- Avoid error '"iconname" may be used uninitialized in this function'
by initializing said variable to NULL.
- Define shell_util_get_file_description as static (like the other
similar functions) to avoid another compiler error.
- Don't save errors from g_key_file_load_from_data_dirs into the
variable "error" (ie. pass NULL to it instead). Without this,
gnome-shell crashes if the key file can't be found (with message
"Error invoking Shell.load_from_desktop_file: Valid key file could
not be found in search dirs").
- Check the result of the load_from_desktop_file() call in places.js,
as it may be null.
Previously, ShellAppSystem only loaded (and cached) the set of
.desktop files from applications.menu and settings.menu, using
the gnome-menus library. The ShellAppInfo structure was
a "hidden typedef" for GMenuTreeEntry.
But we need to support loading an arbitrary .desktop file. Thus,
refactor the ShellAppInfo into a real struct, with a refcount,
and allow it to point to either a GMenuTreeEntry or a GKeyFile.
Also, in the case where we fail to lookup an icon for an
application, ensure we return a 0 opacity texture.
When WellGrid had no child it was doing a division by zero,
which screwed up the calculation for the height request and
having it ask for a negative number. This commit fixes this
by always requesting 0 in this case.
I've done some little modifications to the window positioning code
used in the overlay so that it considers the height of the windows
and they don't overlap or get out of the workspace.
I've also raised the hard-coded scales a bit to have the windows as
big as possible without overlapping (this could use some testing on
a non-widescreen monitor).
Use ShellGenericContainer to implement a fully dynamic layout
for the application well. It's still fixed to 4 columns by default,
but no longer requires a fixed width to be passed in on start.
With another chunk of work, it could likely try to adjust to
the case where we can only fit fewer than 4 items in the well.
Remove the border highlighting on mouseover, since that caused
reallocations, and the grid layout isn't trivial.
Delete the unused shell_global_get_word_with function.
The variables this._yOffset and this._xOffset are included in the
drop coordinates, but as far as I can tell never defined. Looking
back on the commit that introduced this code, they weren't removed
from anywhere else either.
The drop coordinates seem correct without them, so just delete them.
Use the 1x1 actor in the top left corner of the screen to enter and leave
the overview mode by just moving the mouse over to it.
No delay is used for triggering the hot corner because a delay significant
enough to allow moving away the mouse to avoid triggering it ruins the desired
flow when triggering the hot corner is actually intended.
The hot corner is not enabled in the full screen mode because the application
or the virtual system might have a hot corner of its own in that place.
It's easier to explicitly call Main.overlay.hide() instead
of chaining activation signals, this got lost in a mix between
the big dash rewrite and ongoing changes to the Application well.
To better distinguish between vast fields of white of which many
windows are composed, add the application icon to the bottom
right of the window.
We fade them in to avoid an abrupt feel. The icons are in the
workspaces group, not individual workspace groups to avoid
having to adjust them when we scale the workspaces.
Replace Workspace._lookupIndexAndClone with Workspace.lookupIndex,
and make the caller go from index to clone, or clone and index.
Emit the signal at the correct time to take action
on snapback (i.e. after the end of the snapback animation).
Add a boolean to the drag-end signal saying whether it was accepted,
which ensures consumers know whether the drag was successful.
Tweak arrangements with 2,3,4,5 windows in a desktop so:
- Windows are a bit bigger
- All windows for 5 windows are equally sized instead of making
the windows in the bottom row larger
This does cause some more problems with tall windows overlapping
or running off the edge of the workspace, but it's an overall
small improvement to the behavior.
Move the GStreamer initialization from the Javascript code into
shell_recorder_init(). This avoids a dependency on the GStreamer
introspection information and will make it easier to drop the
gir-repository module dependency.
Said function in genericDisplay.js was returning the index of the
actor based upon its position in the entire list, while everywhere
else indexes relative to the currently displayed page were used.
This made actions in the details pane break (bug #590949), so I
replace it with a new function in shell-overflow-list.c,
shell_overflow_list_get_actor_index, which is page based.
This patch is a near-total rewrite of the Dash. First, the dash
code moves into a separate file, dash.js.
Inside dash.js, the components are more broken up into separate
classes; in particular there's now a Pane class and a MoreLink
class. Instead of each section of the dash, when activated,
attempting to close all N-1 other sections, instead there
is the concept of a single "active pane", and when e.g. activating
the More link for documents, if we know there's an active pane
which happens to be the apps, close it.
Many redundant containers were removed from the dash, and all
manual width, height and x/y offsets are entirely gone. We move
the visual apperance closer to the design by using the view-more.svg,
etc.
