This patch fixes the "apps vanish from alt-TAB bug".
If a "package system" rips away and possibly replaces .desktop files
at some random time, we have historically used inotify to detect this
and reread state (in a racy way, but...). In GNOME 2, this was
generally not too problematic because the menu widget was totally
separate from the list of windows - and the data they operate on was
disjoint as well.
In GNOME 3 we unify these, and this creates architectural problems
because the windows are tied to the app.
What this patch tries to do is, when rereading the application state,
if we have a running application, we keep that app around instead of
making a new instance. This ensures we preserve any state such as the
set of open windows.
This requires moving the running state into ShellAppSystem. Adjust
callers as necessary, and while we're at it drop the unused "contexts"
stuff.
This is just a somewhat quick band-aid; a REAL fix would require us
having low-level control over application installation. As long as
we're on top of random broken tar+wget wrappers, it will be gross.
A slight future improvement to this patch would add an explicit
"merge" between the old and new data. I think probably we always keep
around the ShellApp corresponding to a given ID, but replace its
GMenuTreeEntry.
https://bugzilla.gnome.org/show_bug.cgi?id=657990
As extensions are now expected to provide a "disable" function,
they need to remove search providers they added. Implement the
removal functionality and add a public removeSearchProvider()
method.
https://bugzilla.gnome.org/show_bug.cgi?id=657548
We keep track of presence changes by connecting to the
Tp.AccountManager:most-available-presence-changed signal.
However, if multiple accounts are in use, telepathy may
lie to us and emit the signal even when the most available
presence is unchanged.
Work around this by keeping track of the current presence
ourselves.
https://bugzilla.gnome.org/show_bug.cgi?id=657703
Some search providers may want to change their results, or may not
want to block on an external service to get their results (DBus, etc.)
Set up an infrastructure to allow search providers to add their search
results at a later time.
Based on a patch by Jasper St. Pierre and Seif Lotfy.
https://bugzilla.gnome.org/show_bug.cgi?id=655220
This adds contacts search to shell, powered by libfolks.
Changes:
- Add Folks and Gee to the build system
- ShellContactSystem, a backend in C
- ContactDisplay, search frontend in JS
https://bugzilla.gnome.org/show_bug.cgi?id=643018
The current user status menu allow to set the session status,
which also influences the IM status when signed in with
mission-control. However, the way it is presented to the user
makes it hard to figure out how the statuses interact or that
there are two distinct status in the first place.
Therefore, use a separate control for each status, and update the
overall look to match gnome-contacts.
https://bugzilla.gnome.org/show_bug.cgi?id=652837
Given that our menus contain at most two columns, all switch widgets
in menus end up in the last columns, and thus aligned with the right
menu edge.
However, the updated user status menu will contain a section which
ignores the menu's column layout, so the switch might end up in
the middle of the menu if the overall width is determined by said
section.
At least for now, we always want the switch to align with the end,
so just expand switch menu items rather than adding an option.
https://bugzilla.gnome.org/show_bug.cgi?id=652837
Introduce a new menu widget, which displays the active item from
a set of options, and pops up a child menu to allow changing the
active item when activated.
https://bugzilla.gnome.org/show_bug.cgi?id=652837
Allow opening a popup menu from another menu. While the child menu
is open, events on the parent menu are blocked. The parent menu
is kept open when the child menu is closed; the child menu on the
other hand is closed with the parent, e.g. when the focus moves
to another toplevel menu.
This feature will be used to implement combo box menu items.
https://bugzilla.gnome.org/show_bug.cgi?id=652837
We haven't actually been calling the top-right menu "status menu" for
quite some time, so use the upcoming code changes as an excuse for
renaming it to "user menu".
Doing this rather than overdrawing a black rectangle saves us
(pixels in screen) * 8 bytes of memory bandwidth for every frame we draw going
into the overview.
It also allows us to dim the background on non-primary monitors making the
overall overview appearance consistent across all monitors.
https://bugzilla.gnome.org/show_bug.cgi?id=656433
Instances of this class share a single CoglTexture behind the scenes which
allows us to show the background with different rendering options without
duplicating the texture data.
https://bugzilla.gnome.org/show_bug.cgi?id=656433
This commit adds GDM session support.
