]> git.sur5r.net Git - i3/i3/log
i3/i3
10 years agodebian: require libyajl >= 2.0.4 (Thanks Koston)
Michael Stapelberg [Fri, 16 May 2014 16:15:53 +0000 (18:15 +0200)]
debian: require libyajl >= 2.0.4 (Thanks Koston)

10 years agouse size_t for strspn() return value
Michael Stapelberg [Fri, 16 May 2014 16:03:31 +0000 (18:03 +0200)]
use size_t for strspn() return value

10 years ago"somewhen" -> "some time" in i3-config-wizard(1)
Jean-Philippe Ouellet [Sun, 11 May 2014 10:23:34 +0000 (06:23 -0400)]
"somewhen" -> "some time" in i3-config-wizard(1)

10 years agoSeparate border width for floating windows
Aleksi Blinnikka [Fri, 9 May 2014 16:33:38 +0000 (19:33 +0300)]
Separate border width for floating windows

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

10 years agoi3bar: implement custom workspace numbers config
Tony Crisci [Mon, 5 May 2014 17:56:47 +0000 (13:56 -0400)]
i3bar: implement custom workspace numbers config

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

10 years agoi3-input: fix (irrelevant) memory leak
Michael Stapelberg [Thu, 15 May 2014 21:50:09 +0000 (23:50 +0200)]
i3-input: fix (irrelevant) memory leak

Given that the code was exit(0)ing directly after using that memory,
it’s not like this has any effect. However, less false positives on the
clang-analyze report pages is a good thing.

10 years agoi3-dmenu-desktop: don’t use smartmatch (it’s experimental)
Michael Stapelberg [Thu, 15 May 2014 20:52:18 +0000 (22:52 +0200)]
i3-dmenu-desktop: don’t use smartmatch (it’s experimental)

This commit should not change functionality at all (famous last words).

10 years agodocs/userguide: fix default key binding (Thanks Maxime)
Michael Stapelberg [Thu, 8 May 2014 06:45:30 +0000 (08:45 +0200)]
docs/userguide: fix default key binding (Thanks Maxime)

10 years agofix “make dist” target
Michael Stapelberg [Sun, 4 May 2014 21:07:53 +0000 (23:07 +0200)]
fix “make dist” target

This was broken in 13db56255134a71acd2fd5a7e254adfbd0c2983b

10 years agoRerender on button press only when focus changes
Tony Crisci [Sun, 4 May 2014 11:50:23 +0000 (07:50 -0400)]
Rerender on button press only when focus changes

On button press events, the only change in state that would presently
require rerendering the tree is when the focused window changes.

10 years agoRemove yajl major version conditionals
Tony Crisci [Sun, 4 May 2014 01:02:54 +0000 (21:02 -0400)]
Remove yajl major version conditionals

Yajl version ≥ 2 is required.

fixes #1156

10 years agoFeature: improve active window request handling
Tony Crisci [Fri, 2 May 2014 20:27:26 +0000 (16:27 -0400)]
Feature: improve active window request handling

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

10 years agoTest workspace empty event semantics
Marco Hunsicker [Tue, 1 Apr 2014 17:33:37 +0000 (19:33 +0200)]
Test workspace empty event semantics

This patchs provides testcases for the workspace "empty" event.

10 years agoRefactor binding accessor
Tony Crisci [Fri, 2 May 2014 14:22:40 +0000 (10:22 -0400)]
Refactor binding accessor

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

10 years agoi3bar: reinit colors on barconfig update
Tony Crisci [Fri, 2 May 2014 23:56:12 +0000 (19:56 -0400)]
i3bar: reinit colors on barconfig update

Allows bar colors to be updated on barconfig update events, such as with
the ipc command `reload`.

10 years agoDont include dock clients in ewmh lists
Tony Crisci [Fri, 2 May 2014 20:54:34 +0000 (16:54 -0400)]
Dont include dock clients in ewmh lists

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.

10 years agoimprove error messages for i3-internal workspace names
Michael Stapelberg [Sat, 3 May 2014 13:02:51 +0000 (15:02 +0200)]
improve error messages for i3-internal workspace names

10 years agoadd a testcase for the previous commit (internal workspaces)
Michael Stapelberg [Sat, 3 May 2014 13:00:49 +0000 (15:00 +0200)]
add a testcase for the previous commit (internal workspaces)

10 years agoMake all workspaces starting with "__" internal
Alexander Berntsen [Wed, 30 Apr 2014 09:12:22 +0000 (11:12 +0200)]
Make all workspaces starting with "__" internal

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.

