keyring: Don't unregister the prompt when disabled
gnome-keyring provides a fallback in case our builtin prompt fails
to register, so keyring dialogs may still pop up even when they
are supposed to be disabled.
Instead, keep the prompt registered but cancel requests immediately
while disabled.
shell: Make KeyringPrompt.cancel() callable from signal handlers
gcr/gnome-keyring don't handle the case where prompt_password_async()/
prompt_confirm_async() return their result from within the function
very well (or rather: break badly). Calling gcr_prompt_close() instead
is safe, but to avoid these kind of errors in the future, modify
shell_keyring_prompt_cancel() to work as expected in this case.
networkAgent: Don't allow ellipsization of Password label
We don't want password entries to grow when entering more characters
that fit the available width; as labels' ClutterText ellipsizes by
default, the password labels allow entries to grow by shrinking.
Setting the appropriate ellipsize mode fixes this.
keyring: Don't allow ellipsization of Password label
We don't want the password entry to grow when entering more characters
that fit the available width; as labels' ClutterText ellipsizes by
default, the password label allows the entry to grow by shrinking.
Setting the appropriate ellipsize mode fixes this.
We don't make use of any functionality StTable provides over
ClutterTableLayout, so port all users to the Clutter layout
in order to remove our own copy of the code.
We don't make use of any functionality StTable provides over
ClutterTableLayout, so port all users to the Clutter layout
in order to remove our own copy of the code.
main: Close runDialog as necessary on session mode changes
We already do this for looking glass, but it makes even less sense
for the normal run dialog - if a mode sets runDialog to false, the
intention is to not allow executing aribitrary commands.
Carlos Soriano [Tue, 17 Sep 2013 17:41:02 +0000 (19:41 +0200)]
St: Fix fade effect
Commit 4095a58eb9a2921c178b30c56d1fbe7662c2778a introduced a
regression, since we have to take into account four cases,
top, bottom, right, and left, and that can't be merged into
two like that commit did.
So fix it to make fade effect works again.
Carlos Soriano [Thu, 5 Sep 2013 12:48:03 +0000 (14:48 +0200)]
appDisplay: Rework indicators animation
Previously the animation was not entirely according to the mockup.
Now we are closer to the mockup.
The padding for the indicators are decremented, since we need that
to make the animation not too quick. As a drawback, maybe visually
is not as good as before, or the area to click dots is too much little.
Just make that change for now and test it widely, and we can change
that after.
Carlos Soriano [Thu, 5 Sep 2013 22:39:07 +0000 (00:39 +0200)]
appDisplay: Fix indicator animation position
The original position was calculated with the stage and the
transformed position of the indicator when mapped. The values
were wrong on some situations, so lets calculate the position
based on the dots width.
Carlos Soriano [Thu, 5 Sep 2013 22:37:41 +0000 (00:37 +0200)]
theme: Erase appDisplay padding
Since now AllView doesn't have a scrollbar this
padding is not necesary, and will allow to place
the indicators correctly using only the indicators
padding in a upcoming patch.
overviewControls: Use a custom layout to catch all geometry changes
We currently update workspaces geometry when we are notified about
allocation changes of the overview group; however as the geometry
is based on stage coordinates, we miss notifications when the
allocation relative to the parent is unchanged, which happens when
the primary monitor's position changes but not its resolution.
Use a custom layout manager to give us a signal that is emitted
reliably.
st-scroll-view-fade: Reduce number of GLSL uniforms and instructions
The vvalue and hvalue uniforms are only used to decide whether we
should do fade the edges or not based on the fade_edges uniform.
The result does not change accross fragments so there is no reason
to recompute it for every fragment (pixel) so just split the edge
fade into two uniforms and compute the "should we fade the edges"
boolean once for every direction (when setting the uniforms) instead
of for every single fragment twice.
This reduces the number of uniforms as well as the the number of instructions
which are limited on older hardware. It should also be more efficent.
We need to adjust the offset of close buttons, in case the box
pointer has the arrow at the top. To do so, extend close buttons
to hook into a boxpointer (since that's the common use for them)
and automatically adjust their position.
ShellGlobal: use MetaCursorTracker to query the pointer position
Gdk uses Xwayland, so it only sees the events we forward to X11
clients. Instead, we can use the abstraction API provided by
mutter and get the right value automatically.
Also, we need to use MetaCursorTracker to handle the cursor
visibility too.
StEntry: add a hook to set the cursor from libgnome-shell
We need to call into MetaScreen to set the cursor, but we can't
do that from libst, so add a hook that libgnome-shell can fill,
and remove more ClutterX11 usage.
For extremely silly reasons with how the class framework works, the wrapper
method requires "this" to be bound in order for it to work, or else we'll
emit errors in strict mode.
Matthias Clasen [Mon, 26 Aug 2013 22:24:26 +0000 (18:24 -0400)]
Make the search entry behave better in RTL locales
It is expected that the primary and secondary icons in entries
change places in RTL locales. When doing so, the edit-clear
icon must be replaced by an rtl variant too.
projects / gnome-shell.git / log