gtk4.git
9 years agotogglebutton: Move :inconsistent to GtkCheckButton
Timm Bäder [Sat, 28 Jan 2017 07:11:14 +0000 (08:11 +0100)]
togglebutton: Move :inconsistent to GtkCheckButton

9 years agotogglebutton: Move :draw-indicator property to GtkCheckButton
Timm Bäder [Fri, 27 Jan 2017 11:02:47 +0000 (12:02 +0100)]
togglebutton: Move :draw-indicator property to GtkCheckButton

It's not used in GtkToggleButton at all, only in GtkCheckButton and
GtkRadioButton.

9 years agocheckbutton: Remove unused vfunc
Timm Bäder [Fri, 27 Jan 2017 10:16:03 +0000 (11:16 +0100)]
checkbutton: Remove unused vfunc

9 years agocheckbutton: Remove unused function prototype
Timm Bäder [Fri, 27 Jan 2017 10:07:41 +0000 (11:07 +0100)]
checkbutton: Remove unused function prototype

9 years agowidget: Remove gtk_widget_class_list_style_properties
Timm Bäder [Thu, 26 Jan 2017 17:58:10 +0000 (18:58 +0100)]
widget: Remove gtk_widget_class_list_style_properties

Unused.

9 years agogtkbuilder: Remove unused struct
Timm Bäder [Sat, 28 Jan 2017 17:24:30 +0000 (18:24 +0100)]
gtkbuilder: Remove unused struct

9 years agobuild: Stop compressing ui files
Timm Bäder [Thu, 26 Jan 2017 12:31:56 +0000 (13:31 +0100)]
build: Stop compressing ui files

It's slower at runtime. Instead, strip blanks.

9 years agomir: handle surface placement events
William Hua [Sun, 29 Jan 2017 16:24:43 +0000 (11:24 -0500)]
mir: handle surface placement events

This allows the Mir backend to properly emit "moved-to-rect."

9 years agomir: use mir_surface_spec_set_placement for menus
William Hua [Fri, 27 Jan 2017 19:46:10 +0000 (14:46 -0500)]
mir: use mir_surface_spec_set_placement for menus

This API was added to Mir for GTK menus, combo boxes, etc.

9 years agoUpdate Galician translation
Fran Dieguez [Sun, 29 Jan 2017 19:59:21 +0000 (19:59 +0000)]
Update Galician translation

9 years agoUpdate Galician translation
Fran Dieguez [Sun, 29 Jan 2017 19:48:13 +0000 (19:48 +0000)]
Update Galician translation

9 years agoUpdate Slovak translation
Dušan Kazik [Sun, 29 Jan 2017 09:32:03 +0000 (09:32 +0000)]
Update Slovak translation

9 years agoUpdate Slovak translation
Dušan Kazik [Sun, 29 Jan 2017 09:30:05 +0000 (09:30 +0000)]
Update Slovak translation

9 years agoRemove tests/testtreemenu
Daniel Boles [Sat, 28 Jan 2017 20:21:56 +0000 (20:21 +0000)]
Remove tests/testtreemenu

Since GtkTreeMenu became a private class only used by GtkComboBox, all
this test actually did was to show a ComboBox constructed with a custom
CellArea. Now that the latter is no longer possible, the test just shows
a handful of settings that do nothing. Just test GtkComboBox directly.

9 years agoFix translator comments
Piotr Drąg [Sat, 28 Jan 2017 00:10:12 +0000 (01:10 +0100)]
Fix translator comments

They need to be exactly one line above a string to show up in .po files.

9 years agoRevert "notebook: redraw arrows after closing a tab"
Michael Catanzaro [Fri, 27 Jan 2017 14:54:24 +0000 (08:54 -0600)]
Revert "notebook: redraw arrows after closing a tab"

This reverts commit 218beb2ef49d407b547047693605328c10d4877d.

