The Mozilla documentation says: "And as always when working with
reference counted NPObjects, the caller is responsible for calling
NPN_ReleaseObject on the NPObject to drop the reference."
Browsers assume that the plugin does the right thing and always call
NPN_ReleaseObject. At some point the object is released and deallocated
and both the plugin and browser still have references to the object
thinking that it's still alive. That's why the crash is sometimes in the
plugin when it tries to use the np object, and sometimes in the browser.
https://bugzilla.gnome.org/post_bug.cgi
While CoglError is a define to GError, it doesn't follow the convention
of ignoring errors when NULL is passed, but rather treats the error as
fatal :-(
That's clearly unwanted for a compositor, so make sure to always pass
an error parameter where a runtime error is possible
https://bugzilla.gnome.org/show_bug.cgi?id=765061
If cogl_framebuffer_allocate fails in _st_create_shadow_pipeline_from_actor, the
CoglOffscreen* that was allocated earlier in the function is leaked.
https://bugzilla.gnome.org/show_bug.cgi?id=735705
Signed-off-by: Aaron Plattner <aplattner@nvidia.com>
In get_secrets_keyring_cb, we own a ref on the 'attributes' hash table
from secret_item_get_attributes), and a ref on the 'secret' object (from
secret_item_get_secret(), but in the SHELL_KEYRING_SK_TAG case, we unref
these once before breaking out of the loop, and the second time after
breaking out of the loop.
https://bugzilla.gnome.org/show_bug.cgi?id=759708
Note: This is needed to avoid crashes with libsecret 0.18.4 -- Michael
While a channel has pending messages, it will pop up again when
dismissed. That is clearly not what users expect, so clear them
out first before closing a channel.
https://bugzilla.gnome.org/show_bug.cgi?id=747991
The non-interactive requests for 'vpn' settings are forwarded to the UI because
it is able to talk to the auth helpers. However, the VPN requests are identified
by the connection type instead of setting type. That is incorrect and the UI
is not prepared to handle such requests; tries to construct a dialog and fails
miserably:
Gjs-Message: JS LOG: Invalid connection type: vpn
(gnome-shell:13133): Gjs-WARNING **: JS ERROR: Error: No property 'text' in property list (or its value was undefined)
NetworkSecretDialog<._init@resource:///org/gnome/shell/ui/components/networkAgent.js:60
wrapper@resource:///org/gnome/gjs/modules/lang.js:169
_Base.prototype._construct@resource:///org/gnome/gjs/modules/lang.js:110
Class.prototype._construct/newClass@resource:///org/gnome/gjs/modules/lang.js:204
NetworkAgent<._handleRequest@resource:///org/gnome/shell/ui/components/networkAgent.js:724
wrapper@resource:///org/gnome/gjs/modules/lang.js:169
NetworkAgent<._newRequest@resource:///org/gnome/shell/ui/components/networkAgent.js:715
wrapper@resource:///org/gnome/gjs/modules/lang.js:169
https://bugzilla.gnome.org/show_bug.cgi?id=760999
The Next and Sign In buttons are disabled when the username/password
field is empty. However, the user can still bypass this button by
pressing the enter key, leading to some odd glitches with the log in
for 'Not Listed?' users.
This is easy to fix by simply not progressing to the next screen when
the button is disabled.
https://bugzilla.gnome.org/show_bug.cgi?id=746180
Generally a user-changed operation will be uninteresting, but if the
user is currently in the user list and the account changes to locked, we
want to remove it from the list, or if the user is not in the list and
the account changed to unlocked, we want to add it to the list. This
fixes the case where a new user account created in gnome-control-center
does not appear in the user list. The password mode is set in the new
account immediately after it is created, but the operations are not
atomic, so the login dialog considers the new user account when it is
still locked and rejects it from being displayed, then immediately
afterwards the account is unlocked. This commit causes the login dialog
to show the account when this occurs.
The containsUser() check here is not strictly necessary, but reduces
spurious calls to addUser() and removeUser(), since there's no easy way
to check if the locked status of the account has changed (as it's much
easier to connect to one signal on the UserManager than to
notify::locked on each User object).
https://bugzilla.gnome.org/show_bug.cgi?id=758568
LoginDialog has a private _user, but UserListItem has a public user.
Easy to get wrong since _user would be the right thing to type in 90% of
this file.