It provides a user list that talks to GDM,
handles authentication via PAM, etc.
It doesn't currently support fingerprint readers
and smartcards.
https://bugzilla.gnome.org/show_bug.cgi?id=657082
The shell has a number of things that are only relevant for
logged in users (e.g. calendar events, telepathy integration, a
user menu, etc).
This commit moves those user session specific bits into their
own functions in preparation for making the shell code ready
for use at login time.
https://bugzilla.gnome.org/show_bug.cgi?id=657082
Right now the panel code makes the left corner sync up with the
activities button and the right corner sync up with the user menu.
This is fine as long as we have an activities button and a user menu.
The login screen won't have those things, though.
This commit changes the panel corner code to try to figure out which
interface element is the most appropriate to sync up with based on
its position in the panel.
https://bugzilla.gnome.org/show_bug.cgi?id=657082
Images are part of the notification spec, so we should support them.
Marina Zhurakhinskaya provided some code for getting the layout right
for this patch.
https://bugzilla.gnome.org/show_bug.cgi?id=621009
Have LayoutManager automatically deal with sizing and positioning
boxes for the panel and messageTray relative to the monitors.
Also, now that LayoutManager knows exactly where and how tall the
panel and tray are, have it manage the pointer barriers as well.
https://bugzilla.gnome.org/show_bug.cgi?id=612662
A modal dialog in the shell blocks anything but that dialog from
receiving user input. Applications within the session and other
parts of UI are rendered non-reactive.
When GDM gets changed to use the shell for its greeter, the user
list will be presented as a shell dialog. That dialog shouldn't
block access to the panel menus, etc.
This commit adds a shellReactive property that makes the ModalDialog
class continue to block access to applications, but allow the user
to interact with the shell itself.
https://bugzilla.gnome.org/show_bug.cgi?id=657082
The control-center contains user-pertinent settings
panels. These panels don't make sense to show outside
of a user's session, so hide them for session types other
than SessionType.USER.
https://bugzilla.gnome.org/show_bug.cgi?id=657082
We're not going to want an overview at the login screen,
but a lot of code in the shell depends on the overview
existing.
This commit adds a new isDummy constructor property to
allow creating the overview as a non-functional, stub object
that doesn't do anything visible.
https://bugzilla.gnome.org/show_bug.cgi?id=657082
The dash object is currently exposed as a public object.
It's only used outside of the overview for the dash object's
iconSize property though.
This commit makes the dash object private and proxies the dash
iconSize property to the overview.
https://bugzilla.gnome.org/show_bug.cgi?id=657082
Right now, when a user clicks on the panel clock, a menu pops up with a
calendar and a list of events from the user's schedule. The list of
events only makes sense from within a user's session, however.
As part of the prep work for making the shell a platform for the login
screen, this commit makes the events list optional.
https://bugzilla.gnome.org/show_bug.cgi?id=657082
The theme currently hard codes the minimum size of the calendar
menu to make sure there's a designated area for events
(even if there isn't anything currently scheduled).
A side-effect of the hard coded minimum width is that
if the events area is hidden, the menu ends up much
bigger than the calendar. We don't currently ever hide
the events area, but we will in the future.
This commit moves the min-width restriction from the menu
specifically to the events area.
https://bugzilla.gnome.org/show_bug.cgi?id=657082
The chrome layer contains the user interface elements (e.g.,
the panel) that disappear when fullscreen windows get displayed.
Panel menus are currently put in the chrome layer, but don't need
to be, since they are only displayed when the user is interacting
with the shell and not a fullscreen application.
Putting panel menus in the chrome layer does mean they will get
stacked below shell interface elements that aren't in the chrome layer,
though.
This commit changes panel menus to be on the same layer as most other
shell elements, so they get properly stacked above those elements.
https://bugzilla.gnome.org/show_bug.cgi?id=657082
Right now, if buttons get set on a dialog after it is mapped,
they just pop in instantly.
We shouldn't have any harsh transitions like that, though.
This commit changes the buttons to quickly fade in, instead.
https://bugzilla.gnome.org/show_bug.cgi?id=657082
All the system status menus in the panel offer a
menu item to jump to a relevant part of the
control-center.
