There is a fatal crash in Sugar when retrieving mimetypes for some Journal
entries when contructing the icon palette (for both the Journal list view
and detail view). We isolated the problem to xdgmime.c, the call to
xdg_get_mime_parents. Dnarvaez observed that the annotation for the
return argument was missing and that the annotation for the return for
mime_list_parents was wrong. This patch fixes these annotations.
Fixes#4530
This code was moved from sugar expandedentry.py to be available to use
in other places. Some activities like Portfolio or JournalShare
have this code copied.
Signed-off-by: Gonzalo Odiard <gonzalo@laptop.org>
If the icon can't be found, fall back on showing a generic icon.
Useful when activity/content bundles specify a bad icon, which is
unfortunately true for most content bundles in current existance
(because the icon field was previously unused).
Share get_installation_time() between all bundle types so that it works
with content bundles.
Add get_tags() stub, this is called by Sugar.
Remove unused fields: l10n, category, subcategory, etc.
We remove support for running OLPC's content bundle indexer.
This is assumed to go away now that Sugar will support content bundles
as first-class objects.
Some minor behavioural fixes and cleanups.
This now implements http://wiki.sugarlabs.org/go/Content_bundles
and when care is taken to include all the fields required by old
Sugar versions, there is no change in compatibility of existing or
new content bundles.
Required for http://wiki.sugarlabs.org/go/Features/Content_support
Upon import of uuid, that module uses ctypes.util to find libuuid/libc,
which involves forking a shell process which then forks and runs ldconfig.
ldconfig performs a whole bunch of disk accesses. This is a slowdown.
uuid.getnode() is also documented to be slow. Lets avoid using it in
the activity launch path; we don't need such a high degree of
randomness.
This makes activities launch 1-2 seconds faster on XO-1.
The new bundle helpers were being auto-loaded with sugar3.bundle due
to their placement in __init__.py. This was causing Gi to be imported,
breaking GTK2 activity launches.
Move the helpers to a dedicated module.
The bundle API is a bit confusing and inconsistent. Fix up the
ActivityBundle and ContentBundle classes to be more consistent and to
work with the ongoing improvements being made in bundleregistry.
The last change brak compatibility with the use in activities
without set the filter_type parameter, because None is not
a allowed value in the dbus call.
Signed-off-by: Gonzalo Odiard <gonzalo@laptop.org>
It needs to be called as early as possible before unit tests
has a chance to import gtk.
Hopefully will fix random test failures we are seeing in buildbot.
Constants are defined to select the different filter_type
values. A comment was added to document the use.
Signed-off-by: Gonzalo Odiard <gonzalo@laptop.org>
This ensures the test itself is not recognized as a client
by the registry. I'm not sure how that happens since we are
not loading gtk, but it does. Not being a real gtk client with
a mainloop we wasn't answering dbus calls and causing the registry
to hang and fail.
Development continued there for a while. Now I will remove it
and used the toolkit implementation instead.
Changes:
* Retry faster
* Handle GLib errors
* Don't always dump the tree
* Support master atspi api
This extension is used by Sugar to set the HIDDEN attribute for
.Sugar-Metadata. By creating an extension, we avoid the need to add an
additional dependency (fatattr) to Sugar. The code is modeled after
fatattr, using an IOCTL call get and set FAT file attributes.
There is a corresponding patch for model.py in sugar/src/jarabe/journal
It's a first step to make html activities much lighter. It also
just makes a lot more sense to have them interact directly with
the system rather than proxying through the python activity.
We make use of this API in the shell and in the toolkit itself
and never removed those deprecated calls. Let's leave it like
this for now and remove the comment.
Otherwise anything might access the bus.
Note: This will not work if the user moves to another html page. They
will not have a port/key passed in the query string of the URL, so
they won't be able to connect.
The PaletteWindow lost the horizontal padding when it started to be
used as replacement of palette menus. So now each implementation has
to add the padding.
Signed-off-by: Manuel Quiñones <manuq@laptop.org>
When activity bundles are loaded, the activity name and summary are
pulled from an activity.linfo file that is associated with
locale. However, sometimes locale, which uses $LANG has been
superceded by $LANGUAGE as set in
extensions/cpsection/languages/model.py. This patch uses the same
algorithm as gettext.py to find activity.linfo
This problem is seen on the .AU systems which use en_AU for locale and
en_GB for LANGUAGE.
