipc_recv_message in libi3 allocates memory to the location
of the `message` reply in src/ipc_receive_message.c with malloc and must
be freed.
This memory leak was found using valgrind.
Rename `get_binding` to `get_keyboard_binding` and ensure that this
function only accesses bindings of type B_KEYBOARD. Other types of
bindings (e.g. mouse bindings) will be accessed by a different function.
This patch fixes ticket #1168 to extend the window IPC event mechanism
to send IPC events for window focus and title changes. The newly added
window events use the same format as the already established "new"
event.
Specifically this patch:
* Moves the ipc_send_window_event() function from src/manage.c into
src/ipc.c and adds an argument for the change property of the event
* Updates src/manage.c to use the new function signature. To ensure
that the "new" event does not send the same event data as the
"focus" event, setting focus now happens after the "new" event
has been sent
* Adds IPC focus event notification to src/x.c. To workaround a problem
during window close when accessing the window name, a function has been
added to query whether a window is actually attached to its parent. To
avoid obsolete focus notification, a new field has been added to keep
track of the focus without any interference by the click handling
* Adds IPC title event notification to src/handlers.c. To avoid
obsolete title notification, a function has been added to determine
whether a window title has actually changed
* Updates the IPC documentation to include the new events
* Updates testcases/t/205-ipc-windows.t to include the "focus" event
in order to ensure the correct event sequence
* Adds two new unit tests, b/testcases/t/219-ipc-window-focus.t and
b/testcases/t/220-ipc-window-title.t to ensure proper "focus" and
"title" events
The X11 protocol description states:
The window the event is reported with respect to is called the event
window. The event window is found by starting with the source window
and looking up the hierarchy for the first window on which any client
has selected interest in the event.
For the case of urxvt with URxvt.internalBorder > 0, urxvt sets up a
subwindow for its actual contents that is placed “in the middle” of the
urxvt window. In terms of the X11 protocol, the source window is urxvt’s
window, but urxvt does not select ButtonPress events for that.
Therefore, X11 will go up in the hierarchy and deliver the event to i3
for i3’s window decoration, even though this was not actually a click on
the decoration, but into the managed window.
Therefore, we check whether child != XCB_NONE for clicks on window
decorations and then handle them as a click inside the window.
fixes#1176
Calls ewmh_update_current_desktop on startup to set the
_NET_CURRENT_DESKTOP property. Without this change the property only
gets set after the workspaces have been manipulated. Also exclude
hidden workspaces (i.e. those starting with "__" from the workspace
index.
Adds tests for startup and workspace switching.
The pixmap of a borderless leaf container will not be used except
for the titlebar in a stack or tabs.
Make sure these containers do not (really) have a pixmap because it can
only get in the way.
fixes#1013
Create files bindings.[ch] to contain functions for configuring,
finding, and running bindings.
Use the new function `configure_binding` for binding configuration. This
function adds a binding from config parameters.
Export the function `modifiers_from_str` from config_directives.h.
This change is made in preparation for the new bindmouse functionality.
The pixmap of a borderless container will not be used (except for the
titlebar in a stack or tabs). Make sure these containers do not have a
pixmap because it can only get in the way.
fixes#1013
When the _MOTIF_WM_HINTS property of a window specifies it should have
no title bar, or no decorations at all, respond by setting the border
style of that container to BS_PIXEL or BS_NONE respectively.
This comes from the old Motif window manager. It was originally intended
to specify exactly what sort of decorations a window should have, and
exactly what sort of user input it should respond to. The EWMH spec
intended to replace Motif hints with _NET_WM_WINDOW_TYPE, but it is
still in use by popular widget toolkits such as GTK+ and Java AWT.
i3's implementation simply mirrors Gnome's Metacity. Official
documentation of this hint is nowhere to be found.
For more information see:
https://people.gnome.org/~tthurman/docs/metacity/xprops_8h-source.htmlhttp://stackoverflow.com/questions/13787553/detect-if-a-x11-window-has-decorationsfixes#832
Make sure the command `move <direction>` properly sends the workspace
focus ipc event required for i3bar to be properly updated and redrawn.
