If a notification was updated while one of its widgets was focused,
it would lose the grab when that widget was destroyed. Fix that by
moving the focus to a safe place before destroying the old widgets.
https://bugzilla.gnome.org/show_bug.cgi?id=643687
We want to minimize focus stealing from the user. If a non-urgent notification
comes in while the user is interacting with the tray, we add it to the tray
and only show it after the user is done interacting with the tray. If an
urgent notification comes in while the user is interacting with the tray,
we hide the tray and show the urgent notification.
https://bugzilla.gnome.org/show_bug.cgi?id=636838
We want to allow the user to scroll through all notifications from
source by using a single scrollbar. We suppress the individual
scrollbars inside the notifications.
As one exception, we keep the original scrollbar for chat notifications
because it has a distinct look, ending above the text entry box.
https://bugzilla.gnome.org/show_bug.cgi?id=611611
Change the way menu items allocate their contents to take text
direction into account, so they're fully reversed in RTL locales,
and St.Align.START / END are respected.
https://bugzilla.gnome.org/show_bug.cgi?id=645524
While we have menu for an app icon open, we want to show the prelight
for the item instead of removing the prelight when the user mouses
away from the item and into the menu, and if there's a tooltip
(like for the dash), we want to show the tooltip immediately when
the menu is popped up.
https://bugzilla.gnome.org/show_bug.cgi?id=642871
It can be useful to avoid sending enter/leave events to the source actor
of a menu: this would be the case when the source actor isn't a menu item
that should participate in menu navigation but rather is some object
(like an app icon) that we want to indicate corresponds to the menu.
https://bugzilla.gnome.org/show_bug.cgi?id=642871
Inside the Shell, all the UI (including chrome, the overview, and
the actual windows) is not a child of the stage but of a special
ClutterGroup, which is cloned inside the magnifier.
Add function for setting this special actor so that actors added by
St are visible in the magnifier. Nothing yet uses this, but the
tooltip will soon.
https://bugzilla.gnome.org/show_bug.cgi?id=635100
We need to update WorkspacesView._ZoomOut before calling
_updateWorkspacesGeometry() in show(), as otherwise the old
value is kept. This was a problem if we previously left the
overview zoomed out.
The applications have to have a way of keeping resident notifications
updated without unnecessarily notifying the user with the information
the user is already seeing in the application window.
https://bugzilla.gnome.org/show_bug.cgi?id=630847
Make summary sources look more clickable and highlight them when selected.
Highlighting the fully expanded summary source when selected matches the
highlighting in the top bar items and teaches the user that any part of the
expanded summary source can be clicked.
Based on the initial patches by Florian Müllner and Jonathan Strander.
https://bugzilla.gnome.org/show_bug.cgi?id=644788
The lightbox will be sized to the size of its parent container,
so we need to make the parent container reliably the size of the
stage, instead of letting it be auto-sized to the size of its contents.
https://bugzilla.gnome.org/show_bug.cgi?id=644889
Returning true from the event handler seems to avoid getting
duplicate return events; these duplicate events likely are
getting generated by IBus.
https://bugzilla.gnome.org/show_bug.cgi?id=644509
Currently activating a window on a different workspace requires very
long drag distances, which is very inconvenient to use.
Fix that by allowing switching workspaces using the thumbnails which is
consistent with window and launcher dnd and much easier to use.
https://bugzilla.gnome.org/show_bug.cgi?id=643945
Don't do an individual hover fixup for every window overlay, instead
just use the new global.sync_hover() to fix up hovers once we have
finished showing the overview.
Based on a patch from Adel Gadllah <adel.gadllah@gmail.com>
https://bugzilla.gnome.org/show_bug.cgi?id=638613
Main._nWorkspacesChanged was racing with Main._checkWorkspaces.
If _checkWorkspaces won the race, _workspaces was uninitialized.