9 years agowayland: Ensure to set gtk_surface dbus properties after hide()
Carlos Garnacho [Wed, 25 Jan 2017 14:53:55 +0000 (15:53 +0100)]
wayland: Ensure to set gtk_surface dbus properties after hide()

Currently hiding destroys the wl_surface and all related interfaces,
(including the gtk_surface1) so the next time the GdkWindow is mapped,
we don't bother to set the DBus properties. Toggle the check off so
it's actually issued again after the GdkWindow gets a gtk_surface1.

https://bugzilla.gnome.org/show_bug.cgi?id=773686

9 years agoUpdate Kazakh translation
Baurzhan Muftakhidinov [Wed, 25 Jan 2017 09:11:28 +0000 (09:11 +0000)]
Update Kazakh translation

9 years agoUpdate Kazakh translation
Baurzhan Muftakhidinov [Wed, 25 Jan 2017 09:08:46 +0000 (09:08 +0000)]
Update Kazakh translation

9 years agonotebook: redraw arrows after closing a tab
Michael Catanzaro [Sat, 21 Jan 2017 02:19:56 +0000 (20:19 -0600)]
notebook: redraw arrows after closing a tab

https://bugzilla.gnome.org/show_bug.cgi?id=777547

9 years agoUpdate German translation
Mario Blättermann [Tue, 24 Jan 2017 13:32:48 +0000 (13:32 +0000)]
Update German translation

9 years agowidget: Remove gtk_widget_get_preferred_size mention from docs
Timm Bäder [Sun, 22 Jan 2017 14:01:53 +0000 (15:01 +0100)]
widget: Remove gtk_widget_get_preferred_size mention from docs

Mention gtk_widget_measure instead.

9 years agoflowbox: Measure child widget sizes using gtk_widget_measure
Timm Bäder [Sun, 22 Jan 2017 14:01:40 +0000 (15:01 +0100)]
flowbox: Measure child widget sizes using gtk_widget_measure

9 years agowidget: Remove warning about already visible widgets
Timm Bäder [Sun, 22 Jan 2017 13:48:28 +0000 (14:48 +0100)]
widget: Remove warning about already visible widgets

That was just added by mistake of course.

9 years agoRemove unnecessary gtk_widget_show calls
Timm Bäder [Sun, 22 Jan 2017 13:37:56 +0000 (14:37 +0100)]
Remove unnecessary gtk_widget_show calls

9 years agogtkfontbutton: hexpand the font label
Timm Bäder [Sat, 21 Jan 2017 07:45:44 +0000 (08:45 +0100)]
gtkfontbutton: hexpand the font label

To restore the previous look.

9 years ago3to4: Fix typo in previous commit
Daniel Boles [Sat, 21 Jan 2017 22:43:35 +0000 (22:43 +0000)]
3to4: Fix typo in previous commit

9 years ago3to4: Fix suggested replacements for border-width
Daniel Boles [Sat, 21 Jan 2017 22:39:06 +0000 (22:39 +0000)]
3to4: Fix suggested replacements for border-width

The new CSS border-spacing does what Grid::(row|column)_spacing and
Box::spacing already did, i.e. controlling the space added between child
widgets, so it’s not a replacement for Container::border-width.

9 years agocombobox: Stop pointlessly NULL-checking the menu
Daniel Boles [Sat, 21 Jan 2017 20:40:21 +0000 (20:40 +0000)]
combobox: Stop pointlessly NULL-checking the menu

We now have a valid menu to use at all times.

9 years agocombobox: Move area and menu to .ui file
Daniel Boles [Sat, 21 Jan 2017 20:23:27 +0000 (20:23 +0000)]
combobox: Move area and menu to .ui file

Now that priv->area is guaranteed to be constructed by us, and not
passed in by a user, we can move it to the .ui file and stop manually
managing its lifetime altogether. And once the area is there, we can
move the menu there too (and stop pointlessly destroying/rebuilding it).

9 years agocombobox: Remove property cell-area
Daniel Boles [Sat, 21 Jan 2017 12:54:49 +0000 (12:54 +0000)]
combobox: Remove property cell-area

The CellArea is going to become a pure implementation detail & be moved
to the .ui, instead of letting users mess with it (if anyone ever did).

9 years agocombobox: Reuse code rather than reimplementing it
Daniel Boles [Sat, 21 Jan 2017 15:11:40 +0000 (15:11 +0000)]
combobox: Reuse code rather than reimplementing it

We already have cell_layout_is_sensitive() to get whether at least one
cell in a Layout is sensitive, which we need because CellLayout/View
do not implement foreach(). So, since we wrote that, we can use it to
check our CellArea too, instead of doing foreach with a custom callback.