Make `ipc_send_workspace_focus_event publicly available from ipc.h for
more flexible event sending.
Change the behavior of movement into a branch with respect to the
position the moving con will be placed within the branch when the
movement is complete.
The correct position is determined by the direction of movement or the
position of the focused-inactive container within the branch.
If the direction of movement is the same as the orientation of the
branch container, append or prepend the container to the branch in the
obvious way. If the movement is to the right or downward, insert the
moving container in the first position (i.e., the leftmost or top
position resp.) If the movement is to the left or upward, insert the
moving container in the last position (i.e., the rightmost or bottom
position resp.)
If the direction of movement is different from the orientation of the
branch container, insert the container into the branch after the
focused-inactive container.
fixes#1060
When `focus_follows_mouse` configuration option is disabled, do not warp
the pointer when focus changes outputs after rendering.
Rationale: this option is meant to be disabled by users who have a setup
where the mouse is usually in the way. These users tend to move the
mouse to a corner or use a utility to hide the pointer when it is not
active. When this user switches focus between outputs, the mouse is
placed in the center of the screen.
This is clearly against the spirit of disabling `focus_follows_mouse`.
Disabling this option now implies disabling "mouse follows focus".
When the user initiates a drag resize, draw the resize bar on the border
of the two involved containers and snap the pointer.
This solution produces cleaner code than the former approach where the
caller obfuscated the click coordinates of the event. This may confuse
someone expecting a true button press event.
Fixes an issue where the resize cursor is not shown when the resize bar
is clicked until the user begins to drag the mouse.
Fixes an issue where focus is not properly updated after the drag is
complete when `focus_follows_mouse' option is set, leaving the pointer
in an unfocused window in some cases.
Fixes an issue where the resize bar may jump a few pixels when the mouse
is first moved.
(Thanks to pbos for suggesting this fix and providing an example
implementation)
This is done by installing a new check watcher that replaces the main
X11 event handler and calling ev_run with EVRUN_ONCE until the dragging
loop left state DRAGGING.
With this commit, other handlers, most notably the redraw handler for
placeholder windows, get a chance to run when dragging (placeholder!)
windows around.
state->initial is set to false before calling x_push_node() since we
began pushing the window stack before pushing changes. Therefore, the
condition could never be true.
OS X doesn't have posix_fallocate() yet, so put
bf760d0241 in
#if defined(__APPLE__)
the cd fails with:
/bin/sh: line 0: cd: include: No such file or directory
so give it a path relative to the top directory
Since the macro PATH_MAX is not defined on every system (GNU/Hurd being
one of those who do not define it), we remove all references to this
macro. Instead, we use a buffer of arbitraty size and grow it when
needed to contain paths.
The commit title is fairly technical, so I’ll try to explain.
Recently, users of GDM3 (I’m sure) and LightDM (I think) have reported
that when switching to a new workspace, the contents of the previous
workspace are still visible. i3bar updates, though, so it is the X11
root window which is not being updated here.
When using GDM3, X11 will be started with -background none, and no
background pixmap or pixel is set. Then, apparently,
gnome-settings-daemon will display a fade animation from whatever is
currently on the window to the destination contents. I think this is to
avoid flickering when logging in, which would occur when just setting a
specific background pixmap or pixel.
So, this commit will, when i3 starts first (not on restarts), copy the
contents of the X11 root window (typicall a grey background, at least on
my machine with GDM3) into a pixmap and set that pixmap as background
pixmap. That way, the content will be preserved and one has a
background, instead of what is perceived as a bug :).
This commit has some chance of breakage, so I’m prepared to revert it
unless we can figure out the issues and roll forward.
Every container 'above' (in the hierarchy) the workspace content should
not be closed when the last child was removed.
Add a check for output content containers and do not handle them. These
cons are at the root of the output container with dockarea siblings.
They may be run through this callback when an output is killed with
RandR.
