There are two ways in which the input focus can change: Lua can request
a change and the X11 server can inform us that the input focus changed
(because some application changed it).
In the first case, we still have to inform the X11 server about the
desired change, in the second case we must not to avoid races due to
X11's asynchronous nature.
However, there was a case where we screwed up: When a focus change is
still pending, meaning that Lua assigned the focus elsewhere, but we
have not yet sent this focus change to the X11 server, we could get an
event from the X11 server telling us that the focus changed. To make
sure that the pending focus change is not lost, we sent the focus change
out in this case (call to client_focus_refresh() in
event_handle_focusin()). After sending out this pending call, we updated
the internal state to record that whatever the X11 server just told us
had the focus. The intention was that our just sent-out focus change
will cause the X11 server to send a new event and our to-be-focused
client then has the focus.
However, if the pending focus change was for a client which only showed
up in this event loop iteration, the client was still banned. This means
that client_focus_refresh() would call client_unban() to be able to give
the focus to this client. However, since awesome (partly) allows to
"focus" currently banned clients, client_unban() recorded that there is
a pending focus change. This caused confusion later on.
In this specific bug, a main window opened a dialog, and when this
dialog was closed, a new dialog window was opened immediately. When the
first dialog was closed, Lua (the focus history) gave the input focus to
the main window. Now, a new dialog showed up and Lua focused it. Next,
we received the event from the X11 server telling us that the main
window was focused. Because there was still a pending focus change to
the new dialog window, event_handle_focusin() called
client_focus_refresh() to send out this focus change. This set
globalconf.focus.need_update to false and continued. However, because
the new dialog only just now appeared, it was still banned, meaning that
client_focus_refresh() had to call client_unban(). This set
globalconf.focus.need_update to true. Thus, when client_focus_refresh()
returned, globalconf.focus.need_update was incorrectly true. Next,
event_handle_focusin() recorded that the main window had the focus.
Thus, it now appeared as if there was a pending focus change for the
main window. Next, we got the event from the X11 server telling us that
the dialog is now focused, and because focus.need_update was set,
awesome now send out a focus change request for the main window.
Fix this race by unsetting globalconf.focus.need_update at the end of
client_focus_refresh() and not at the beginning, thus making sure that
client_unban() cannot set this flag again.
Fixes: https://github.com/awesomeWM/awesome/issues/2220
Signed-off-by: Uli Schlachter <psychon@znc.in>
When a window has a WM_TRANSIENT_FOR property that is later unset,
awesome would still keep c.transient_for pointing to the previous
"parent client". This commit fixes that.
First, property_update_wm_transient_for() is fixed so that it unsets
c->transient_for_window if the WM_TRANSIENT_FOR property is deleted.
Additionally, this then calls client_find_transient_for() to update the
c->transient_for pointer.
Secondly (and a bit unrelated), this changes client_find_transient_for()
so that it always sets c->transient_for. Previously, if updating this
property would introduce a cycle in the transient_for relation, it would
just leave c->transient_for with its old value. After this change, it
gets explicitly set to NULL instead.
Signed-off-by: Uli Schlachter <psychon@znc.in>
This makes motif wm hints available on clients as c.motif_wm_hints.
Actually interpreting all the values is up to Lua. The definition of the
necessary properties is taken from motif.
Signed-off-by: Uli Schlachter <psychon@znc.in>
The getters for properties already get the object as their second
argument, so there is no need to get the object again from Lua.
Fixes: https://github.com/awesomeWM/awesome/issues/2299
Signed-off-by: Uli Schlachter <psychon@znc.in>
When a new client appears, we reparent its window into a window of our
choice so that we can add decorations. However, windows can be difficult
making the reparenting fail. For example, some snapshots of GTK+ set a
parent-relative background, which means that its parent windows have to
have the same depth as the window itself. This makes reparenting fail if
transparency is used/supported.
Make the failure mode... different, by checking if reparenting
succeeded. If it failed, we print an error message and unmanage the
window immediately. This means that those GTK+ windows just do not
appear at all, instead of behaving weirdly.
This idea of this patch came from i3.
Related-to: https://github.com/awesomeWM/awesome/issues/2279
Signed-off-by: Uli Schlachter <psychon@znc.in>
_NET_WM_STATE could contain _NET_WM_STATE_HIDDEN, which we interpret as
minimized. Minimized clients have WM_STATE set to WM_STATE_ICONIC, but
the code in client_manage() would later overwrite this to
WM_STATE_NORMAL.
Fix this by setting the initial WM_STATE_NORMAL by doing so before
processing _NET_WM_STATE.