9 years agocombobox: some code cleanups
Daniel Boles [Sat, 21 Jan 2017 14:44:30 +0000 (14:44 +0000)]
combobox: some code cleanups

 * Merge adjacent if statements with the same condition
 * Add a space between a function and its declaration

9 years agomenushell: Fix select_last() to get last, not 1st
Daniel Boles [Sat, 21 Jan 2017 18:53:38 +0000 (18:53 +0000)]
menushell: Fix select_last() to get last, not 1st

It was broken in the same way as select_first(), as explained in
commit a5211adb225172195d6778d652b58632c38afcbf

9 years agocombobox: Fix documentation typo bx => box
Daniel Boles [Sat, 21 Jan 2017 16:10:32 +0000 (16:10 +0000)]
combobox: Fix documentation typo bx => box

9 years agomenushell: Make select_first() do that, not last…
Daniel Boles [Sat, 21 Jan 2017 15:50:54 +0000 (15:50 +0000)]
menushell: Make select_first() do that, not last…

It was looping over all items, not breaking out when it found the first
selectable one, and then selecting the _last_ selectable one (if any)
found. So, it did exactly the opposite of its name. This made me quite
baffled when opening a submenu with right-arrow put me at its last item.

Originally, the loop set to_select and broke if the current item was
selectable and not tear-off, meaning that it would correctly select the
first suitable item. However, when tear-off functionality was removed
in commit 4ed9452e9028bb90589474783cebac57ab66e2c9, so was the break.

9 years agoUpdate Polish translation
Piotr Drąg [Fri, 20 Jan 2017 22:47:04 +0000 (23:47 +0100)]
Update Polish translation

9 years agoFix CSS spelling in a translatable string
Piotr Drąg [Fri, 20 Jan 2017 22:46:38 +0000 (23:46 +0100)]
Fix CSS spelling in a translatable string

9 years agomenushell: Don’t dereference before typechecking
Daniel Boles [Fri, 20 Jan 2017 22:15:42 +0000 (22:15 +0000)]
menushell: Don’t dereference before typechecking

9 years agoUpdate POTFILES.in and POTFILES.skip
Piotr Drąg [Fri, 20 Jan 2017 21:11:22 +0000 (22:11 +0100)]
Update POTFILES.in and POTFILES.skip

9 years agocombobox: If typechecking, don’t reference before
Daniel Boles [Fri, 20 Jan 2017 20:43:19 +0000 (20:43 +0000)]
combobox: If typechecking, don’t reference before

9 years agoDon't hide windows and popovers in init()
Timm Bäder [Fri, 20 Jan 2017 09:37:29 +0000 (10:37 +0100)]
Don't hide windows and popovers in init()

Instead, don't make them visible in gtk_widget_init in the first place.

9 years agofontbutton: Be a GtkWidget
Timm Bäder [Thu, 19 Jan 2017 19:15:43 +0000 (20:15 +0100)]
fontbutton: Be a GtkWidget

9 years agoassistant: Don't forget to remove child widgets
Timm Bäder [Thu, 19 Jan 2017 14:02:59 +0000 (15:02 +0100)]
assistant: Don't forget to remove child widgets

9 years agotoolitemgroup: Properly remove internal child widget
Timm Bäder [Thu, 19 Jan 2017 13:37:51 +0000 (14:37 +0100)]
toolitemgroup: Properly remove internal child widget

We now need to use gtk_widget_unparent and can't just destroy it.

9 years agopopovermenu: Remove forall implementation
Timm Bäder [Thu, 19 Jan 2017 13:34:42 +0000 (14:34 +0100)]
popovermenu: Remove forall implementation

9 years agobutton: Don't explicitly show child widgets
Timm Bäder [Thu, 19 Jan 2017 11:30:23 +0000 (12:30 +0100)]
button: Don't explicitly show child widgets

9 years agorecorder: Adapt to widget visibility changes
Timm Bäder [Thu, 19 Jan 2017 11:26:56 +0000 (12:26 +0100)]
recorder: Adapt to widget visibility changes

We now need to explicitly hide the details label.