Fixes an issue that caused content cons to become urgent when the output
is killed with RandR.
fixes: #1121
This removes code duplication, which will be useful for a subsequent
commit.
Furthermore, we now don’t open X11 connections unnecessarily in some
corner cases.
When 'move <direction>' is issued in the context of a container that
borders a workspace, and there is no suitable place within this
workspace for which this container can move, move the container to the
closest output in this direction instead.
Add DRAG_ABORT to enum drag_result_t. DRAG_ABORT will indicate the drag
operation cannot be completed.
Return DRAG_ABORT on UnmapNotify, or when the keyboard or pointer cannot
be grabbed.
Add DRAGGING to return value for drag_result_t. DRAGGING is used
internally by drag_pointer to indicate the drag is in progress.
Change DRAG_CANCEL to DRAG_REVERT to clarify the distinction between
"abort" and "revert/cancel" actions.
Fixes an issue that caused i3 to crash when a user is dragging or
resizing a floating window that becomes destroyed.
[Michael]
This commit should fix problems with people using a non-bourne shell as
login shell, e.g. fish or rc. AFAICT, $SHELL should only be used for
interactive shells, but we just want a bourne shell, not an interactive
shell.
When focusing/moving to outputs, the method of getting the correct
output for a given container fails if the container in question is
floating and only partially mapped on an output screen. This patch
introduces a fail-safe retrieval of the output for any container.
When i3 begins to manage a window, if the window opens on a workspace
that is not visible, the urgency hint on the newly managed window will
be set.
fixes#1088
Reply to _NET_REQUEST_FRAME_EXTENTS by settings _NET_FRAME_EXTENTS
property of the window to widths of the respective borders added by i3.
fixes#1093fixes#1069
This patch introduces a yerror() macro in src/commands.c and also
removes some unused yajl helper macros from src/config_directives.c.
Signed-off-by: Vivien Didelot <vivien.didelot@savoirfairelinux.com>
i3 current behavior hides the binding mode indicator when
workspace buttons are disabled.
This patch adds a new configuration for i3bar called
'binding_mode_indicator' which acts like the workspace_buttons.
It is now possible to configure i3bar to hide the
workspace buttons and keep showing the binding mode indicator.
This should make the hide workspace buttons configuration
more convenient for those who are heavily using binding
modes.
Default value for binding_mode_indicator is true.
If the currently focused window is in fullscreen mode, and a new window
is opened with WM_STATE_FULLSCREEN set, the new window now becomes the
new fullscreen window and gains focus.
This reverts commit 0659a0d98c.
It was a fix for dzen2 setups, but broke several other use cases.
Because dzen2 is not really important since most people use i3bar
nowadays, let’s revert this.
fixes#1062fixes#1068fixes#1070
The effect is that the error handling is much better. posix_fallocate()
will allocate all the requested space, whereas ftruncate() does not.
Before this commit, in case the /dev/shm filesystem is too small to hold
the _contents_ of the log file, i3 will SIGBUS when writing to the shm
logfile. With this commit, it will print an error message on startup,
but continue to run without logging.
This ClientMessage can be used to estimate how big the window will be
before opening it. Java always sends the ClientMessage and checks the
atom that should be set by the window manager, but it seems that the
fallback code path has a race condition.
Let’s see if the situation gets better with this change. I have been
running this patch for about two weeks and have not seen any issues with
it.
fixes#934fixes#709
Add debuglog command that takes toggle|on|off. Add get_debug_logging()
to be able to toggle. Make t/187-commands-parser.t expect 'debuglog'.
Document the debuglog command in userguide.
This commit only goes to “next” because I am not sure whether it
actually makes things better in all cases and want to give it some
testing first.
There was no documented reason behind using the
proportional_{width,height} variables, so I suppose that code was just
stupidity on my part (it was written merely a month after I started this
project in 2009).
fixes#1032
Add shmlog command that takes <size>|toggle|on|off. Separate logbuffer
management into open_logbuffer() and close_logbuffer(). Make
t/187-commands-parser.t expect 'shmlog'. Add update_shmlog_atom() to
update the SHMLOG_PATH. Document the shmlog command in userguide.