10 years agorender resize windows retina-correctly
Michael Stapelberg [Fri, 2 May 2014 15:11:48 +0000 (17:11 +0200)]
render resize windows retina-correctly

10 years agos/he/she/, s/his/her/ (Thanks lasers)
Michael Stapelberg [Wed, 30 Apr 2014 07:43:32 +0000 (09:43 +0200)]
s/he/she/, s/his/her/ (Thanks lasers)

10 years agoImplement the window::fullscreen_mode event
Tony Crisci [Mon, 28 Apr 2014 22:38:06 +0000 (18:38 -0400)]
Implement the window::fullscreen_mode event

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.

10 years agoFeature: send complete config on barconfig_update
Tony Crisci [Sun, 27 Apr 2014 05:33:58 +0000 (01:33 -0400)]
Feature: send complete config on barconfig_update

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.

10 years agoDelegate click handling to dock clients
Tony Crisci [Wed, 30 Apr 2014 01:29:17 +0000 (21:29 -0400)]
Delegate click handling to dock clients

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.

10 years agoBugfix: ipc window event crash with no window
Tony Crisci [Mon, 28 Apr 2014 22:21:29 +0000 (18:21 -0400)]
Bugfix: ipc window event crash with no window

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.

10 years agoImplement bindsym mouse configuration
Tony Crisci [Sun, 27 Apr 2014 05:48:28 +0000 (01:48 -0400)]
Implement bindsym mouse configuration

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.

10 years agotruncate DPI to an integer before multiplying (Thanks eeemsi)
Michael Stapelberg [Sat, 26 Apr 2014 20:50:31 +0000 (22:50 +0200)]
truncate DPI to an integer before multiplying (Thanks eeemsi)

This fixes i3bar workspace button rendering on non-DPI screens
(oldschool!)

10 years agodo click handling and tray padding retina-correctly
Michael Stapelberg [Sat, 26 Apr 2014 15:30:27 +0000 (17:30 +0200)]
do click handling and tray padding retina-correctly

10 years agodraw workspace buttons and padded text blocks retina-correctly
Michael Stapelberg [Sat, 26 Apr 2014 15:27:57 +0000 (17:27 +0200)]
draw workspace buttons and padded text blocks retina-correctly

10 years agoi3bar: render separators retina-correctly
Michael Stapelberg [Sat, 26 Apr 2014 15:20:21 +0000 (17:20 +0200)]
i3bar: render separators retina-correctly

10 years agoInitialize default window border width retina-correctly
Michael Stapelberg [Sat, 26 Apr 2014 15:18:21 +0000 (17:18 +0200)]
Initialize default window border width retina-correctly

10 years agolibi3: add logical_px() for Retina display support
Michael Stapelberg [Sat, 26 Apr 2014 15:17:37 +0000 (17:17 +0200)]
libi3: add logical_px() for Retina display support

10 years agoi3-dmenu-desktop: also quote the %c field code (Thanks bo)
Michael Stapelberg [Sat, 26 Apr 2014 12:25:12 +0000 (14:25 +0200)]
i3-dmenu-desktop: also quote the %c field code (Thanks bo)

The spec says:

“Implementations must take care not to expand field codes into multiple
 arguments unless explicitly instructed by this specification. This means
 that name fields, filenames and other replacements that can contain
 spaces must be passed as a single argument to the executable program
 after expansion.”

fixes #1240

10 years agoadd missing docs/layout-saving-1.png
Michael Stapelberg [Sat, 26 Apr 2014 10:46:25 +0000 (12:46 +0200)]
add missing docs/layout-saving-1.png

10 years agodebian: add docs/layout-saving
Michael Stapelberg [Sat, 26 Apr 2014 10:40:30 +0000 (12:40 +0200)]
debian: add docs/layout-saving

10 years agoMove check_for_duplicate_bindings to bindings.[ch]
Tony Crisci [Wed, 23 Apr 2014 01:41:28 +0000 (21:41 -0400)]
Move check_for_duplicate_bindings to bindings.[ch]

Additionally add a check for the same input_type (mouse or keyboard).
Bindings with different input types cannot be duplicates.