To complete the removal of height/width calculations from the dash,
we also had to do the same for GenericDisplay. Also clean up
the positioning inside overlay.js so calculation of children's
positioning is inside a single function that flows from screen.width
and screen.height, so in the future we can stop passing the width
into the Dash constructor and call this once and work on screen
resizing.
The previous application monitoring code was originally designed
to be based on WM_CLASS, which was then resolved on a server.
We have that resolution code locally now, so instead
of saving WM_CLASS data, save application IDs.
Also, inside the WM we have a much better
infrastructure for tracking windows. In particular, rather
than polling, we can just watch for focus notification on
the display, and window add/remove.
Instead of polling XScreensaver, use DBus to watch org.gnome.Session
which already has an idle time watch.
Now there is no polling at all inside the monitor.
Add a Notebook class which we then use to pack in separate Hierarchy
and Properties tabs.
Split out the inspector into its own class to avoid bloat in the
main class.
5 seconds is should help ensure we lose work less often on
Alt-f2 restart.
Avoid saving empty lines to history, and filter them out if
we find them.
Minor fixes for the still-not-enabled inspector.
This patch attempts to adjust if we have many tray icons; currently
the simple algorithm is to drop down to a spacing of 8 if we have
more than 6. In the future we should fix the panel layout so that
the clock moves to the side.
Add a dropdown pane triggered by Alt-F2, "lg" which supports interactive
JavaScript evaluation, saving/restoring a history file, as well as
an inspector element to pick by using the mouse.
Using an internal boolean rather than the visibility property seems
more reliable to me. Add a list of internal functions rather than
an if/else chain, so for example an extension could hook something on.
Delete the javascript evaluator in favor of the upcoming lookingGlass.js.
We had duplicate code in appDisplay and workspaces for handling activating
a window; unify that inside workspaces, add an API to Main.overlay to
access it from both contexts.
Also, explicitly raise the clone we're activating to the top
before starting the animation to leave the overlay.
Two calls to _repositionDetails() were done too early,
before the visibility of the results pane had changed,
so they had no effect. Thus when a search was done while
a details pane was active, the pane with the results was
hidden by the details pane.
In addition to fixing this, move the two remaining calls to
the line after changing the results pane's visibility
to make it clearer why the calls are there.
Don't use a clone of an actor that's not part of the scene graph for the
item previews. This patch fixes previews in the details pane for documents
for which we don't have full previews and for applications.
Use create_icon_texture() from AppInfo instead of looking up the file
for gicon when creating an application icon for the details pane.
Now clicking on the information icon of the same item a second
time will hide the details pane. Clicking it another time will
show it again, etc.
This is achieved by adding a 'toggle-details' signal which
switches the visibility of the details pane.
Other than the necessary changes, function _selectIndex (in
genericDisplay.js) has been restructured a bit to avoid
duplicating checks.
The experiment with avoiding ellipsization was a definite failure;
several translations have very long names, and we'd end up
with a single column.
Also fix extra padding; we only want some space at the top, not
left/right.
Rename DocsWidget to RecentDocsWidget (as widgets for most used docs,
docs related to the currently open documents, etc. may be added in the
future), and change the title so that it doesn't abbreviate 'Docs' (for
consistency with the overlay).
Corresponding with the design, if an application is in a running
state (has > 0 windows open), draw a glow behind the name.
To make the display look a bit nicer, set the width of each item
to be equal to the longest word among all the items.
Make dash background darker so that the blue color used for indicating
running apps can be visible. Use the dark blue (almost black) color from
Jeremy's mockup.
Make the dash height be the full screen height minus the height of the panel.
Don't use padding on top or on the bottom.
Remove the border from the main dash, but leave it for the results and details
panes. Make the border slightly transparent.
Make sure the details pane is correctly positioned by not applying the
additional padding when determining its x position.
Timeout callback function should not be rescheduled again with the same timeout
because we compute a new timeout and schedule it again instead. So the callback
function needs to return false to not be scheduled again by default.
Make sure that we calculate the next update time correctly.
Store timeout time instead of the timeout delta, so that it doesn't get outdated.
Create a new callback when the time update happens for the original callback.
Make sure last visited time is updated in the details pane by keeping track
of the description actors created for the detail actors.
Add comments to the new functions.
Extend ShellTextureCache by adding the concept of a policy, which
we expose to the public API for loading URIs.
This lets us have the shell tell the cache to keep the information
icon texture around forever.
Secondly, fix the caching of recent info; we shouldn't always be
loading the backup pixbuf. Move recent info loading entirely
into ShellTextureCache.