Because of this, _checkWorkspaces only noticed workspaces
with windows on them, leading it to believe the last workspace wasn't
empty, and added a new, empty workspace.
https://bugzilla.gnome.org/show_bug.cgi?id=645343
In the case where the original actor is destroyed, we don't
show a snap back animation, so we need to destroy drag actor
as we would do in _onAnimationComplete.
https://bugzilla.gnome.org/show_bug.cgi?id=640781
If you have XFixes 5 (and corresponding xserver support) then we
add barriers on the panel and in the message tray corner so that
its easy to reach the corners even when there are monitors to the
sides of the primary monitor.
https://bugzilla.gnome.org/show_bug.cgi?id=622655
If the user clicks a trayicon in the overview, drop out of the
overview before passing the click on to the icon. (We have to actually
wait for the overview animation to complete, in case the icon wants to
get a pointer grab, which it would not be able to do with the overview
active.)
https://bugzilla.gnome.org/show_bug.cgi?id=641853
If the pointer moves on or off the stage while another process has a
grab, we will lose track of it. One example of this is that if you use
a popup menu from a message tray trayicon, the tray will stay up after
the menu goes away, because the shell never saw the pointer leave it.
Add a new method shell_global_sync_pointer() that causes clutter to
recheck what actor is under the pointer and generate leave/enter
events if appropriate.
Of course, we can't actually tell for sure when another process has a
grab, so we need a heuristic of when to call this. Currently we call
it from Chrome._windowsRestacked(), which is not really the right
thing at all, but does fix the menu-from-trayicon case...
https://bugzilla.gnome.org/show_bug.cgi?id=630842
If the user clicks on the title of a trayicon's SummaryItem, forward
that click to the trayicon. Also adjust
gnome_shell_plugin_xevent_filter() so that if the trayicon takes a
grab as a result of this, we don't hide the message tray.
https://bugzilla.gnome.org/show_bug.cgi?id=630842
If you run a command from Alt+F2 that tries to get a server grab (eg,
xmag), it will fail if it starts up before the run dialog is finished
hiding.
Additionally, the run dialog currently stays focused while it is
fading out, potentially stealing keystrokes (or causing the user to
accidentally launch two copies of a program).
Change ModalDialog.close() to call popModal() immediately
Add a ModalDialog.popModal method, and call that before running the
RunDialog command. If the command succeeds, close the dialog as
before. If it fails, call ModalDialog.pushModal() to put things back
to normal before displaying the error.
https://bugzilla.gnome.org/show_bug.cgi?id=644857
When right-clicking on an AppWellIcon, the icon will become focused,
which (presumably via style-changed) invalidates its current
allocation, causing "icon.y" to return 0 until it has been
reallocated, messing up our idea of where in the AppDisplay the icon
is. Work around this by calling get_allocation_box() instead.
https://bugzilla.gnome.org/show_bug.cgi?id=645162
If there is a monitor to the right it is very easy to overshot the
expanding thumbnails and enter the next monitor. So, in that case
we just always show it.
https://bugzilla.gnome.org/show_bug.cgi?id=641877
Specify x-fill and y-fill true for the bin that contains the status
icon so the status icon will always be sized to our specified icon
size (24x24). This prevents pathological behavior for legacy status
icons embedded in the tray where an initial allocation at 1x1 before
they had content would "stick", and the icon would permanently
end up 1x1.
https://bugzilla.gnome.org/show_bug.cgi?id=634820
Right now, the user status menu always contains actions to logout
and lock the screen, and the user switching action only depends on
the technical availability of the functionality.
All those items should honor the lockdown settings defined in
org.gnome.desktop.lockdown.
https://bugzilla.gnome.org/show_bug.cgi?id=645335
To avoid having hot corners that accidentally trigger when e.g. trying
to hit the panel on the primary monitor we add hot corners only to
monitors that are "naturally" top left (top right for RTL).
For instance, we'd like a hot corner here:
corner -> +-------------
| |
+---------+ |
|=========| |
| | |
| | |
| | |
+---------+------------+
But not here:
unexpected hot corner
↓
+---------+-------+
|=========| |
| | |
| +-------+
+---------+
https://bugzilla.gnome.org/show_bug.cgi?id=645116
The fade effect when switching tabs should only be applied when
switching from a previously selected tab, not when selecting the
initial one - otherwise, the window previews are faded in the first
time the overview is shown.
https://bugzilla.gnome.org/show_bug.cgi?id=644389
When we were knocking off workspace height to fix the ratio problems, we
weren't adding spacing in between workspaces, so they smooshed up against
each other whenever we took height off, causing them to be visible.
Different methos are being used to launch the control-center panels of each
status icon. Standarize on Shell.AppSystem.
This also fixes the network icon using a non-existant Util.spawnDesktop()
method.