Fixes: https://github.com/awesomeWM/awesome/issues/2095
Signed-off-by: Uli Schlachter <psychon@znc.in>
The code here made sure that clients were not moved outside of the root
window. However, that's not enough, because clients can still end up
inside the root window, but outside of anything that is visible in some
output. Thus, just remove this.
Related-to: https://github.com/awesomeWM/awesome/issues/318
Signed-off-by: Uli Schlachter <psychon@znc.in>
Aborting the process is sometimes a bit harsh for a failed assertion.
This adds a non-fatal assert() macro called "check()" and uses it in
some places where we might be able to survive the error.
Signed-off-by: Uli Schlachter <psychon@znc.in>
Every call to client_ignore_enterleave_events() must be paired with a
following call to client_restore_enterleave_events(). In between these
two calls, no other calls to client_ignore_enterleave_events() is
allowed.
The code in banning_refresh() sometimes broke these rules. This can
happen because the code causes signals to be emitted and Lua code can do
basically anything.
Fix this by moving the calls into the called functions.
Fixes: https://github.com/awesomeWM/awesome/issues/1746
Signed-off-by: Uli Schlachter <psychon@znc.in>
This adds c.icon_sizes which is a table containing the width and height
of each available icon. With c:get_icon(i), Lua can query the i-th icon
as a lightuserdata.
Signed-off-by: Uli Schlachter <psychon@znc.in>
Clients can provide various icons in their _NET_WM_ICON property. Up to
now we only saved a single one, now we save all of them.
Signed-off-by: Uli Schlachter <psychon@znc.in>
Before this commit, do this:
c.maximize_hoizontal = true
c.maximize = true
c.maximize = false
assert(c.maximize_hoizontal)
Would not work because the states were not preserved individually.
This commit fixes that. Awesome wont be confused about it's own
state anymore.
This may seem pointless, but when it come to undoing these
maximizations, it was ambiguous.
Before 4.0, maximizing could only be done in 2 operations.
4.0 add an unified way, but kept doing 2 operations. The old
Lua EWMH code to serialize the 2 operations was dropped when
the codepath was simplified and replaced by a generic version
in awful.placement. However this version never implemented
combining multiple mementos into 1.
This commit unify the maximize C code, drop the ugly macro
template and actually fixes a couple more issues that were
caused because request::geometry was sent twice.
I explicitly did not add client_shape_input property since querying the
input shape of the client window seems useless to me.
Signed-off-by: Uli Schlachter <psychon@znc.in>
These warnings might help catching some problems in the future. These
could be asserts, but printing a warning is a lot nicer than dying.
Signed-off-by: Uli Schlachter <psychon@znc.in>
X11 does not allow to resize a window to size 0x0. Also, there are some
possibilities of integer overflows in our case. We tried to handle this
already, but there was a loop-hole: If the too-small-value is only
produced after applying size hints, then this was not caught.
Fix this by applying size hints before checking if the resulting size is
valid. However, this means some check needs to be duplicated to handle
the possibility of integer underflows while applying size hints.
Helps-with: https://github.com/awesomeWM/awesome/issues/1340
Signed-off-by: Uli Schlachter <psychon@znc.in>
The code in luaA_client_swap() is incorrect, because
luaA_object_emit_signal() already pops the arguments to the signal.
Still, the code here tried to remove the arguments from the Lua stack
again, thereby corrupting the stack (removing more items than there are
in the stack).
Normally, popping more things from the stack than it has entries
silently corrupts the Lua stack. Apparently this doesn't necessarily
cause any immediate issues, because this code has been broken since nine
months and no one noticed. This mistakes was introduced in commit
55190646.
This issue was only noticed by accident. Thus, this commit also adds a
small integration test that exercises this bug. This test catches the
issue, but only on Travis, because there we are building our own version
of Lua 5.3 and that one has assertions enabled.
Signed-off-by: Uli Schlachter <psychon@znc.in>
It does not provide much value. The version number is already known to
ldoc globally in the "description" variable.
Signed-off-by: Uli Schlachter <psychon@znc.in>
Daniel sees a short flicker of his wallpaper when he closes a client.
This happens because the window is destroyed immediately, but other
clients are re-arranged only shortly later. In the mean time, the X
server updates the display and repaints the root window (= wallpaper
becomes visible).
Work around this by delaying the destruction of frame windows to the end
of the current main loop iteration. This means that we first update the
position of all other windows and later destroy the window that was
actually closed.
Signed-off-by: Uli Schlachter <psychon@znc.in>
There are some situations where we do things that can make the mouse pointer
enter another window. We do not want to react to these "self inflicted" mouse
enter and leave events, because they aren't "real" (= generated by the user).