Some apps such as Mathematica send a synthetic UnmapNotify event without
properly unmapping their window first. This change makes sure that
happens and fixes an annoying bug with Mathematica where some unmanaged
windows would stay around, but you couldn’t do anything with them.
Thanks to psychon (current awesome maintainer) for helping with the
diagnosis!
fixes#787
The hidden_state and mode of each i3bar instance can now be controlled from within i3.
Therefore, two new i3 command were introduced:
_
bar hidden_state show|hide|toggle [<bar_id>]
show: always show the bar
hide: normal hide mode
toggle: toggle between show and hide (individually for each bar)
_
bar mode dock|hide|invisible|toggle [<bar_id>]
hide,dock: like before
invisible: always keep the bar hidden
toggle: toggle between dock and hide (individually for each bar)
This patch introduces a hidden_state ("hidden_state hide|show") in the
barconfig, which indicates the current hidden_state of each i3bar
instance. It only affects the bar when in hide mode. Additionally, a new
invisible mode was introduced. In order to change the hidden_state or
mode of the bar from i3, a barconfig-update event was introduced, for
which a bar can subscribe and the bar then gets notified about the
currently set hidden_state and mode in its barconfig.
For convenience, an id field ("id <bar_id>") was added to the barconfig, where one can
set the desired id for the corresponding bar. If the id is not specified, i3 will
deterministically choose an id; otherwise, with the previous random approach for finding
a new id, which is actually not shared with i3bar, as it would determine its id on
startup, the event-subscription would be destroyed on reload. Still, this issue remains
when manually changing the bar_id in the config and then reloading.
fixes#833, #651
When moving a fullscreen window to scratchpad with 'move scratchpad', the
focused window would stay fullscreen.
Also, when having a container in fullscreen mode and then focusing a child of
this container and moving it to scratchpad, it would enable fullscreen for
the child window.
This patch fixes both problems, so the scratchpad window is always floating.
In order to distinguish split layouts from tabbed/stacking layouts when
only one window is shown, do not draw the indicator border for tabbed or
stacking layouts.
Note that this is ineffective for dunst’s notifications because dunst
re-raises them as soon as they get obscured. It does work for dzen2
however, which was the original use-case.
fixes#569
This is the only sane way I can think of to deal with the problem that
the screen size may not be dividable by the amount of tabbed children
(e.g. 1280 / 41 = 31.219512…).
fixes#645fixes#791
What I do in this patch is:
1. Split the con_new() function, so I can create a Con without actually
create a window.
2. Store the depth of Cons in the layout file when i3 is restarting.
Fix typos and mis-staged files in previous patch.
_NET_WM_STATE_DEMANDS_ATTENTION indicates that some action in or with
the window happened. It's a weaker hint than urgency flag of WM_HINTS,
but some applications and almost all Qt applications use it instead of
WM_HINTS' urgency flag (one example is Skype).
During smart popup fullscreen handling, display all transient popups
that belong to the respective fullscreen application. A popup window
belongs to another window if the latter is reachable via the path
induced by the WM_TRANSIENT_FOR hints.
fixes#881
Added new event id (I3_IPC_EVENT_WINDOW) so that a an IPC client can
subscribe to events on windows. Added a basic window event that gets
triggered when a window gets successfully reparented. This new event
also dumps the container data, so that IPC clients can get the initial
window name. IPC clients wishing to see window events should subscribe
to 'window'.
This patch adds the following features:
1) Configure a color of the separator via config. It is done like
bar {
colors {
separator #000000
}
}
2) A block can have an integer entry "separator_block_width" which
sets the width of the gap which would follow after the current block.
3) A block can have a boolean entry "separator" and if it is set
to false, then the drawing of the separating line would be disabled.