Bug #645091
Commit fcfd17e was overzealous when simplifying the previous spinner
animation, as a result the spinner now stays around when switching
to another application while the animation is ongoing.
If a workspace becomes empty due to a window changing to/from the
primary monitor, but not changing its original workspace then we
were not noticing this. This can happen for instance if you drag
a thumbnail window to a non-primary window.
https://bugzilla.gnome.org/show_bug.cgi?id=609258
We clip the entire WorkspacesDisplay to its allocation to avoid things
like the WorkspaceThumbnails sticking out of the primary monitor into
another monitor.
https://bugzilla.gnome.org/show_bug.cgi?id=609258
We used to do this only on automatic workspace switch, but that
doesn't work for the multiple monitors case where we want to reserve
space on the extra monitors.
https://bugzilla.gnome.org/show_bug.cgi?id=609258
We create a Workspace with a null metaWorkspace for each
non-primary monitor, showing the windows on these monitors.
These are saved in WorkspaceView.extraWorkspaces.
https://bugzilla.gnome.org/show_bug.cgi?id=609258
This means a bunch of windows will not be visible at all in the overview.
Those will be added back with per-screen workspaces on the non-primary
monitors.
https://bugzilla.gnome.org/show_bug.cgi?id=609258
Notification was connecting to button-release-event to decide when to
be dismissed, which caused problems with widgets inside the
notification that reacted to button-press-event but not
button-release-event. Fix this by wrapping the Notification's table in
an StButton and connecting to 'click'.
https://bugzilla.gnome.org/show_bug.cgi?id=642978
In commit 9bd22dc0, I introduced an API to load an arbitrary
.desktop file, not necessarily from the menu path. It turns
out this function was broken because it created ShellApp instances
that were *different* from ones that were cached normally.
As far as I can tell, we didn't initially use it. Then later
Util.spawnDesktop was created which used this function.
Remove this broken function and all callers; if we're loading
.desktop files from *outside* the menu path, we can look at
readding.
This patch also kills off Util.spawnDesktop in favor of callers
talking to ShellAppSystem directly, now that the latter reports
errors.
https://bugzilla.gnome.org/show_bug.cgi?id=644402
Move the "system notification error" handling out of
util.js, and add it to ShellGlobal so we can start
calling it from across the codebase better (including
C).
https://bugzilla.gnome.org/show_bug.cgi?id=644402
Simplify the accounting of which windows we should dim by checking
the current state of windows rather than trying to track changes,
and by keeping a list of dimmed windows rather than a list of windows
with a dimmed parent. Remove windows from the list of dimmed windows
when they are destroyed.
This should fix problems where destroyed windows could end up in
the list of dimmed windows.
https://bugzilla.gnome.org/show_bug.cgi?id=644167
Adds an implementation of nm-applet in javascript. Uses the new
introspection from NetworkManager, and temporarily requires
nm-applet to be running for the secret service.
Features a renewed interface, with each device controllable through
a switch, which if toggled off disconnects, and if toggled on
connects to the most recently used valid connection. More esoteric
features like creation of ad-hoc networks have been moved to the
control center panel.
https://bugzilla.gnome.org/show_bug.cgi?id=621707
Inside the inner loop, use the inner iterator, not that of the outer
loop. At the same time, refactor the code to rely less on private
properties appended to foreign objects.
https://bugzilla.gnome.org/show_bug.cgi?id=644858
The polkit authentication dialog contains logic for
falling back to dispalying a user's username if that
user has no real name.
This logic is no longer needed because gdmuser does it
internally now.
https://bugzilla.gnome.org/show_bug.cgi?id=644765
Commit 0207f1f29b landed a new
way of zooming, but was causing all sorts of window positioning
weirdness because the positions were supposed to be working against
a proportional workspace.
https://bugzilla.gnome.org/show_bug.cgi?id=644542
Do a basic job of converting font sizes from pixels to points, so they
will scale will the global GNOME scale factor. Some other sizes that are
clearly related to the font sizes are changed to ems, but no comprehensive
attempt is made to get rid of px units.
https://bugzilla.gnome.org/show_bug.cgi?id=636868
During application startup, we used to display a rotating spinner
which also moved from left to right, revealing the application title.
The result looks rather busy, so remove the horizontal movement.
https://bugzilla.gnome.org/show_bug.cgi?id=640782
Update the Source title when an contact's alias changes, and
also also add a minor meta message like the current timestamps.
