Instead of using a GtkGrid and GtkBox to mimic what GtkStack and
GtkStackSwitcher already does for us, let's just use the latter. As a
result each source now has its separate GtkIconView. The pictures
source has a nested GtkStack to switch between the 'empty' box and the
view, which should simplify size-related issues.
https://bugzilla.gnome.org/show_bug.cgi?id=736366
When the GrlSourceResultCb callback is called, the media's ownership is
given to the callback function (as per documentation that says "media"
is "transfer full").
There are three cases which affect our size calculation:
- When the dialog is without a parent in the test application.
- When the dialog is maximized on small screens.
- When the dialog is used with an unmaximized parent.
The width of the chooser is much more than half the width of the parent
in the third case. So there is no point in doing it.
The width that we set for the grid in the content area is actually what
is useful.
https://bugzilla.gnome.org/show_bug.cgi?id=736366
When selecting a background from the Pictures source, make sure that
the placement is reset from whatever it was set to before, otherwise
it will make use of the previous one, leading to tiled wallpapers if a
colour was selected before, for example.
https://bugzilla.gnome.org/show_bug.cgi?id=703057
We were assuming that setting stock-size would affect the emblems in
GEmblemedIcons, but not the icons themselves. This is a bit weird.
GtkCellRendererPixbuf:gicon is meant to work with
GtkCellRendererPixbuf:stock-size, and this was only working so far
because GTK_ICON_LOOKUP_FORCE_SIZE was not being used when loading
the icon.
Let's composite the emblems ourselves so that we don't have to depend
on this quirky interpretation of stock-size.
Unfortunately, we can not directly use the pixbufs because they are
unaware of the scale factor and GTK+ will scale them on HiDpi
displays. Since our pixbufs already have enough pixels to work well
with such devices, scaling them further will lead to giant, fuzzy
thumbnails. Hence, we use GtkCellRendererPixbuf:surface with the
scale factor codified in it.
https://bugzilla.gnome.org/show_bug.cgi?id=732375
This is deprecated in newer automake versions, and this causes warnings
with automake 1.14:
panels/printers/Makefile.am:3: warning: 'INCLUDES' is the old name for
AM_CPPFLAGS' (or '*_CPPFLAGS')
https://bugzilla.gnome.org/show_bug.cgi?id=732189
... so that it is possible to override virtual functions and change
the order of the object creation and destruction process without having
to write a bunch of prototypes. Having the class_init below the others
helps us do that.
https://bugzilla.gnome.org/show_bug.cgi?id=731713
When using a small screen, make sure that the whole of the dialogue
is visible, including the confirmation buttons. This means that the
dialogue will be maximised when on smaller screens (just like its
parent), and that we won't set a number of columns for the icon
view.
We will also stop making the dialogue much bigger than the
Settings window itself.
https://bugzilla.gnome.org/show_bug.cgi?id=692495
A completely black background will make it impossible to see
whether there is one more workspace in gnome-shell, and as we
do not want to design gnome-shell around the possibility of
a solid black coloured background, remove black from the options.
https://bugzilla.gnome.org/show_bug.cgi?id=649426
There is no need to pass the source-uri as a parameter as we can get it
from the file itself. This would help in using the same code paths for
local and remote pictures.
https://bugzilla.gnome.org/show_bug.cgi?id=707569
We do not wait until the first thumbnail is created to start filling up
the store. This avoids the "No Pictures Found" message from showing up
on slow machines were I/O and decoding the data may take a while.
However, placeholders are not shown for PNGs since they might be
screenshots.
https://bugzilla.gnome.org/show_bug.cgi?id=708943