Merge PluginData and PluginObject structs into a single one and create
the scriptable object associated to the plugin instance in NPP_New. Then,
when NPPVpluginScriptableNPObject is requested we just return the
scriptable object associated to the given instance. This caused the
crashes in NPN_InvokeDefault with WebKit, since we had multiple
scriptable objects for the same instance, but only one of those objects
had the onchange listener installed. Firefox seems to cache the
scriptable object for the instance and therefore NPPVpluginScriptableNPObject
is requested only once.
https://bugzilla.gnome.org/show_bug.cgi?id=737932
NPAPI plugins are windowed by default, so we need to set
NPPVpluginWindowBool value to FALSE on startup. This way the browser
will not create a GtkSocket for a GtkPlug that we are not going to
create. It doesn't make sense to claim that we need XEmbed either.
https://bugzilla.gnome.org/show_bug.cgi?id=757940
NPAPI plugins are windowed by default, so we need to set
NPPVpluginWindowBool value to FALSE on startup. This way the browser
will not create a GtkSocket for a GtkPlug that we are not going to
create. It doesn't make sense to claim that we need XEmbed either.
https://bugzilla.gnome.org/show_bug.cgi?id=757940
This ensures that the module will not be unloaded, since GObject types
registered statically can't be reloaded. This should fix crashes with
browsers that correctly unload the plugins.
https://bugzilla.gnome.org/show_bug.cgi?id=737932
If we are trying to render a shadow at a size that is very large in one
direction, but small in the other direction (so that we don't 9-slice
the texture), then allocating the backing texture for the offscreen
buffer may fail due to texture-size limits. Don't crash in that case.
https://bugzilla.gnome.org/show_bug.cgi?id=757150
There are quite a few crashes in retrace.fedoraproject.org that are a result of
of cairo_pattern_get_surface() failing, then a subsequent call to
cairo_image_surface_get_width() crashing because no surface was returned to the
out parameter. Knowing what causes these is hard - my best guess is widgets getting
allocated at ridiculous sizes - but avoiding the crash makes sense in any case.
See https://bugzilla.redhat.com/show_bug.cgi?id=1206754https://bugzilla.gnome.org/show_bug.cgi?id=756983
It's very unexpected that a spinner animation would
preempt idles from running.
This commit runs the spinner animation with a low
priority to ensure it doesn't take over the main
loop.
https://bugzilla.gnome.org/show_bug.cgi?id=754814
Right now the spinner animation updates every 14ms.
60 frames per second would be one frame per 16.667ms,
so we're waking up more frequently than we need to.
This commit changes the wakeup to happen after 16ms.
https://bugzilla.gnome.org/show_bug.cgi?id=754814
The text-scaling-factor GSetting was not being properly propagated
to clutter and the Pango font map; under X this is done by Clutter,
which listens to XSETTINGS directly.
https://bugzilla.gnome.org/show_bug.cgi?id=756447
If the source actor is sized 0x0, the grabHelper will close the menu
on button release if the menu ends up flipped because in that case the
release event happens when the pointer is neither over the source
actor (since it's 0x0) or over the menu actor. A zero sized source
actor works for the non-flipped menu case because the menu's actor
itself ends up underneath the pointer.
https://bugzilla.gnome.org/show_bug.cgi?id=756605
If we get another effect on the same actor, we should make sure to
remove the clone through the "overwrite" methods provided by Tweener, or
there will be a race that might end up with a stray clone being left
around.
https://bugzilla.gnome.org/show_bug.cgi?id=756714
The translation should describe the difference between the fullscreened
and unfullscreened position of the window - however we are currently
assuming a fullscreen position of (0, 0) instead of the monitor's origin,
which causes glitches on dualscreen setups.
https://bugzilla.gnome.org/show_bug.cgi?id=756697
For menus, it makes more sense to pick a width that fits a reasonable amount
of content rather than a fixed amount of screen estate, so use font-relative
sizes instead of pixel values.
https://bugzilla.gnome.org/show_bug.cgi?id=754581
Some labels in the system status menu - namely network names - are out
of our control, and may thus grow the width "infinitively" unless we
restrict the menu width. So far we have been doing this by setting a
fixed width or max-width, but any value we put there might end up
being too restrictive in some locales. Instead, request a width that
fits all the labels we want to show unellipsized and use that instead
of an arbitrary limit.
https://bugzilla.gnome.org/show_bug.cgi?id=708472
There is nothing preventing callers from replacing the internal
layout manager, and as long as the replacement is a (or derives
from) ClutterBoxLayout, everything should work fine except for
losing a bit of automatic property mapping - and the latter is
easily fixable by moving the setup out of the constructor.
https://bugzilla.gnome.org/show_bug.cgi?id=708472