This means each status icon has the same, or nearly the
same bit of code to:
- Add a new "action" menu item and listen for its activation.
- Hide the overview if it's showing when the menu item is activated
- Find the relevant control-center panel from its desktop file
- Launch the control-center to the relevant panel
This commit consolidates all those details in a new method,
addSettingsAction. This refactoring reduces code duplication and
slight inconsistencies in the code resulting from that duplication.
It will also make it easier in subsequent commits to hide settings menu
items when the shell is used in the login screen.
https://bugzilla.gnome.org/show_bug.cgi?id=657082
A separator only makes sense if there are items on both
sides of it. There is quite a lot of code written
throughout the shell that manages the process of showing
and hiding separators as the items around those separators
change.
This commit drops all that code in favor of changes to the menu
implementation to dynamically hide or show separators as
appropriate, so the callers don't have to deal with it.
https://bugzilla.gnome.org/show_bug.cgi?id=657082
Wireless and 3g dialog code has moved to gnome-control-center, so
we can stop calling out to nm-applet. Also, we can now enable the
notifications provided by the shell and kill a bit of code about
auth that is not actually needed.
https://bugzilla.gnome.org/show_bug.cgi?id=650244
Using the new ShellNetworkAgent, show a system modal dialog
(similar to the PolicyKit one) when NetworkManager needs secrets
for connecting to wireless.
https://bugzilla.gnome.org/show_bug.cgi?id=650244
The order of indicators depends on the order of calls to
Panel.addToStatusArea. To have it consistent across enabling and
disabling of extensions, we need to place the core ones first.
https://bugzilla.gnome.org/show_bug.cgi?id=653205
This way all standard indicators have a shell implementation
provided, which prevents issues with extensions enabling/disabling
(in particular with xrandr-indicator)
https://bugzilla.gnome.org/show_bug.cgi?id=653205
Extensions often want to add items to the system status area, so it
is useful to add a convenience API for it. Also, we now allow
for cleaner destruction of panel objects, by just calling destroy()
on it.
Based on a patch by Jasper St. Pierre.
https://bugzilla.gnome.org/show_bug.cgi?id=653205
The "id" variable was being sporadically reset to null, and as far as
Florian and I could determine, this is actually a Spidermonkey bug.
The issue has something to do with:
1) use of "let" for the variable
2) Nesting a dynamic closure inside of a for() loop
Work around it here for now - I tried to create a minimized test case
to hand to the Spidermonkey developers, but failed. A big part of
the problem is it's only sporadically reproducible.
Direction containers group all contiguous messages in the same direction into
their own parent container, allowing for smarter styling of similar messages.
https://bugzilla.gnome.org/show_bug.cgi?id=640271
This adds a new DBus method: InstallExtensionRemote(uuid : s, url : s)
Pass it the UUID of an extension and the URL of a manifest file: the same as a
metadata.json, but with a special key, '__installer', which is an HTTP location
that points to an zip file containing the extension. The Shell will download
and use it to install the extension. In the future, the manifest file may be
used to automatically detect and install updates.
https://bugzilla.gnome.org/show_bug.cgi?id=654770
The two similar keys were hard to manipulate to have specific effects, so just
remove one. Now there is an *explicit* whitelist: all extensions must be in the
'enabled-extensions' for them to be loaded.
https://bugzilla.gnome.org/show_bug.cgi?id=654770
It's generally more useful to see when a person sent a message instead of when
we received it. Also, a recent change in Telepathy made the received timestamp
be 0 for messages we send.
https://bugzilla.gnome.org/show_bug.cgi?id=640271
Adds methods to shell_global to allow taking screenshots
save the result into a specified png image.
It exposes three methods via shellDBus applications like
gnome-screenshot:
*) Screenshot (screenshots the whole screen)
*) ScreenshotWindow (screenshots the focused window)
*) ScreenshotArea (screenshots a specific area)
https://bugzilla.gnome.org/show_bug.cgi?id=652952
It is not possible to connect to hidden access points without
knowing the SSID, and it should be done using the control center
panel and the appropriate dialog. At the same time, this should
fix some warnings from libnm-glib and dbus-glib.
https://bugzilla.gnome.org/show_bug.cgi?id=646454
The shell should only notify in case no other client handles the message.