This change ensures a scratchpad window is still centered on the screen
if it has first been shown on another screen of a different
resolution. Moving or resizing the scratchpad manually disables this
behavior.
fixes#809
This makes the moving behavior more consistent. If you want to focus the
workspace you are moving to, just chain the keybinding in your config
file:
bindsym $mod+Shift+1 move workspace 1; workspace 1
It looks like the code which was removed with this commit was not
necessary anyways since con_move_to_workspace() by now checks on its own
whether it moves to the scratchpad.
fixes#913
We eagerly marked all KeyRelease bindings as eligible to match without
checking modifiers, while this should only be allowed for the key
binding which actually was pressed.
fixes#915
When an X11 window is closed (say, urxvt), i3 gets an UnmapNotify event
and destroys (DestroyWindow) the window decorations. Before this commit,
the DestroyWindow call was sent immediately.
This lead to a situation where — due to the DestroyNotify — EnterNotify
events were generated that would cause the focus to be set to the
underlying window.
With this commit, i3 first renders the tree and pushes changes to X11
before calling DestroyWindow. Therefore, the surrounding containers will
take up any space that was freed by the window which was closed and no
EnterNotify will be generated.
fixes#660
• our function names use underscores
• rewrote the function’s comment
• function comments must be in the source _and_ in the header
• no blank lines after function signatures
Add a `current' and `old' properties to the `focus' change type,
containing the current and old workspace respectively. These additions
are not necessary anywhere else because `focus' is always triggered when
changing ws.
The corresponding command is 'rename workspace to <name>'. As a side-effect
this fixes the command 'rename workspace 1 to to'.
Signed-off-by: Michael Walle <michael@walle.cc>
With this commit, the default behavior is to display popups while there
is a fullscreen application only if the popup belongs to that
application (as determined by the WM_TRANSIENT_FOR hint which
applications have to set properly).
fixes#663
Focusing child from a workspace should now skip over the floating con and
go directly to its child. Focusing parent from that grandchild should leave
the workspace focused again.
As the workspace might be reached via recursion (e.g. moving from the edge
of a fullscreen split container), it's necessary to check for a fullscreen
container whenever a workspace is reached.
If a window with _NET_STARTUP_ID set is moved to another workspace, it
will delete any associated startup sequence immediately. This will also
occur if a window has a leader with _NET_STARTUP_ID set, if the leader
has no container (never been mapped).
A startup sequence may also be deleted if it's matched by
startup_workspace_for_window() and its 30-second timeout has elapsed.
This avoids a case where a fullscreen container could be moved onto a
workspace that already had its own fullscreen container, leading to
two fullscreen containers on top of each other.
Previously, while the loop would not be executed anymore, the callback
(for actually repositioning the window after a MotionNotify) would still
get called once. This commit avoids that, thus fixing the crash.
fixes#819
Fix a crash that occured when moving a window to the scratchpad that is
seemingly floating to the user, but actually a descendant of a floating
parent con (and itself non-floating). If that is the case, move the
floating parent container to scratchpad instead of the window.
fixes#740
This commit breaks floating window keyboard focus order
(t/135-floating-focus.t) when you have > 2 floating windows. Since
keyboard focus is more important than saving one click to raise floating
windows, I revert the commit.
Note that we cannot implement this without keeping a third list (beneath
floating_windows and focus) for the z coordinate of a floating window.
This seems not worth it.
This reverts commit 064be457e5.
This makes our signal handler useless and leads to infinite SIGSEGV
loops because the ev callback handler gets called only from within the
event loop, and control doesn’t necessary get to the event loop…
This reverts commit 514265b529.
If there is a single floating con on a separate workspace that is not
focused, and this con becomes urgent, switching back to that workspace
may result in a crash of i3. This is because while setting the urgency
of parent containers, 'parent' may become NULL in case of floating
containers. This commit checks the validity of parent.
fixes#821
We need to verify that setting the event mask works, and we need to
include StructureNotify to get unmap events at any point in time.
Thanks darkraven for the pointer.
fixes#718
Previously, when you had an urgent container in a stack on some
invisible workspace (say urxvt) and you switched to it, the urgency hint
was not properly cleared.