10 years agoBugfix: resize window check should check for null
Tony Crisci [Fri, 18 Apr 2014 18:14:22 +0000 (14:14 -0400)]
Bugfix: resize window check should check for null

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

10 years agoReset SIGIPE handler before executing a command
Petr Písař [Thu, 24 Apr 2014 18:35:17 +0000 (20:35 +0200)]
Reset SIGIPE handler before executing a command

This commit:

commit fb6d117c42ce3d9988ff44c079814b3840b1e37f
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

10 years agoadd docs/layout-saving
Michael Stapelberg [Wed, 23 Apr 2014 17:49:50 +0000 (19:49 +0200)]
add docs/layout-saving

Feedback on this new document is very much appreciated, please don’t
hesitate to state anything that is hard to understand/could be improved.

10 years agoFix i3.mk to not rebuild test.* all the time
Michael Stapelberg [Wed, 23 Apr 2014 17:34:45 +0000 (19:34 +0200)]
Fix i3.mk to not rebuild test.* all the time

This problem was introduced in 5181055a

10 years agobuild i3-save-tree(1) manpage (Thanks xeen)
Michael Stapelberg [Wed, 23 Apr 2014 17:33:22 +0000 (19:33 +0200)]
build i3-save-tree(1) manpage (Thanks xeen)

10 years agofix parallel make
Julian Ospald [Tue, 22 Apr 2014 15:09:23 +0000 (15:09 +0000)]
fix parallel make

fixes #1232

10 years agorespect CFLAGS in linking command
Julian Ospald [Tue, 22 Apr 2014 15:09:23 +0000 (15:09 +0000)]
respect CFLAGS in linking command

fixes #1233

10 years agoDon’t overwrite existing windows with placeholder windows
Michael Stapelberg [Fri, 18 Apr 2014 18:36:56 +0000 (20:36 +0200)]
Don’t overwrite existing windows with placeholder windows

This could happen when you appended a layout on a workspace where there
are already other windows.

10 years agolayout restoring: append at the nearest split container (or workspace) (Thanks chris)
Michael Stapelberg [Fri, 18 Apr 2014 18:30:27 +0000 (20:30 +0200)]
layout restoring: append at the nearest split container (or workspace) (Thanks chris)

Before this commit, leaf containers (such as terminal emulators) would
get children appended, which is not intended.

fixes #1223

10 years agoMove switch_mode to bindings.[ch]
Tony Crisci [Mon, 14 Apr 2014 18:52:56 +0000 (14:52 -0400)]
Move switch_mode to bindings.[ch]

10 years agoDont set input focus and send WM_TAKE_FOCUS
Tony Crisci [Thu, 10 Apr 2014 17:28:14 +0000 (13:28 -0400)]
Dont set input focus and send WM_TAKE_FOCUS

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

10 years agoMaintain the _NET_CLIENT_LIST property
Tony Crisci [Tue, 8 Apr 2014 18:27:40 +0000 (14:27 -0400)]
Maintain the _NET_CLIENT_LIST property

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

10 years agoBugfix: clear wm_size_hints if they are not set
Michael Stapelberg [Wed, 9 Apr 2014 20:14:50 +0000 (22:14 +0200)]
Bugfix: clear wm_size_hints if they are not set

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.

10 years agoFix t/158-wm_take_focus, it was not properly verifying events (Thanks TonyC)
Michael Stapelberg [Wed, 9 Apr 2014 19:47:32 +0000 (21:47 +0200)]
Fix t/158-wm_take_focus, it was not properly verifying events (Thanks TonyC)

See also http://cr.i3wm.org/patch/500 for TonyC’s patch. This version
reduces code duplication.

10 years agoBugfix: Use DOM loaded event for TOC (Thanks TonyC)
Michael Stapelberg [Wed, 9 Apr 2014 18:37:24 +0000 (20:37 +0200)]
Bugfix: Use DOM loaded event for TOC (Thanks TonyC)

See http://code.i3wm.org/i3-website/commit/?id=d8e239e6 for the original
commit for the i3 website. This one is for documentation built from git.