9 years agoFix some widget visibility fallout
Timm Bäder [Thu, 19 Jan 2017 10:00:30 +0000 (11:00 +0100)]
Fix some widget visibility fallout

9 years agowidget: Remove show-all property
Timm Bäder [Thu, 19 Jan 2017 09:31:03 +0000 (10:31 +0100)]
widget: Remove show-all property

Doesn't make sense anymore now that gtk_widget_show_all is gone.

9 years agoRemove gtk_widget_show_all
Timm Bäder [Thu, 19 Jan 2017 09:02:04 +0000 (10:02 +0100)]
Remove gtk_widget_show_all

9 years agowidget: Remove show_all vfunc
Timm Bäder [Thu, 19 Jan 2017 07:50:37 +0000 (08:50 +0100)]
widget: Remove show_all vfunc

9 years agofilechooserbutton: Be a GtkWidget
Timm Bäder [Wed, 18 Jan 2017 21:21:42 +0000 (22:21 +0100)]
filechooserbutton: Be a GtkWidget

Instead of subclassing GtkBox to have 2 child widgets, subclass
GtkWidget and only render one of them.

9 years agosizerequest: Use gtk_widget_measure to measure widget sizes
Timm Bäder [Mon, 16 Jan 2017 17:47:31 +0000 (18:47 +0100)]
sizerequest: Use gtk_widget_measure to measure widget sizes

9 years agogtkbuilder: Improve error messages
Timm Bäder [Sat, 14 Jan 2017 09:08:53 +0000 (10:08 +0100)]
gtkbuilder: Improve error messages

9 years agoshortcutswindow: Unset search_bar in dispose
Timm Bäder [Thu, 12 Jan 2017 16:13:19 +0000 (17:13 +0100)]
shortcutswindow: Unset search_bar in dispose

And check for null-ness in unmap. This prevents a crash when the window
is first disposed and then unmapped.

9 years agoshortcutswindow: Use GtkButton:icon-name for search button
Timm Bäder [Thu, 12 Jan 2017 12:10:14 +0000 (13:10 +0100)]
shortcutswindow: Use GtkButton:icon-name for search button

9 years agoshortcutsshortcut: Be a GtkWidget
Timm Bäder [Thu, 12 Jan 2017 11:59:06 +0000 (12:59 +0100)]
shortcutsshortcut: Be a GtkWidget

Instead of subclassing GtkBox and overriding _add to not work, subclass
GtkWidget and add a normal GtkBox as a child widget.

9 years agogtkshortcutswindow: Adapt to GtkWidget visibility changes
Timm Bäder [Thu, 12 Jan 2017 10:26:55 +0000 (11:26 +0100)]
gtkshortcutswindow: Adapt to GtkWidget visibility changes

9 years agowidget-factory: Remove some gtk3-widget-factory leftovers
Timm Bäder [Thu, 12 Jan 2017 09:12:31 +0000 (10:12 +0100)]
widget-factory: Remove some gtk3-widget-factory leftovers

9 years agofilechooser: adapt to GtkWidget:visible changes
Timm Bäder [Thu, 12 Jan 2017 09:04:35 +0000 (10:04 +0100)]
filechooser: adapt to GtkWidget:visible changes

9 years agowidget-factory: Adapt to GtkWidget visibility changes
Timm Bäder [Thu, 12 Jan 2017 08:47:50 +0000 (09:47 +0100)]
widget-factory: Adapt to GtkWidget visibility changes

Make the infobar and the toolbar invisible by default and remove all the
lines showing the other widgets.

9 years agocoloreditor: Hide overlay boxes
Timm Bäder [Thu, 12 Jan 2017 08:43:45 +0000 (09:43 +0100)]
coloreditor: Hide overlay boxes

9 years agoHide popovers and toplevels by default
Timm Bäder [Thu, 12 Jan 2017 08:43:17 +0000 (09:43 +0100)]
Hide popovers and toplevels by default

9 years agowidget: Be visible by default
Timm Bäder [Thu, 12 Jan 2017 08:29:59 +0000 (09:29 +0100)]
widget: Be visible by default

9 years agocombobox: popdown() the menu during unmap()
Daniel Boles [Fri, 20 Jan 2017 03:30:36 +0000 (03:30 +0000)]
combobox: popdown() the menu during unmap()

combo_box_popdown() currently skips popping down our menu if it is NULL.
But the required call to this at end-of-life was in destroy(), by which
point dispose() already NULLed the menu, so Menu::popdown() would never
run, even if it should. Fix this by trying popdown() earlier in unmap().
Also, add a converse assurance that we don’t popup() while not mapped.