Previously, i3 would send width=0, height=0 to windows which were put on
workspaces created by an assignment (that is, invisible workspaces,
which do not get rendered normally).
fixes#653
The removed code was add by commit 61b8a62 to fix#564. That bug is cause
by rendering the mplayer window again as a floating window (even if it
has been rendered before as a fullscreen window, at line 202). So simply
checking for fullscreen window is enough to solve this problem. Treating
floating/tiling fullscreen window differently is not needed.
A good visualization of the new algorithm is this:
+--------+
| |
+--------+=| S1 |========================
| | | |
| S0 | +--------+
| | +--------+
+--------+=========| |================
| S2 | +--------+
| | | |
+--------+ | S3 |
| |
+--------+
When focus is on S0, 'focus output right' will first match S1 (the
closest output which overlaps in the highlighted area), then S2, but not
S3 (since S3 does not overlap into the highlighted area).
fixes#669fixes#771
While it’s certainly better and clearer to specify it, we should do the
right thing when the unit is missing, just like CSS for example
(margin: 0; is okay, margin: 0px; too).
If there is a client with an urgency hint on another workspace and
switching to this workspace would cause the urgency to be reset (by
moving the focusing to the client), delay the reset by some time. This
gives the user the chance to see it.
This commit adds the possibility to configure the urgency delay timer
duration using the 'force_display_urgency_hint' directive. Also,
documentation and a testcase was added to allow for automated checks of
the intended behavior.
fixes#482
this implements both the "move container to workspace back_and_forth" command
and movements to the same workspace when auto_back_and_forth is set.
it includes documentation and test suite additions by michael.
it also simplifies the workspace_show_by_name function (making use of
workspace_get accepting NULL pointers).
Introducing a new event to subscribe called mode. It's fired up
when i3 changes binding mode (like switching from default to resize).
IPC guide adjusted also.
Functions such as fprintf() might be unsafe to use in a signal handler,
see http://stackoverflow.com/questions/3941271/#answer-3941563
By using ev_signal, libev will use a tiny signal handler which just
passes on the information and then calls (outside of the signal handler)
our callback function which can use fprintf() and other unsafe
functions.
fixes#803
Before commit 4976fa3350, setting the
layout of workspaces to something else than the default would just mess
up the parent container of the workspace (the content container).
After that commit, it would create an unnecessary split container when
you change the layout _before_ opening any containers. To avoid this, we
now store the layout (similar to how the 'workspace_layout'
configuration directive works) and apply it when the first container is
attached to the workspace.
Fixes#796
If the focused application is in fullscreen mode and 'scratchpad show'
is executed, toggle the fullscreen mode before showing the scratchpad
application. Otherwise, the scratchpad app would not be visible anyways
but focus would be stolen.
fixes#613
The problem was that scratchpad_move() didn’t check whether the source
workspace was focused. Therefore, 'move scratchpad' only worked reliably
interactively , but not when used with criteria.
For the following binding:
# Simulate ctrl+v upon pressing $mod+x
bindsym --release $mod+x exec --no-startup-id xdotool key --clearmodifiers ctrl+v
you can now use either:
1. press $mod, press x, release x, release $mod
2. press $mod, press x, release $mod, release x
fixes#485
The implementation is naive because the user has to generate exactly the
event he specified. That is, if you use this binding:
bindsym --release $mod+x exec import /tmp/latest-screenshot.png
Then it will only be triggered if you hit $mod, hit x, release x,
release $mod. It will not be triggered if you hit $mod, hit x, release
$mod, release x. The reason is that the KeyRelease event in the latter
case will not have the modifier in its flags, so it doesn’t match the
configured binding.
In tabbed mode, the available width (say 1280) is divided by the amount
of child containers (say 3). Before this commit, we just truncated the
result and would end up with 426 + 426 + 426 = 1278 pixels that we
render to. Now we render a bit too much, but that’ll at least not give
us graphics corruption on any side :).
fixes#791
When the workspace layout (formerly orientation) was forced to change
due to a move command, the split container we created was not marked as
split = true, which caused tree_flatten() to errnously kill the contents
of it and thus one window ended up unmanaged.