10 years agoSend last event timestamp with WM_TAKE_FOCUS msg
Tony Crisci [Sat, 29 Mar 2014 04:25:52 +0000 (00:25 -0400)]
Send last event timestamp with WM_TAKE_FOCUS msg

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.

10 years agoOpenBSD does not have librt
David Coppa [Wed, 26 Mar 2014 09:01:50 +0000 (10:01 +0100)]
OpenBSD does not have librt

10 years agoi3-nagbar: add missing newline when printing version information
Arun Persaud [Sun, 23 Mar 2014 17:49:20 +0000 (10:49 -0700)]
i3-nagbar: add missing newline when printing version information

10 years agoi3-nagbar.man: updated man page to include all options
Arun Persaud [Sun, 23 Mar 2014 17:49:52 +0000 (10:49 -0700)]
i3-nagbar.man: updated man page to include all options

several options were not documented in the man page.

10 years agoi3bar: Send mouse wheel events to child too
Quentin Glidic [Thu, 13 Mar 2014 12:58:15 +0000 (13:58 +0100)]
i3bar: Send mouse wheel events to child too

Signed-off-by: Quentin Glidic <sardemff7+git@sardemff7.net>
10 years agoFree owindow when no con_id match or window
Tony Crisci [Wed, 19 Mar 2014 08:56:34 +0000 (04:56 -0400)]
Free owindow when no con_id match or window

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.

10 years agoFree match after criteria initialization
Tony Crisci [Wed, 19 Mar 2014 08:34:42 +0000 (04:34 -0400)]
Free match after criteria initialization

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.

10 years agodocs: fix typo
jeroentbt [Mon, 17 Mar 2014 00:24:15 +0000 (01:24 +0100)]
docs: fix typo

10 years agofix typo
jeroentbt [Sun, 16 Mar 2014 23:59:01 +0000 (00:59 +0100)]
fix typo

10 years agoTest 222: call exit_gracefully
Tony Crisci [Wed, 19 Mar 2014 02:38:01 +0000 (22:38 -0400)]
Test 222: call exit_gracefully

Not calling exit_gracefully causes some other test to fail during a
complete_run sometimes.

10 years agoTest 175: use BAIL_OUT instead of `die`
Tony Crisci [Wed, 19 Mar 2014 03:07:01 +0000 (23:07 -0400)]
Test 175: use BAIL_OUT instead of `die`

BAIL_OUT produces a friendlier message explaining why the tests cannot
continue on the command line. `die` produces a cryptic message that some
test failed for some reason.

10 years agoMerge branch 'master' into next
Michael Stapelberg [Sat, 15 Mar 2014 17:05:45 +0000 (18:05 +0100)]
Merge branch 'master' into next

10 years agoMerge branch 'fix-resize'
Michael Stapelberg [Sat, 15 Mar 2014 17:05:45 +0000 (18:05 +0100)]
Merge branch 'fix-resize'

10 years agoBugfix: don't try to resize docks by command
Tony Crisci [Sat, 15 Mar 2014 02:59:08 +0000 (22:59 -0400)]
Bugfix: don't try to resize docks by command

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

10 years agoMove translate_keysyms to bindings.[ch]
Tony Crisci [Thu, 13 Mar 2014 12:02:07 +0000 (08:02 -0400)]
Move translate_keysyms to bindings.[ch]

Additionally add a check so the function only handles bindings of type
B_KEYBOARD to prepare for the new bindmouse feature.

10 years agoAdd valgrind suppression file to testcases
Tony Crisci [Mon, 10 Mar 2014 08:47:40 +0000 (04:47 -0400)]
Add valgrind suppression file to testcases

The suppression file makes valgrind output more readable by hiding
reports of memory leaks for GObject-related initialization functions in
Pango and Cairo.

10 years agoFree window ran_assignments on tree_close
Tony Crisci [Mon, 10 Mar 2014 08:29:44 +0000 (04:29 -0400)]
Free window ran_assignments on tree_close

10 years agoUpdate ipc library list documentation
Tony Crisci [Mon, 10 Mar 2014 01:33:26 +0000 (21:33 -0400)]
Update ipc library list documentation

Add a link to the new JavaScript, Python, and Lua ipc libraries.

Add notice that the older Python libraries are not maintained.

Sort the list alphabetically.

