The asynchronous nature of extension loading, session loading, and more,
makes the code racy as to what is initialized first, and hard to debug.
Additionally, since gjs is single-threaded, the only code we're running
in a thread anyway is readdir, which is going to be I/O bound, so the
code here is actually likely to be faster.
Drop this in favor of some good old fashioned synchronous loading.
We currently only ensure that width and height are positive, so it
is still possible to pass in values that don't make any sense at all
(which may even result in a crash when exceeding limits imposed by
X11).
There is nothing to screenshot outside the actual screen area, so
restrict the parameters to that.
app-system: Lazily create ShellApps for apps we care about
Rather than create all ShellApps up-front, create them lazily. We really
had no reason to do this before as we were scanning GMenu to get all the
apps, but doing this can remove a need for get_all, which is slow and
memory-hungry.
We want to transition to a system in the future where we have a desktop
file cache. As we no longer differentiate categories or similar, it no
longer makes sense to have app visibility based on categories. Thus,
we no longer need to use gnome-menus to list all apps. The potential
issue here is reloading all desktop files when new files are created,
but this can be dealt with individually.
The "All Applications" view still uses gnome-menus.
Since appDisplay.js makes its own GMenu tree, it's not necessary
anymore. This does mean that searches will show apps in NoDisplay
categories, but that's an obscure enough edge case not to matter.
app-system: Don't use gmenu_tree_entry_get_desktop_app_info
It's a broken method when it comes to giving us a useful GDesktopAppInfo,
and it's hard to fix libgmenu properly, so simply recreate the app info
using the desktop file ID that libgmenu has.
We want to move away from gnome-menus eventually, so the simple
utility method isn't really worth keeping around. Reimplement it
in the one place that uses it.
search: Make the internal search interface callback-based
Long ago, the search system worked in a synchronous manner: providers
were given a query, and results were collected in a single array of
[provider, results] pairs, and then the search display was updated
from that.
We introduced an asynchronous search system when we wanted to potentially
add a Zeitgeist search provider to the Shell in 3.2. For a while, search
providers were either async or sync, which worked by storing a dummy array
in the results, and adding a method for search providers to add results
later.
Later, we removed the search system entirely and ported the remaining
search providers to simply use the API to modify the empty array, but the
remains of the synchronous search system with its silly array still
lingered.
Finally, it's time to modernize. Promises^WCallbacks are the future.
Port the one remaining in-shell search engine (app search) to the new
callback based system, and simplify the remote search system in the
process.
Florian Müllner [Fri, 1 Nov 2013 03:33:10 +0000 (04:33 +0100)]
system: Restore support for 'disable-restart-buttons'
The org.gnome.login-screen schema contains a key to disable the
power/restart buttons; our support for this fell victim to the
new combined status menu, add it back.
workspacesView: Simplify the workspacesOnlyOnPrimary implementation
Before, workspacesOnlyOnPrimary was implemented in quite a crazy manner:
* If workspacesOnlyOnPrimary was false, we'd create one WorkspacesView per
monitor, with the primary one being a bit special.
* If workspacesOnlyOnPrimary was true, we'd create one WorkspacesView, and
additional montiors would be handled inside that WorkspacesView as
"extra workspaces".
This caused numerous bugs as the two modes weren't consistently
implemented, and a lot of code was duplicated between all the modes.
Fix this by always creating WorkspaceViews, even if it only handles
one interface. We do this by having two different WorkspacesView-ish
classes: WorkspacesView handles the traditional combination of lots
of workspces, and a new ExtraWorkspaceView is in control of only one
workspace.
workspacesView: Make each WorkspacesView own its set of workspaces
Right now, the workspace update code is complex and spread across parts:
WorkspacesView takes a set of workspaces and looks like it owns them, but
WorkspacesDisplay is actually in charge of setting them up and creating
new ones for each WorkspacesView.
Change initialization and handling to move all of the creation/destruction
responsibilities to WorkspacesView.
We pass in monitorIndex into each WorkspacesView, which is a lie in the
workspacesOnlyOnPrimary case, as the primary WorkspacesView currently has
the responsibility of handling the extra workspaces on all the other
monitors. The commit will clean this up and punt the responsibility back
to WorkspacesDisplay.
overviewControls: Move translation-x to SlideLayout
Not because ClutterActor is bad or wrong, but because I always get
confused on the difference, and having them both in SlideLayout
makes the code a bit easier to read and understand.
The parent SlidingControl had an onOverviewShowing, but we had
overridden it with the same code in both subclasses. Just move it
back to SlidingControl.
When we create a result actor, cache it, so it can be used for
subsearches of the same initial. For now, to keep memory usage
and the stage graph relatively clean, don't persist the actors
across searches, but maybe we should do this in the future.
This also means that we don't query getResultMetas for items
that we've seen in the same initial search.
viewSelector: Move all of the provider-loading logic to SearchSystem
The existing provider system is split between a confusing mess of
RemoteSearch, SearchSystem, SearchDisplay, and ViewSelector, partly
because of the vestigal in-shell search system. Move most of the
logic to search.js so it's easier to read.
We fetch and store the list of providers from the search system when we
construct SearchResults, but we never update this list when providers are
changed at runtime, causing various bugs making the search not seem as
snappy as it should be. Make sure to always fetch the list of providers
from the search system.
Florian Müllner [Mon, 28 Oct 2013 10:26:17 +0000 (11:26 +0100)]
notificationDaemon: Save notifications on source destruction
While the existing comment is correct in that a source's notifications
will be destroyed first, the code takes a shortcut which prevents the
Source::count-updated signal from being emitted. Given that the purpose
of the signal is to keep notification counters up-to-date which is
pointless when the source is about to be destroyed, the shortcut makes
sense; just save notifications explicitly in that case.
search.js used to do a lot more, but now that most of the
functionality has been moved to the remote search system,
it doesn't do a lot. Merge searchDisplay.js into it.
It's been broken for quite a bit since we removed Panel.Animation,
and hasn't really ever worked with our new search results. It's also
the only non-remote provider left.
Maybe we'll add it back as a remote provider later, but for now, just
ditch it.
projects / gnome-shell.git / log