Also, the logic in tree_flatten() was inverted due to commit de94f6da.
fixes#790
This changes the fact that Firefox would not be launched on the correct
workspace because it marked the startup sequence as completed *before*
actually mapping all of its windows.
To test this, go to workspace 3 and run this command in a terminal:
i3-msg 'exec iceweasel; workspace 4'
That will make i3 start iceweasel (and create a proper startup
notification context for it), then immediately switch to workspace 4
(before iceweasel could possibly start).
The iceweasel window(s) should appear on workspace 3.
Basically, this is the same fix as commit 914ca6cf :-/. Once again, we
called exit() instead of _exit(), but this time it lead to a kill(0,
SIGTERM), effectively killing all processes in the i3 process group,
including i3 itself. The cause for the kill(0) is that nagbar_pid is set
to 0 by fork(), signaling we’re in the child process. The cleanup
handler only checks for nagbar_pid being -1 as a special value, however.
Previously, in case 'layout stacked' (for example) had been called
interactively, con_set_layout would be called with focused->parent,
while with for_window, it’d be called on the actual matching container.
This difference in behavior was the cause for the inability to use
'for_window [class="XTerm"] layout tabbed', which now works \o/, but
more on that below.
The change also allows us to handle the case of the user selecting a
CT_WORKSPACE container properly, that is, by using the special case and
creating a new split container on the workspace which gets all the
contents, but a new layout.
Now, before you are enthusiastic about the change and try to use
for_window magic in your config file, keep in mind: The 'layout' command
acts on the parent split container. That is, when using a line such as
this one:
for_window [class="XTerm"] layout tabbed
…and opening an XTerm when on a workspace with one single other window,
the whole workspace will be set tabbed (just as previously when you
opened an XTerm and sent 'layout tabbed' manually).
Therefore, to open XTerm in its own tabbed split container, you need to
split before:
for_window [class="XTerm"] split v, layout tabbed
The comma here is important! It says that the second command should not
be treated as an entirely unrelated command, but it should also relate
the matching window (while it does work with a ';', that is prone to
race-conditions and should be avoided).
fixes#358
When you have a tabbed container which has had more than one container
but currently has precisely one container, there was a bit of flickering
when switching workspaces occasionally. This commit fixes it by properly
setting the height of the deco_rect (and thus the X11 window) to not
make the old window contents show up for a minimum period of time.
fixes#777
This fixes a compatibility issue with gnome-terminal and xfce’s
terminal, where fullscreening would lead to moving the window and not
displaying the contents properly.
fixes#788
When resizing floating windows, changing the height was not correctly
handled. This commit fixes that and adds testcases for shrinking and
growing the width and height of floating windows.
To automagically do the right thing when rotating monitors with regards
to splith/splitv layout (depending on width/height of the monitor), we
change the orientation of existing workspaces and the first child.
If that first child happens to be a stacked/tabbed con, we cannot change
the layout unconditionally (previously, the orientation was not in the
layout, so we never noticed this problem).
fixes#768
The latter is actually wrong. For example, when running i3
--moreversion, it will print $(pwd)/i3 instead of $(which i3). In my
previous tests, this coincidentally was the same.
This changes the SHM log format, it doesn’t use 0-bytes to separate
entries anymore. Instead of using lots of printf() calls in i3-dump-log,
we now do precisely one big write().
So, to be clear: i3-dump-log and i3 both need to be upgraded.
Mismatching versions will lead to garbage output (no crashes of i3, just
garbage output).
The -f flag uses an inter-process pthread_cond_t in the shared memory
header to broadcast the arrival of new messages to all i3-dump-log
processes. This internally uses futexes and thus doesn’t even mean a
kernel call in most cases. inter-process pthread_cond_ts require NPTL
(the Native Posix Thread Library, introduce in Linux 2.6).
Michael Stapelberg