This is currently breaking the activity list unit test where
the profile has not been initialized. But, more in general,
it seems like we should not be crashing if our settings has
unexpected values.
And use it in the sugar3.mime modules. This allows
to get rid of the pygtk generated sugarbase module,
along with the wrapping code.
Also add more cases to test_time to make sure
everything is still working.
The uitree module exposes the at-spi tree. We can use it to do
functional tests of the UI, by checking if the expected
widgets exists, clicking them etc.
A simple example of how this can be used is in the test, which
runs a window and check that it has the expected button.
The expected parent window did likely change, for example
this can happen when we switch the Home Views while a Palette
of a canvas icon is popping up (SL #4221). In that case
send the 'popdown' signal so that for example the hovering
state feedback can be cleared.
Signed-off-by: Simon Schampijer <simon@laptop.org>
Acked-by: Manuel Quiñones <manuq@laptop.org>
If the default locale is not set (see SL #4450 for a use case), get_locale_path
does now return None so that the sugar-activity script can handle that case
cleanly.
Signed-off-by: Simon Schampijer <simon@laptop.org>
Reviewed-by: Daniel Drake <dsd@laptop.org>
This is the same behaviour as for the RadioToolButton and the
ToolButton. See 5a1b380dd6 where
we did the same for the RadioToolbutton.
Updated the toolbuttons.py test to be able to test that code
path.
Signed-off-by: Simon Schampijer <simon@laptop.org>
Acked-by: Manuel Quiñones <manuq@laptop.org>
Remove the vertical padding from the box that contain the toolbar, and
change the order of drawings in the container widget so that the
outline looks right. This has been tested with screenshots, see the
gif animation attached to the ticket.
Signed-off-by: Manuel Quiñones <manuq@laptop.org>
Acked-by: Simon Schampijer <simon@laptop.org>
A new icon widget useful to display progress. It is compatible with
sugar3.graphics.icon.Icon . The progress is represented filling the
icon.
Signed-off-by: Manuel Quiñones <manuq@laptop.org>
Acked-by: Simon Schampijer <simon@laptop.org>
Otherwise the internal state is messed up, GTK+ grabs may be held on
unrealized widgets.
Signed-off-by: Carlos Garnacho <carlos@lanedo.com>
Acked-by: Simon Schampijer <simon@laptop.org>
This was lost when PaletteMenuItem was added in
2d0ac2a952 of sugar source code.
That class does not handle the text_maxlen argument (default to 60). I
used this argument to set the Gtk.Label's size:
label.set_max_width_chars and use MIDDLE for its EllipSizeMode as
sugar3.graphics.palette.Palette does for its title and subtitle.
With these settings, the title of the palette and its items share the
same behaviour and are rendered with the same width.
Signed-off-by: Manuel Kaufmann <humitos@gmail.com>
Acked-by: Manuel Quiñones <manuq@laptop.org>
There are cases where there is no user interaction but we do
receive a clicked signal: in the clipboard we do have
GtkRadioToolButton which are derived from the GtkToggleToolButton [1].
The 'clicked' signal is emitted when the 'active' property
changes [2]. We use the 'active' property to indicate the
current active clipping. In the Invoker we do check now
if the event originated a user interaction or if it was
generated due to a for example a property change.
[1] http://developer.gnome.org/gtk3/3.4/GtkToggleToolButton.html
[2] http://developer.gnome.org/gtk3/3.4/GtkToggleToolButton.html#GtkToggleToolButton--active
Signed-off-by: Simon Schampijer <simon@laptop.org>
Reviewed-by: Carlos Garnacho <carlos@lanedo.com>
Acked-by: Manuel Quiñones <manuq@laptop.org>
'import gio' was importing static gobject, and mixing static and
dynamic bindings provoked a crash.
Signed-off-by: Manuel Quiñones <manuq@laptop.org>
Acked-by: Simon Schampijer <simon@laptop.org>
In the callback, the allocation can be 1x1 if it is yet unknown [1].
This is the case of trays that start hidden by default, like the one
in Browse activity.