Even once we remove all the now-pointless NULL checks, destroy() was the
wrong place to call combo_box_popdown(), and unmap() is the right place.

9 years agoRevert "wayland: Push NULL buffer when hiding a GdkWindow"
Carlos Garnacho [Fri, 20 Jan 2017 13:04:29 +0000 (14:04 +0100)]
Revert "wayland: Push NULL buffer when hiding a GdkWindow"

This reverts commit 6beb0b91c91dd4a018b0fe2c39f5687a00938821.

9 years agoupdate zh_CN translation
Mandy Wang [Fri, 20 Jan 2017 12:37:01 +0000 (20:37 +0800)]
update zh_CN translation

9 years agogtkmain: Fix build on Windows
Chun-wei Fan [Fri, 20 Jan 2017 10:38:53 +0000 (18:38 +0800)]
gtkmain: Fix build on Windows

gtk_init() removed its support for supporting arguments, so we ought to do
likewise for Windows, which actually defines items that call gtk_init()
the old way (and also get rid of argument support in those functions,
since the direction is to not support them).

https://bugzilla.gnome.org/show_bug.cgi?id=773299

9 years agoVisual Studio builds: Update GSK project
Chun-wei Fan [Fri, 20 Jan 2017 10:35:52 +0000 (18:35 +0800)]
Visual Studio builds: Update GSK project

Another GSK Vulkan renderer source was added for the build, so include it
here.

9 years agogtk: Register GSK resources at startup
Benjamin Otte [Wed, 18 Jan 2017 21:41:20 +0000 (22:41 +0100)]
gtk: Register GSK resources at startup

That way they show up in the resource list in the inspector even when
running with the Cairo renderer.

9 years agocombobox: Remove dead code & unneeded type checks
Daniel Boles [Thu, 19 Jan 2017 22:28:41 +0000 (22:28 +0000)]
combobox: Remove dead code & unneeded type checks

Commit fdc0c6426b6fb751653f3a8536a344f69a6407ac removed the appears-as-
list style property, & hence the ability to put the ComboBox into list
mode – but it left behind a pile of hijinks that were only used in said
mode & so were now doing absolutely nothing. This commit deletes those.

While doing that, I got carried away…so this also stops pointlessly type
checking popup_widget, as that can never be anything but a GtkTreeMenu.
It still checks for NULL everywhere, which shouldn’t be needed, but (A)
this commit is already too big, & (B) simply removing such checks where
they _seem_ unnecessary causes bad times. I’ll puzzle through that later

9 years agotestcombo—Don’t set appears-as-list style property
Daniel Boles [Thu, 19 Jan 2017 21:51:22 +0000 (21:51 +0000)]
testcombo—Don’t set appears-as-list style property

List mode no longer exists, so this had no effect.

9 years agocombobox: Make wrap-width and grid mode work again
Daniel Boles [Thu, 19 Jan 2017 21:23:51 +0000 (21:23 +0000)]
combobox: Make wrap-width and grid mode work again

Commit fdc0c6426b6fb751653f3a8536a344f69a6407ac for removing (partly!)
appears-as-list also deleted the code that propagated wrap-width to the
TreeMenu and thus put us into “grid mode”. This restores that code.

And as Benjamin noted, calling check_appearance() here is wrong, so bye.

9 years agoAdd a section about initialization to the migration guide
Matthias Clasen [Thu, 19 Jan 2017 18:52:52 +0000 (13:52 -0500)]
Add a section about initialization to the migration guide

Just the facts.

9 years agoUpdate callers
Matthias Clasen [Wed, 28 Dec 2016 13:53:22 +0000 (08:53 -0500)]
Update callers

Adapt all our tests and examples to the new initialization api.

9 years agoDrop gdk init api
Matthias Clasen [Wed, 28 Dec 2016 21:39:22 +0000 (16:39 -0500)]
Drop gdk init api

Supporting separately initialized gdk is just not worth it.

9 years agoDrop arguments from gtk_init
Matthias Clasen [Wed, 28 Dec 2016 13:49:37 +0000 (08:49 -0500)]
Drop arguments from gtk_init

This is our ultimate goal: no more commandline argument handling.
Lots of callers will have to be updated.