Updating the alias of a 'presenced' contact will overwrite the
current title, and it will also not update the summary item title
right now due to limitations of the message tray.
https://bugzilla.gnome.org/show_bug.cgi?id=642793
WorkspacesDisplay removes its dragMonitor in _dragEnd, but
this was never called in when a xdnd drag ended causing
dragMonitors to stack up and handling events multiple times.
Fix that by making sure that _dragEnd is called when xdnd ends.
https://bugzilla.gnome.org/show_bug.cgi?id=644642
Monkey-patch Date.prototype.toLocaleFormat() with a version that uses
g_date_time_format() since the Spidermonkey built-in can't handle
format strings with Unicode characters.
https://bugzilla.gnome.org/show_bug.cgi?id=643350
If we don't update every second, we may show the wrong time for up to
a minute on 1. resume; or 2. when changing the time; or 3. when
changing the timezone. This is both annoying and and leads to people
thinking that the tool for changing the time / timezone is broken.
https://bugzilla.gnome.org/show_bug.cgi?id=635840
Signed-off-by: David Zeuthen <davidz@redhat.com>
The mockups are here
https://live.gnome.org/GnomeShell/Design/Whiteboards/AuthorizationDialog
Detailed changes
- Don't use an icon for root
- For root, show Administrator in red
- Nuke icons for info and error messages
- Make error messages yellow
- Use 10pt size for error and message labels, not 12px
- Don't make the dialog change size when (single-line) error/info
messages appear
- Spacing fixes
- Show "Sorry, that didn't work. Please try again" if authentication fails
- Don't cancel the PolkitAgentSession if the session has already completed
https://bugzilla.gnome.org/show_bug.cgi?id=644737
Signed-off-by: David Zeuthen <davidz@redhat.com>
In the mockups the slider does not have an uniform color but uses
one color (shade of blue) to indicate the current value and one
to indicate "the rest" (shade of grey).
So adjust the slider to look like that to be closer to the look
in the mockups and thus to the design.
https://bugzilla.gnome.org/show_bug.cgi?id=644600
GJS complains when a NaN is passed in place of an integer, and
parseInt returns that from non numeric string. Pass a sentinel
that cancels the operation in that case.
https://bugzilla.gnome.org/show_bug.cgi?id=642978
Around 2.91.90, gnome-session-save was renamed to gnome-session-quit.
This commit restores compatibility with the older gnome-session, for
those testing under GNOME 2.32 or below, by calling the DBus methods
directly.
https://bugzilla.gnome.org/show_bug.cgi?id=644591
Add the idea of an 'id' for a tab, and add a public switchTab method
so you can switch to 'applications' or 'windows'. This will be useful
for performance tests that test tab switching performance.
https://bugzilla.gnome.org/show_bug.cgi?id=644266
* Run gnome-shell-perf-helper during performance tests
* Use MUTTER_WM_CLASS_FILTER to omit all other windows
* Add new Scripting methods: createTestWindow,
waitTestWindows, destroyTestWindows
* Create a single 640x480 test window for testing overview
animation performance.
https://bugzilla.gnome.org/show_bug.cgi?id=644265
Don't enter the overview at startup, or when we we remove the
last window on the first workspace, but only when we remove a
workspace and there are windows on the other workspaces.
https://bugzilla.gnome.org/show_bug.cgi?id=644541
When new messages come in we want to scroll down so that the user
sees the incoming messages. The current implementation does not work
because it relies on a synchronous allocation hack which does not work
for unmapped notifications.
Fix that by connecting to adjustment::changed and scroll whenever the
adjustment changes which equals "new messages", "new timestamp" or
"presense change", but don't interference with the user's scroll actions
i.e when the user scrolls back to read something don't scroll to the bottom.
https://bugzilla.gnome.org/show_bug.cgi?id=614977
Remove the hack from Notification.scrollTo because it is unreliable,
the caller should make sure to call scrollTo when it will actually
have the desired effect.
https://bugzilla.gnome.org/show_bug.cgi?id=614977
If we're dragging a window around and we need to reposition the windows,
due to e.g. the sliding in of the thumbnails or some other reason, then we
need to consider the original position of the dragged window, rather than
the currend drag position. Otherwise we will unnecessarily rearrange the
other windows for instance on snap-back if you moved the dragged window
past some other window.
https://bugzilla.gnome.org/show_bug.cgi?id=643786
We currently show the workspace in the overview in a rectangle
with the same aspect ratio as the screen. Originally this was
probably done since it showed the desktop, but we don't do this
anymore, and the positioning of the windows in the overview is
strictly a grid, so its not in any way related to monitor geometry.
