| Commit message (Collapse) | Author | Age | Files | Lines |
... | |
| |
|
|
|
|
|
|
|
|
|
|
|
| |
Shells can fix docks, panels and other similar things which reduce the
available space to draw on. This is called the canvas margin.
When positioning things relative to icons within containers we need to
adjust by this margin, otherwise they will be shifted away from the
target. Do this when we calculate the bounding box.
https://bugzilla.gnome.org/show_bug.cgi?id=759002
|
| |
|
|
|
|
| |
Conflict resolution in commit e6587b86b4c75 went wrong on my part.
|
|
|
|
|
|
|
| |
We were assuming trash and native_path as local, but not recent,
which is wrong.
So assume recent as local, with the benefit that we can use it
for mime type polling and remove some dead code now.
|
|
|
|
| |
https://bugzilla.gnome.org/show_bug.cgi?id=758898
|
|
|
|
| |
https://bugzilla.gnome.org/show_bug.cgi?id=758632
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
We were checking how many days ago the file was modified by
checking directly the current time with the file modification
time, which is wrong if the file was modified a few hours ago
but the day already passed.
What happened is that it was showing the time as in if it was
modified in the same day, instead of saying "Yesterday".
To fix it, check the time difference with the midnight time.
https://bugzilla.gnome.org/show_bug.cgi?id=757272
|
| |
|
| |
|
| |
|
| |
|
|
|
|
|
| |
They shouldn't need to hold a ref to the whole view and makes
ownership management more complex.
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
We were leaking it, causing that the view was not free if
there were some templates, since templates menu items were
holding a ref to the view and are only freed when the closure
of the action gets freed, which is the view action group.
This was making the view to get never finalized, and therefore
making all kind of problems afterwards.
Upcoming patch will also make template not hold a ref to the view,
even if the issue is fixed now, I think templates menu items holding
a reference to the view is wrong.
|
|
|
|
|
| |
We were leaking them refing and not unrefing.
ref sink them and unref them on finalize.
|
| |
|
| |
|
|
|
|
|
|
|
| |
Maintenance is not that much as long as we don't show
it in any UI, and was requested by several people.
https://bugzilla.gnome.org/show_bug.cgi?id=757375
|
|
|
|
| |
Probably if didn't crash before because we are leaking elsewhere...
|
| |
|
|
|
|
| |
So we can debug ref counting issues more comfortably.
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
| |
Since bgo#757826 introduced g_application_set_resource_base_path,
we don't need to use a GtkBuilder to load the appmenu.
GtkApplication automatically loads menus from the GtkBuilder resource
located at "gtk/menus.ui", relative to the application's resource
base path (see g_application_set_resource_base_path ()). The menu
with the ID "app-menu" is taken as the application's app menu.
https://bugzilla.gnome.org/show_bug.cgi?id=757946
|
|
|
|
| |
For behave accordingly with gtkplacessidebar and gtkplacesview.
|
| |
|
|
|
|
|
|
|
| |
Use GTK+'s new shortcut window to document our shortcuts.
See: https://wiki.gnome.org/Initiatives/GnomeGoals/ShortcutWindows
https://bugzilla.gnome.org/show_bug.cgi?id=757826
|
| |
|
| |
|
|
|
|
| |
(cherry picked from commit 9f8f4fb12ecc8985ccf5bca5d0feeae4ce3a0473)
|
| |
|
| |
|
| |
|
| |
|
|
|
|
|
| |
Make sure we don't eat errors without at least printing them on the
terminal.
|
|
|
|
|
| |
The "unhelpful error dialog" code already takes care of freeing the
activation parameters. Make sure it does not do that twice.
|
|
|
|
|
| |
We might not have a PackageKit name owner. Check for that case, or the
system will think that it can install extra applications.
|
| |
|
| |
|
|
|
|
|
|
|
| |
We need to check at runtime if we are actually running under X,
before calling gdk_x11 API.
http://bugzilla.gnome.org/show_bug.cgi?id=758026
|
| |
|
| |
|
| |
|
|
|
|
|
|
|
| |
On languages different than English was not enough, making the
text not understandable.
https://bugzilla.gnome.org/show_bug.cgi?id=754968
|
|
|
|
|
|
|
|
|
| |
Last patch now makes view reference counting works, however
NautilusFilesView sink the ref on creation, and NautilusPlacesView
was not, making NautilusPlacesView holding a reference less than other
views and making nautilus crash.
Follow what other views do and sink the reference.
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This was leaking and making that no view was finalized, leading
to all kind of problems and crashes due to signals and files updating
the dangling view, and not letting files to be released.
This was creating multiple non-related reports on Fedora Tracker making
nautilus one of the top crashers.
Downstream tracker bug:
https://bugzilla.redhat.com/show_bug.cgi?id=1277625
The upcoming patch is also necessary for a complete fix.
|