9 years agoSimplify initialization code
Matthias Clasen [Wed, 28 Dec 2016 12:46:56 +0000 (07:46 -0500)]
Simplify initialization code

Now that gtk_get_option_group and gtk_parse_args are no longer
public, we can get rid of them.

9 years agoDrop the --gtk-debug and --gtk-no-debug options
Matthias Clasen [Tue, 27 Dec 2016 23:49:30 +0000 (18:49 -0500)]
Drop the --gtk-debug and --gtk-no-debug options

We have environment variables that cover this.

9 years agoDrop the --g-fatal-warnings argument
Matthias Clasen [Tue, 27 Dec 2016 23:48:02 +0000 (18:48 -0500)]
Drop the --g-fatal-warnings argument

We don't need it, GLib has an environment variable for this.

9 years agoDrop the --gtk-modules argument
Matthias Clasen [Tue, 27 Dec 2016 23:46:45 +0000 (18:46 -0500)]
Drop the --gtk-modules argument

We don't need it; the environment variable is enough.

9 years agoDon't export gtk_get_option_group
Matthias Clasen [Tue, 27 Dec 2016 23:45:53 +0000 (18:45 -0500)]
Don't export gtk_get_option_group

We want to get rid of commandline option handling in GTK+.
This is a step in that direction.

9 years agoStop exporting gtk_parse_args
Matthias Clasen [Tue, 27 Dec 2016 23:36:19 +0000 (18:36 -0500)]
Stop exporting gtk_parse_args

We want to stop handling commandline arguments, and that is the
sole purpose of this function. So it has to go

9 years agoDrop gtk_init_with_args
Matthias Clasen [Tue, 27 Dec 2016 23:31:03 +0000 (18:31 -0500)]
Drop gtk_init_with_args

We want to simplify our initialization code and remove all commandline
argument handling from it. The first stop for this is to reduce the
number of gtk_init variants we have.

9 years agogdk: Skip GdkVulkanContext methods when building introspection data
Emmanuele Bassi [Thu, 19 Jan 2017 18:18:07 +0000 (18:18 +0000)]
gdk: Skip GdkVulkanContext methods when building introspection data

We cannot use Vulkan types through introspection anyway.

9 years agowayland: Push NULL buffer when hiding a GdkWindow
Carlos Garnacho [Thu, 19 Jan 2017 15:07:13 +0000 (16:07 +0100)]
wayland: Push NULL buffer when hiding a GdkWindow

This is how windows are meant to be hidden as per the wayland
protocol, there's no need to destroy the xdg_surface and other
interfaces.

Also, rename gdk_wayland_window_hide_surface() to clear_surface(),
as that's what it does.

https://bugzilla.gnome.org/show_bug.cgi?id=773686

9 years agocombobox: Avoid a pointless assignment
Daniel Boles [Thu, 19 Jan 2017 03:26:09 +0000 (03:26 +0000)]
combobox: Avoid a pointless assignment

Don’t get the active item pointer before the grid/non-grid conditional,
because if we’re in grid mode, we re-get it before selecting it anyway.

9 years agocombobox: Also preselect active item in grid popup
Daniel Boles [Thu, 19 Jan 2017 03:12:32 +0000 (03:12 +0000)]
combobox: Also preselect active item in grid popup

i.e. when wrap-width > 0. This was only being done for non-grid cases.
So, ComboBoxes in grid mode did not indicate their selection when popped
up and required users to keynav from ‘nothing’ (at the top-left) to the
item they wanted to select. By selecting the active item in advance, now
it’s highlighted & acts as the starting point for keynav around the grid

9 years agoscale: Really fix documentation of set_digits()
Daniel Boles [Thu, 19 Jan 2017 00:44:35 +0000 (00:44 +0000)]
scale: Really fix documentation of set_digits()

This previously only mentioned its effect on the displayed value, and
even after the previous commit, its rounding of the actual value upon
change still reads like too much of an afterthought. Worse, it wasn’t
mentioned at all in the doc for the @digits parameter. Change this to
emphasise rounding always occurs and the displayed value is secondary.