Additionally, in the multihead case the screen aspect ratio is
very different from the overview monitor geometry, so a lot of
space is lost.
So, instead we just fill the entire inner rectangle of the overview
with the workspace. However, the way the zoom into and out of the
workspace right now is by scaling the workspace so that it covers
the entire monitor. This cannot really work anymore when the workspace
is a different aspect ratio. Furthermore the coordinates of the
window clone actors are of two very different types in the "original
window" case and the "window in a slot case". One is screen relative,
the other is workspace relative. This makes it very hard to compute
the cost of window motion distance in computeWindowMotion.
In order to handle this we change the way workspace actor positioning
and scaling work. All workspace window clone actors are stored in
true screen coordingates, both the original window positions and the
in-a-slot ones. Global scaling of the workspace is never done, we
just reposition everything in both the initial zoom and when the
controls appear from the side.
There is one issue in the initial and final animations, which is that
the clip region we normally have for the workspacesView will limit the
animation of the clones to/from the original positions, so we disable
the clip region during these animations.
https://bugzilla.gnome.org/show_bug.cgi?id=643786
When closing a workspace due to the last window on that workspace
closing, switch to the overview and show the always empty workspace
rather then just going to the adjacent workspace.
Based on a patch from Adel Gadllah <adel.gadllah@gmail.com>.
https://bugzilla.gnome.org/show_bug.cgi?id=642188
During a drag-and-drop, our pointer grab keeps enter/leave events from
being delivered. That means that after the DND ends, whatever actor is
under the pointer won't have received the enter event it should have,
and any state or hover effect dependent on that won't work right.
By paying attention to the actors we leave and enter we can figure out
what widgets we need to call st_widget_sync_hover() on after the drag.
https://bugzilla.gnome.org/show_bug.cgi?id=640974
Showing the right click menu causes errors when ungrabbing focus in this case.
It will soon be impossible to get to the right click menu anyway when a new
notification is showing, because we are never going to show the summary
and the new notification at the same time.
Add Ctrl-Alt-Tab support to ViewTab, and fix the Applications pane to
scroll to track the keyboard focus.
The Windows pane can be switched to, but navigation within the pane is
not yet implemented.
https://bugzilla.gnome.org/show_bug.cgi?id=618887
Fix the "panel" icon to be symbolic. Make the overview parts only show
up when in the overview, and the non-overview parts (eg, the Desktop
window, if there is one) only show up when not in the overview. Sort
the different items consistently with their locations on the screen.
https://bugzilla.gnome.org/show_bug.cgi?id=618887
Unset this._expandedSummaryItem if it is the summary item that is being removed.
This avoids "this._sourceTitle.clutter_text is null" error.
Destroy the summary item actor only after calling _unsetClickedSummaryItem()
that disconnects from one of its signals.
https://bugzilla.gnome.org/show_bug.cgi?id=644043
PopupMenuManager was pretending that it knew nothing about the menu's
sourceActors, while also trying to handle keynav between them. This
was a big mess, and resulted in bugs in navigation between panel menus
and the Activities button, and it totally gets in the way when trying
to add keynav to the dash (whose menu sources are arranged vertically
rather than horizontally).
Fix this up by moving the panel-specific parts to PanelMenuButton
instead.
https://bugzilla.gnome.org/show_bug.cgi?id=641253
It already doesn't work right, because the PanelMenuButton code
assumes that Left and Right won't be used as part of keynav within a
menu. And the gnome-panel calendar isn't keyboard accessible either,
so this isn't a regression. To be fixed later.
https://bugzilla.gnome.org/show_bug.cgi?id=641253
To deal with different CSS in RTL locales, we used to manually add
an :rtl pseudo class to some actors. With automatically assigned
:ltr/:rtl selectors this is no longer necessary.
https://bugzilla.gnome.org/show_bug.cgi?id=643835
CtrlAltTabPopup was using a St.BoxLayout and relied on anchor_gravity
center for positioning. This does not guarantee correct pixel alignment,
so use a St.GenericContainer instead and do the positioning similar to
that of the appSwitcher.
https://bugzilla.gnome.org/show_bug.cgi?id=643820
Make GSettings support optional, refactor text entry handling,
fix some off-by-one bugs in the management itself, use Params
for parsing, fix other typos and bugs.
https://bugzilla.gnome.org/show_bug.cgi?id=642793
Currently the icon texture is only updated on style changes when
the icon size is set from CSS and differs from the previously used
icon size.
