If the view is not active, our menus are not merged in, so we shouldn't
try to update the relative actions underway, as they're not in the
action group.
This fixes crashers/warnings when doing a file operation with more than
one view open.
This doesn't really work well, it is inconsistent with the file chooser,
isn't use anywhere else in GNOME, and we want to enhance the list
view to cover this case.
https://bugzilla.gnome.org/show_bug.cgi?id=676531
Themes can nowadays set padding on GtkLabel widgets (e.g. Ambiance and
Radiance do this for labels in primary toolbar buttons). This breaks our
hack to force a size request for the pathbar labels, since we measure
the PangoLayout directly instead of measuring the GtkLabel it's part of
(which includes the border/padding values from the theme).
Fix this by measuring the size requisition of GtkLabel directly; for
this to work effectively, we need to pack an (invisible) additional
label in the button GtkBox, always set the text to both labels and
update the requisition of the non-bold one to
MIN(MAX_POSSIBLE_WIDTH, MAX (width, bold_width)) every time a
size-request cycle is called.
https://bugzilla.gnome.org/show_bug.cgi?id=678341
This was introduced in 241e462024 - the
original code did
if (!is_compact || zoom > standard)
So since is_compact is now always false, the commit should have removed
the if block completely.
https://bugzilla.gnome.org/show_bug.cgi?id=678351
This changes the way the wait dialog cancellation callback works. Instead of
cleaning up early we just cancel the operation and wait for proper
rename_callback() carrying the error.
This fixes an assertion failure when the wait dialog has been displayed and user
pressed the Cancel button. Unlikely to happen, found during gvfs testing.
Or if something (e.g. another process via xsettings) causes the
GtkSettings to change the value of gtk-shell-shows-app-menu after a
window is finalized, our callback will run and we'll crash trying to
access freed memory.
https://bugzilla.gnome.org/show_bug.cgi?id=677945
Change the way DnD works in the places sidebar.
After this commit, DnD on the bookmarks area will always mean
"copy/move the DnD source into the bookmark" instead of "add a bookmark
to this location" (DnD of bookmark rows in their section will still
reorder them), since that's the most common action you want to do.
This also makes it consistent with the default action obtained by
dropping on the built-in places below.
https://bugzilla.gnome.org/show_bug.cgi?id=645353
Clean up a11y code and make sure inheritance for
EelCanvasAccessible->NautilusIconContainerAccessible and for
EelCanvasItemAccessible->NautilusIconCanvasItemAccessible work properly.
This makes the ATK state machinery in EelCanvasItemAccessible work
again, which fixes
https://bugzilla.gnome.org/show_bug.cgi?id=677509
If the final class does not override the get_accessible method, it will
inherit it from the parent class; what we're interested in is the first
class for which get_accessible is != from our class get_accessible.
For EelEditableLabel, we need to create an object derived from the
accessible type of the parent class; since that will be a
GtkWidgetAccessible, which is private, we can't just subclass it using
the regular GObject mechanisms.
This functions creates a derived GType by peeking at the parent
accessible class and type.
https://bugzilla.gnome.org/show_bug.cgi?id=672498
Set a maximum allowed width of 250px for pathbar button requisitions,
and ellipsize after that. This fixes buttons disappearing completely
from the pathbar in case there was not enough space to show the complete
folder name,
I am not sure I completely like this approach in all the situations,
since there might be some value in showing more of a very long folder
name if there's space available on screen, but unfortunately, it's not
really possible without restructuring completely the way NautilusPathBar
allocates children.
Adapted for master, and slightly modified from an initial patch by
Ted M Lin <tedmlin@gmail.com>
https://bugzilla.gnome.org/show_bug.cgi?id=313854