10 years agoSet con window to NULL on tree_close
Tony Crisci [Mon, 10 Mar 2014 04:38:52 +0000 (00:38 -0400)]
Set con window to NULL on tree_close

Prevents invalid reads on the freed window struct.

10 years agoAdd test for floating types
Tony Crisci [Sat, 8 Mar 2014 13:18:56 +0000 (08:18 -0500)]
Add test for floating types

Add a test that verifies that windows with properties that indicate they
should be floating are indeed opened floating.

10 years agoImproved detection of windows that want floating
Kernc [Sat, 15 Feb 2014 15:49:41 +0000 (16:49 +0100)]
Improved detection of windows that want floating

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

10 years agoMerge branch 'fix-recv-memleak'
Michael Stapelberg [Sat, 8 Mar 2014 19:59:18 +0000 (20:59 +0100)]
Merge branch 'fix-recv-memleak'

10 years agoMerge branch 'master' into next
Michael Stapelberg [Sat, 8 Mar 2014 19:59:18 +0000 (20:59 +0100)]
Merge branch 'master' into next

10 years agoBugfix: ipc_receive_message reply leak
Tony Crisci [Sat, 8 Mar 2014 04:56:25 +0000 (23:56 -0500)]
Bugfix: ipc_receive_message reply leak

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.

10 years agoUpdate ipc COMMAND reply documentation
Tony Crisci [Fri, 28 Feb 2014 18:45:22 +0000 (13:45 -0500)]
Update ipc COMMAND reply documentation

The COMMAND reply consists of a list of results for each command that
was parsed.

10 years agoMove keyboard binding accessor to bindings.[ch]
Tony Crisci [Sat, 22 Feb 2014 00:10:21 +0000 (19:10 -0500)]
Move keyboard binding accessor to bindings.[ch]

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.

10 years agoFix current_workspace outputs reply member docs
Tony Crisci [Sun, 23 Feb 2014 09:36:39 +0000 (04:36 -0500)]
Fix current_workspace outputs reply member docs

The current_workspace member of an ipc OUTPUTS reply is the name of the
workspace as a string.

10 years agoSend IPC window events for focus and title changes
Marco Hunsicker [Sat, 22 Feb 2014 10:52:01 +0000 (11:52 +0100)]
Send IPC window events for focus and title changes

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

10 years agotests: sync_with_i3 before warping pointer
Michael Stapelberg [Tue, 25 Feb 2014 19:44:17 +0000 (20:44 +0100)]
tests: sync_with_i3 before warping pointer

To be honest, I’m not entirely sure where the race condition originates
from, but when making sure that there are no pending events
(which is what sync_with_i3 accomplishes) before warping the pointer, we
have less flaky testsuite runs.

closes #1189

10 years agoman/i3.man: Update font doc to reflect i3.config
Alexander Berntsen [Sat, 15 Feb 2014 12:00:14 +0000 (13:00 +0100)]
man/i3.man: Update font doc to reflect i3.config

10 years agoMove grab_all_keys to bindings.[ch]
Tony Crisci [Fri, 14 Feb 2014 23:38:01 +0000 (18:38 -0500)]
Move grab_all_keys to bindings.[ch]

Also add checks for binding input_type to filter bindings that are not
keyboard bindings.

10 years agoi3bar: Do not kill watcher on EOF
Tony Crisci [Fri, 14 Feb 2014 01:17:18 +0000 (20:17 -0500)]
i3bar: Do not kill watcher on EOF

Wait for the child process to exit on its own before freeing
watcher-related resources.

i3bar shows the last received status line until the process exits.

Fixes a race condition where the termination signal was sometimes not
received to display a meaningful error message.

10 years agoImprove drag_pointer error handling
Tony Crisci [Wed, 12 Feb 2014 10:16:37 +0000 (05:16 -0500)]
Improve drag_pointer error handling

Log errors generated by failed pointer grabs and ungrab the pointer
after failed keyboard grabs.

10 years agot/130-close-empty-split: clean up old workspace
Michael Stapelberg [Thu, 13 Feb 2014 20:05:42 +0000 (21:05 +0100)]
t/130-close-empty-split: clean up old workspace

This hopefully increases the reliability of the test. Empty containers
created by “open” sometimes get spurious UnmapNotify events, and in at
least one run, this test failed because of such an event.