As the style change may have been triggered by an icon theme change,
textures that are created for themed icons should always be recreated;
given that this is the case for most uses (with the exception of
file thumbnails), recreate the icon texture unconditionally to avoid
complexity.
https://bugzilla.gnome.org/show_bug.cgi?id=643738
Commit b1654af406 moved the panel positioning (and thus that of its
corners) before the panel startup animation. As the panel corners now
are mapped while the panel animation is active, the initial style-changed
signal which triggers another repositioning is received after the
initial layout and the corners end up at wrong positions.
To fix, animate the corner positions as well during the startup
animation - if anyone could actually see the animation, the corners
should animate with the panel anyway ...
https://bugzilla.gnome.org/show_bug.cgi?id=643804
Mutter really expects this, as this is how app-specified struts
happen. For instance, if the primary display is beside a taller
screen and is not positioned at the top, then we extend the struts
for the panel with the size of the unused area above the primary
monitor.
https://bugzilla.gnome.org/show_bug.cgi?id=642881
Right now we require a strut to be as wide as the full screen to
create a TOP strut. This means that in a multi-monitor scenario (at
least if the primary monitor is leftmost) we will make the panel strut
be Meta.Side.LEFT due to the random side picking for corner objects.
This changes the width/height comparison to the primary monitor rather
than the screen to get this right. Also adds some docs about how
struts work in a multi-monitor situation.
https://bugzilla.gnome.org/show_bug.cgi?id=642881
We need to do the initial relayout before we start up the startup
animation, and the startup animation can't hardcode the position
of the panel to zero.
https://bugzilla.gnome.org/show_bug.cgi?id=642881
1. disconnect destroy signals in popModal (actors can have great lifetime and then pushed again)
2. incorrect pushModal/popModal pair in overview
https://bugzilla.gnome.org/show_bug.cgi?id=64078
Increase the icon size of search results for consistency with the
application view. To account for the larger icons, only display
a single row of results per section.
https://bugzilla.gnome.org/show_bug.cgi?id=643632
Search results' meta info currently is expected to have an 'icon'
property holding a Clutter.Texture of a fixed icon size. This
property is used to implement the createIcon() function of BaseIcon,
which is used to actually display the result, ignoring the size
parameter due to the fixed icon size.
Given that all available search providers create this property for
the desired icon size using a function with the same signature, it
appears logical to replace the fixed-sized 'icon' property with
such a function, so that the icon size will be defined by the display
actor rather than the search system.
https://bugzilla.gnome.org/show_bug.cgi?id=643632
Currently section headers in the search view are reactive and run
the corresponding provider's expandSearch() function when clicked,
which should launch an external program displaying all search
results for the section. Unfortunately it is only implemented for
the "Settings" provider, and does not work properly (as it ignores
the search and just launches System Settings).
Also current mockups deemphasize the section header, making the
feature pretty much indiscoverable (except by accident when
interfering with swipe-scrolling).
In conclusion, the feature does not make much sense in its current
form, so remove it.
https://bugzilla.gnome.org/show_bug.cgi?id=643632
Currently the total number of matches for a particular section is
displayed on the left of the section's header. While it made sense
in the old layout where it was close to the section's title, it is
now rather disconnected, and has been removed in current mockups.
https://bugzilla.gnome.org/show_bug.cgi?id=643632
When UPower has not yet collected enough data to calculate a time
to discharge, it will report 0 minutes. Show "Estimating..." in
that case instead.
https://bugzilla.gnome.org/show_bug.cgi?id=642753
Currently we remove a workspace when last window on that workspace closes,
which turned out to be too agressive as some apps open either a splashscreen
or a "I have crashed last time" dialog at startup.
Try to detect such apps and give them a chance to open their "real window(s)"
on the workspace.
https://bugzilla.gnome.org/show_bug.cgi?id=642188