Before this commit, this is done by going through all windows and toggling the
"please send us enter and leave events"-bit on them. This becomes slower when
many windows are visible and floods the server with requests.
This commit changes this to a constant-time logic. Each event contains the
sequence number of the last request that the X11 server handled. Thus, we just
remember the right sequence numbers and ignore any events that comes in whose
sequence number falls into the ignored range.
In detail, we keep a list of "begin" and "end" sequence numbers and ignore any
enter and leave events that fall in this range. If we get any event with a
sequence number higher than "end", we remove this pair from the list, since it
is no longer needed.
To generate these pairs, we use a GrabServer request in
client_ignore_enterleave_events(). This gives us a sequence number and makes
sure that nothing else besides us can cause events. The server is ours! In
client_restore_enterleave_events(), we first do a NoOperation request to
generate the sequence number for the end of the pair and then do UngrabServer.
Any event that is generated after UngrabServer will have at least the sequence
number of the UngrabServer request and thus no longer fails between begin and
end.
Fixes: https://github.com/awesomeWM/awesome/issues/1107
Signed-off-by: Uli Schlachter <psychon@znc.in>
This commit makes the function only call client_ignore_enterleave_events() when
it actually has to. Since we expect that most of the time, no client's geometry
is changed, this means that most of the time this function is not called.
Fixes: https://github.com/awesomeWM/awesome/issues/1107
Signed-off-by: Uli Schlachter <psychon@znc.in>
This should "protect" the user from some stupidities that Lua code might be
doing that e.g. makes a client jump to another position and then immediately
back to where it was before. Only the last change in a single main loop
iteration will actually have any effect.
Original idea by Daniel here: https://github.com/awesomeWM/awesome/pull/174
Signed-off-by: Uli Schlachter <psychon@znc.in>
A client c could have no c.machine or no c.pid because the corresponding
properties are not set on its window. Previously, the C code would return an
empty string or 0 for these values. This commit makes the C code give Lua no
value instead (not even a nil).
Signed-off-by: Uli Schlachter <psychon@znc.in>
Instead of computing the workarea whenever some Lua code asks for it, it is now
remembered explicitly as a property on a screen. This allows us to only emit
property::workarea if the workarea actually changed.
Fixes: https://github.com/awesomeWM/awesome/issues/756
Signed-off-by: Uli Schlachter <psychon@znc.in>
Apparently the spec allows to set the _NET_STARTUP_ID value on the property that
WM_CLIENT_LEADER points to instead of the window itself. Thus, if we don't find
a _NET_STARTUP_ID on the window itself, check again on the client leader window.
Apparently GTK even does this (for whatever reason...)...
Signed-off-by: Uli Schlachter <psychon@znc.in>
Remove request::fullscreen and request::maximized_* and use
a single request for them. The other client resizing features
will soon also start to use this.
In the case where one want to put the cursor at the middle of the
workarea, it is logic to do:
x=screen.workarea.x+screen.workasrea.width/2
However, this can cause floating points. This commit move the
burden back to the C-API so the Lua placement code doesn't have
to add a large number of rounding methods. Given 1 type of rounding
cover a vast majority of use cases for each types of coordinates,
the C-API can take care of it in peace. For the other corner cases,
it is still possible for the Lua code to do the rounding there, but
no longer necessary. The convenstions are:
'x' and 'y': use round (move to the closest point)
'width' and 'height': use ceil (to avoid involontary truncating)
This change catches things like c:geometry { width = -42 }.
Helps-a-bit-with: https://github.com/awesomeWM/awesome/pull/820 (fixes X errors)
Signed-off-by: Uli Schlachter <psychon@znc.in>
Instead of focusing the root window, we now create a "focus window" inside of
our frame window. This window is placed so that it is not visible, but we can
grab key bindings on it to simulate the window having the input focus.
Fixes: https://github.com/awesomeWM/awesome/issues/699
Signed-off-by: Uli Schlachter <psychon@znc.in>
Even when a screen is just an integer, the code becomes a bit more
self-documenting. Even better, if we start to handle screen objects to Lua
instead of screen indicies, there will only be one place that needs to be
changed.
Signed-off-by: Uli Schlachter <psychon@znc.in>
When we manage the transient before the main window, the client object's
.transient_for property would stay nil. This happens because the property points
to a window which we don't know (yet) and thus is ignored.
Fix this by remembering the value of WM_TRANSIENT_FOR and checking in
client_manage() if the new client is the "missing window we did not find
before".