10 years agohandle ButtonPress events with child != XCB_NONE (Thanks xeen)
Michael Stapelberg [Sun, 9 Feb 2014 13:00:43 +0000 (14:00 +0100)]
handle ButtonPress events with child != XCB_NONE (Thanks xeen)

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

10 years agofix warnings: unsigned check for >= 0 is always true
Michael Stapelberg [Sat, 8 Feb 2014 20:03:26 +0000 (21:03 +0100)]
fix warnings: unsigned check for >= 0 is always true

10 years agoBugfix: `tree_split` should not split floating cons
Tony Crisci [Fri, 7 Feb 2014 15:07:56 +0000 (10:07 -0500)]
Bugfix: `tree_split` should not split floating cons

Use the more robust test of `con_is_floating` for testing a container
for being floating.

fixes #1177

10 years agoDoc update: title right click resizes floating windows
Jonas Maaskola [Fri, 7 Feb 2014 10:52:14 +0000 (11:52 +0100)]
Doc update: title right click resizes floating windows

This documentation update describes the change implemented in
905440d6d1c6d2a872e3bbb0da4c963bedaf5708
It avoids repetitive usage of the word 'also'.

10 years agoSet EWMH desktop properties on startup.
Steve Jones [Sat, 1 Feb 2014 16:09:51 +0000 (16:09 +0000)]
Set EWMH desktop properties on startup.

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.

10 years agoResize floating windows by right-clicking decoration
Jonas Maaskola [Wed, 5 Feb 2014 16:29:42 +0000 (17:29 +0100)]
Resize floating windows by right-clicking decoration

10 years agoDocument the existence of a C ipc project
Tony Crisci [Tue, 4 Feb 2014 15:52:52 +0000 (10:52 -0500)]
Document the existence of a C ipc project

Add a link to https://github.com/acrisci/i3-ipc which is a new ipc
library in the design phase of development. When it is stable, it will
provide bindings to many high-level scripting languages with
GObject-introspection.

This project aims to replace the unmaintained Python library and offer
an ipc library in new languages such as Lua and JavaScript.

10 years agoReally do not create con pixmap when not needed
Tony Crisci [Tue, 4 Feb 2014 18:36:20 +0000 (13:36 -0500)]
Really do not create con pixmap when not needed

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

10 years agoi3bar: fix resource leak: statusline_ctx needs to be freed first
Michael Stapelberg [Sun, 2 Feb 2014 10:22:42 +0000 (11:22 +0100)]
i3bar: fix resource leak: statusline_ctx needs to be freed first

In practice this is rarely noticeable, unless you very often switch
screen resolutions, I think.

fixes #1172

10 years agoMerge branch 'fix-i3bar-leak'
Michael Stapelberg [Sun, 2 Feb 2014 10:23:54 +0000 (11:23 +0100)]
Merge branch 'fix-i3bar-leak'

10 years agoi3bar: fix resource leak: statusline_ctx needs to be freed first
Michael Stapelberg [Sun, 2 Feb 2014 10:22:42 +0000 (11:22 +0100)]
i3bar: fix resource leak: statusline_ctx needs to be freed first

In practice this is rarely noticeable, unless you very often switch
screen resolutions, I think.

fixes #1172

10 years agoRevert "Do not create container pixmap when not needed"
Michael Stapelberg [Thu, 30 Jan 2014 21:55:26 +0000 (22:55 +0100)]
Revert "Do not create container pixmap when not needed"

This reverts commit 6578976b6159437c16187cf8d1ea38aa9fec9fc8.

fixes #1170

10 years agoAbstract binding configuration to bindings.[ch]
Tony Crisci [Tue, 28 Jan 2014 03:45:01 +0000 (22:45 -0500)]
Abstract binding configuration to bindings.[ch]

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.

10 years agocontrib: add per-workspace-layout.pl
Michael Stapelberg [Sun, 26 Jan 2014 15:51:32 +0000 (16:51 +0100)]
contrib: add per-workspace-layout.pl

10 years agocontrib: add sticker that was actually printed
Michael Stapelberg [Sun, 26 Jan 2014 15:43:02 +0000 (16:43 +0100)]
contrib: add sticker that was actually printed

While sticker_stickma_black.svg is the file that we originally provided,
the file that we got back from stickma and actually ordered is the new
one.