[1] http://developer.gnome.org/gtk3/stable/GtkWidget.html#gtk-widget-size-allocate
Signed-off-by: Manuel Kaufmann <humitos@gmail.com>
Signed-off-by: Manuel Quiñones <manuq@laptop.org>
Acked-by: Simon Schampijer <simon@laptop.org>
In the Journal, CellRendererActivityIcon is calling its do_render
method after the temporal file is removed. This is because we wrap
the icon path in a TempFilePath to control its deletion, but as we
don't maintain any reference to it, its __del__ method deletes the
temporal file.
Note that returning the path directly has one disadventage: the
temporal file is left in /tmp .
As a comment in the get_icon method of ActivityBundle say, ideally we
should return the icon data. But that implies a major change in the
Journal model and views, and in the CellRendererIcon class.
Signed-off-by: Manuel Quiñones <manuq@laptop.org>
Acked-by: Simon Schampijer <simon@laptop.org>
In the palette reimplamentation, commit 48ad255a, set_border_width was
removed. Add it again to the corresponding widget, and get the value
from the theme. Previously the value was given by:
self.get_style().xthickness . Now we can use the widget border.
Signed-off-by: Manuel Quiñones <manuq@laptop.org>
Acked-by: Simon Schampijer <simon@laptop.org>
XInput2 events are coerced by active grabs on the alt-tab handling
code in sugar shell, even if the keygrabber itself does core events.
Signed-off-by: Carlos Garnacho <carlos@lanedo.com>
Acked-by: Simon Schampijer <simon@laptop.org>
1. if the palette is up, draw the black background
2. draw the button itself
3. if the palette is up, draw the grey outline
Exactly as the ToolButton does. Otherwise the outline is not visible.
Signed-off-by: Manuel Quiñones <manuq@laptop.org>
Acked-by: Simon Schampijer <simon@laptop.org>
The Palette is popped down when an item is activated. We can
listen on the 'button-release-event' on that widget. We make
sure the widget is of type PaletteMenuItem. This is similar
to what the GtkMenu is doing, see gtk_menu_shell_button_release [1].
[1] http://git.gnome.org/browse/gtk+/tree/gtk/gtkmenushell.c#n915
Signed-off-by: Simon Schampijer <simon@laptop.org>
Acked-by: Manuel Quiñones <manuq@laptop.org>
Do the same as in ToolButton, update do_draw method to do the drawing
in the code.
Signed-off-by: Simon Schampijer <simon@laptop.org>
Acked-by: Manuel Quiñones <manuq@laptop.org>
Now the do_draw method is exactly the same as the one for ToolButton.
This considers the border that the radio button can have, which is
needed for the changes made in the theme to get back the original
style.
Signed-off-by: Manuel Quiñones <manuq@laptop.org>
Acked-by: Simon Schampijer <simon@laptop.org>
When a CanvasIcon is right-clicked __button_press_event_cb and
__palette_popup_cb are called but not __button_release_event_cb. So,
ACTIVE flag is set but immediately removed by __palette_popup_cb and
PRELIGHT is set.
Now, if the user right-click it again while the palette is popped up,
__button_press_event_cb is called and it sets ACTIVE flag but
__palette_popup_cb is not called again because the palette is already
shown and that makes the icon to keep in ACTIVE state.
This patch checks if the palette is popped up when
__button_press_event_cb is called and if the palette is not popped up
the ACTIVE flag is set.
Signed-off-by: Manuel Kaufmann <humitos@gmail.com>
Acked-by: Manuel Quiñones <manuq@laptop.org>
We have to convert the boxes to EventBoxes because otherwise the
background is not themeable [1]
[1] https://bugzilla.gnome.org/show_bug.cgi?id=678790
Signed-off-by: Manuel Quiñones <manuq@laptop.org>
Acked-by: Simon Schampijer <simon@laptop.org>
If we pass the active property to the constructor do_clicked seem
to get called. Before chaining up to the Gtk.RadioToolButton
constructor we need to initialize the variable.
This is a followup of 5a1b380dd6
Signed-off-by: Simon Schampijer <simon@laptop.org>
Tested-by: Gonzalo Odiard <gonzalo@laptop.org>
Acked-by: Manuel Quiñones <manuq@laptop.org>