Fixes: https://github.com/awesomeWM/awesome/issues/181
Signed-off-by: Uli Schlachter <psychon@znc.in>
This creates a new helper function for setting the transient_for property of a
client. This is a preparation for a following commit. No behaviour changes
intended.
Signed-off-by: Uli Schlachter <psychon@znc.in>
To quote from ICCCM (§4.1.2): "The window manager will not change properties
written by the client."
We tried to do this anyway to update WM_HINTS so that the current urgency state
is reflected. Apparently, Chrome does a similar read-modify-set cycle and the
resulting race condition meant that the "accepts input" hint on Chromium's
window was permanently disabled.
This helps with https://github.com/awesomeWM/awesome/issues/670, but I still
think that Chrome shouldn't try to implement "please don't focus me when I do
the following" by temporarily claiming "please don't ever focus me".
Signed-off-by: Uli Schlachter <psychon@znc.in>
ICCCM specifies when the WM has to send a ConfigureNotify. Java does not care
and wants one all the time. Meh.
Fixes: #248
Signed-off-by: Uli Schlachter <psychon@znc.in>
This allow layout "arrange" to be called less often and react on
the cause of the change itself rather than it's consequences
(usually, the "focus" signal).
Previously, the layout were re-arranged everytime the focus changed.
Now, with "raised" and "lowered", it require less "arrange".
"swapped" allow smarted layouts. Currently, swapped cause a full
re-arrange. It re-read the "index" list from scratch and create
a "new" layout. With "swapped", incremental layout changes are
possible.
Fixes https://github.com/awesomeWM/awesome/issues/616
When a client is unmanaged, we know emit mouse::leave on its titlebar before the
client object is invalidated, so that Lua can still work with it. Before, this
event was emitted only when we got a LeaveNotify from the X11 server.
Fixes: #620
Signed-off-by: Uli Schlachter <psychon@znc.in>
Before this, we grabbed the keys on the frame window. That meant we only got key
events for things that nothing else grabbed directly on the key window.
After this, we grab directly on the client window itself and so we "fight" with
everything else which wants to grab keys. I don't actually know how the winner
is decided... First come, first serve, the rest gets an error?
Signed-off-by: Uli Schlachter <psychon@znc.in>
The bit that indicates that the base size is set is
XCB_ICCCM_SIZE_HINT_BASE_SIZE. However, instead this code checked
XCB_ICCCM_SIZE_HINT_P_SIZE which is set to indicate how the initial window
position is chosen. So we were checking the complete wrong bit. Whoops...
Fixes: https://github.com/awesomeWM/awesome/issues/456
Signed-off-by: Uli Schlachter <psychon@znc.in>
Together with the previous changes, this also fixes the initial positions for
metacity's test-gravity.c.
Signed-off-by: Uli Schlachter <psychon@znc.in>
Whenever client.focus == nil, we set the input focus to the root window to
express "nothing has the input focus". However, thanks to the way X11 input
works, this means that whatever is under the mouse cursor gets keyboard input
events. This can easily be reproduced with urxvt and some small addition to the
config to unfocus things.
This commit changes things. Instead of focusing the root window, we create a
special "no focus" window that gets focused if we want nothing to have the
focus.
Closes https://github.com/awesomeWM/awesome/pull/470.
Signed-off-by: Uli Schlachter <psychon@znc.in>
This makes it possible to add something similar to a __index / __newindex
metamethod to all our C objects. Based on this, Lua can then easily implement
arbitrary properties on our capi objects.
From http://standards.freedesktop.org/wm-spec/latest/ar01s05.html:
> _NET_WM_STATE_FULLSCREEN indicates that the window should fill the
> entire screen and have no window decorations. Additionally the Window
> Manager is responsible for restoring the original geometry after a
> switch from fullscreen back to normal window. For example, a
> presentation program would use this hint.
awesome prefers fullscreen internally already. With this patch, the
previous maximized state will be restored after leaving fullscreen mode.
Fixes https://github.com/awesomeWM/awesome/issues/245.
Closes https://github.com/awesomeWM/awesome/pull/418.
This will skip calling `client_resize_do` in case of honored size hints.
This could be also done in `client_resize_do`, but it appears to be
meant to force the resize.
Closes https://github.com/awesomeWM/awesome/pull/383.
Instead of `client.client`, the client object is now referred to as
`client.object` and the client class as `client.class`.
This moves the documentation of `client.focus` to the class.
Closes https://github.com/awesomeWM/awesome/pull/349.
Uli Schlachter
Daniel Hahler
Emmanuel Lepage Vallée
mergify-bot
Uli Schlachter
Daniel Hahler
Luke Bonham
Vitja Makarov
blueusername