Empathy will ack the message if focused, so we don't want to step on its
toes.
Use the existing setting
org.gnome.desktop.default-applications.office.calendar.exec
as calendar application instead of the hard-coded evolution. Evolution
is still the fallback if that setting is cleared (it defaults to
evolution).
https://bugzilla.gnome.org/show_bug.cgi?id=651190
Since almost all of the callers of shell_app_activate were using the
default workspace (by passing -1), remove that parameter.
Add a new shell_app_activate_full() API which takes a workspace as
well as a timestamp; previously we might have been ignoring event
timestamps from elsewhere.
https://bugzilla.gnome.org/show_bug.cgi?id=648149
This dramatically thins down and sanitizes the application code.
The ShellAppSystem changes in a number of ways:
* Preferences are special cased more explicitly; they aren't apps,
they're shortcuts for an app), and we don't have many of them, so
don't need e.g. the optimizations in ShellAppSystem for searching.
* get_app() changes to lookup_app() and returns null if an app isn't
found. The semantics where it tried to find the .desktop file
if we didn't know about it were just broken; I am pretty sure no
caller needs this, and if they do we'll fix them.
* ShellAppSystem maintains two indexes on apps (by desktop file id
and by GMenuTreeEntry), but is no longer in the business of
dealing with GMenuTree as far as hierarchy and categories go. That
is moved up into js/ui/appDisplay.js. Actually, it flattens both
apps and settings.
Also, ShellWindowTracker is now the sole reference-owner for
window-backed apps. We still do the weird "window:0x1234beef" id
for these apps, but a reference is not stored in ShellAppSystem.
The js/ui/appDisplay.js code is rewritten, and sucks a lot less.
Variable names are clearer:
_apps -> _appIcons
_filterApp -> _visibleApps
_filters -> _categoryBox
Similarly for function names. We no longer call (for every app) a
recursive lookup in GMenuTree to see if it's in a particular section
on every category switch; it's all cached.
NOTE - this intentionally reverts the incremental loading code from
commit 7813c5b93f. It's fast enough
here without that.
https://bugzilla.gnome.org/show_bug.cgi?id=648149
Add a helper function (mostly copied from gtkcalendar.c) for getting
the first week day for the current locale, using nl_langinfo if
available and falling back to the GTK+ gettext fallback otherwise.
Use that function in the calendar, so that the LC_TIME setting is
used if possible.
https://bugzilla.gnome.org/show_bug.cgi?id=649078
Instead, create three ripples and keep tweening them. This gives a dramatic
speedup when entering the overview, but means that we can't have the same animation
running twice. In this case, we "reset" the currently running ripple animation, but
it is hard to notice unless looking for it.
https://bugzilla.gnome.org/show_bug.cgi?id=656125
Extension developers may be confused about why their extensions aren't working:
the LookingGlass isn't a very obvious place, or even which errors are theirs.
To remedy this, save all errors per-UUID which allows them to be retrieved
later.
https://bugzilla.gnome.org/show_bug.cgi?id=654770
Add ShellVersion, designed for detecting OUT_OF_DATE extensions so they can't
be installed, as well as ApiVersion, designed for backwards-compatibility with
the SweetTooth web-app, which must support all shell versions.
https://bugzilla.gnome.org/show_bug.cgi?id=654770
GetExtensionInfo() takes a UUID and returns a JSON object with information
about that extension including their metadata, path and current state.
ListExtensions() takes no arguments and returns a JSON object mapping UUIDs
to the same information objects described above.
https://bugzilla.gnome.org/show_bug.cgi?id=654770
As an effort to prevent a string freeze to land timestamps on 3.0, we reused
translations for the calendar. Now that the string freeze is long gone, make
some proper strings.
https://bugzilla.gnome.org/show_bug.cgi?id=640271
The notification spec supports the concept of a 'default' action:
"The default action (usually invoked my clicking the notification)
should have a key named "default". The name can be anything, though
implementations are free not to display it."
Support this by invoking the 'default' action rather than a emitting
the 'clicked' signal when clicking notifications which specifie a
default action.
Also don't add an action button for the default action.
https://bugzilla.gnome.org/show_bug.cgi?id=655818