9 years agoscale: doc: set_digits doesn’t round retroactively
Daniel Boles [Wed, 18 Jan 2017 23:42:55 +0000 (23:42 +0000)]
scale: doc: set_digits doesn’t round retroactively

Whether it should is an open question, but for now, the documentation
should clearly indicate that currently rounding is only applied upon
changes to the value, not to the existing value when ::digits changes.
This is already clear in the doc for the underlying Range::round-digits.

https://bugzilla.gnome.org/show_bug.cgi?id=358970

9 years agoScale: Always sync ::digits to Range::round-digits
Daniel Boles [Wed, 18 Jan 2017 22:54:42 +0000 (22:54 +0000)]
Scale: Always sync ::digits to Range::round-digits

The documents state that gtk_scale_set_digits() “causes the value of the
adjustment to be rounded off to this number of digits, so the retrieved
value matches the value the user saw.” Note the lack of any condition.

But in fact, if draw-value was false, rounding was disabled on the base
Range, so values that weren’t displayed weren’t rounded. This made the
docs wrong and made an apparently cosmetic detail alter functionality.

Fix by ensuring the number of digits set on Scale is always propagated
along to gtk_range_set_round_digits(), thus rounding to it in all cases
when the value changes, regardless of whether the value is displayed.

This doesn’t address the other idea from Bugzilla: that changing the
number of digits should clamp the _existing_ value if it’s more precise.
This contradicts digits docs in the base Range, but the above from Scale
can be read as implying it’ll happen. For now, that’s an open question.

https://bugzilla.gnome.org/show_bug.cgi?id=358970

9 years agocombobox: Work around popup handler altering model
Daniel Boles [Wed, 18 Jan 2017 22:22:52 +0000 (22:22 +0000)]
combobox: Work around popup handler altering model

GtkFileChooserButton installs a handler for the popped-up signal, which
refilters the menu, in order to hide the “(None)” item from the popup
if it was previously selected in the ComboBox. This oddity means that:

 • Until recently, this item would be selected in the menu shell, which
   would then be popped up and change the selection away from that item.
   This was therefore redundant (more on which below!) but benign.

 • After the patch for https://bugzilla.gnome.org/show_bug.cgi?id=771242
   however, this causes a critical assertion fail, as now we stash the
   originally selected item in a pointer so that it can be selected only
   after realisation/popup – but by that stage, the model has just been
   refiltered and the previous pointer no longer refers to a valid item.

This commit works around this problem by, after popping up the menu,
getting the active item again, in case a popped-up handler has gone and
invalidated the pointer to the active item that we saved before popup.

If a handler does this, everything done to find/use the original item is
pointless. But this avoids the ugly critical in FileChooserButton, while
not harming every other ComboBox that doesn’t mess with its model while
popping up (hopefully the vast majority), and it’s very difficult to
imagine a way to check if the active item is /going to/ be hidden later)

9 years agocombobox: Don’t select active item if it’s hidden
Daniel Boles [Wed, 18 Jan 2017 22:17:37 +0000 (22:17 +0000)]
combobox: Don’t select active item if it’s hidden

I hope no one ever actually brings such a silly item into this world,
but this achieves symmetry with the similar checks immediately after.

9 years agocombobox: Move variables into narrowest scopes
Daniel Boles [Wed, 18 Jan 2017 21:20:10 +0000 (21:20 +0000)]
combobox: Move variables into narrowest scopes

9 years agoUpdate Slovak translation
Dušan Kazik [Wed, 18 Jan 2017 20:50:55 +0000 (20:50 +0000)]
Update Slovak translation

9 years agovulkan: Fix copy/paste error
Benjamin Otte [Wed, 18 Jan 2017 15:46:04 +0000 (16:46 +0100)]
vulkan: Fix copy/paste error

9 years agovulkan: Add compiled shaders
Benjamin Otte [Wed, 18 Jan 2017 15:42:01 +0000 (16:42 +0100)]
vulkan: Add compiled shaders

I forgot to include them when adding the shaders.

9 years agovulkan: Compute right size for outset shadow
Benjamin Otte [Wed, 18 Jan 2017 15:39:27 +0000 (16:39 +0100)]
vulkan: Compute right size for outset shadow

We were emitting the box and not the shadow box as the rectangle. So we
didn't draw any shadows. Oops.

9 years agocss: Remove deprecated number-as-pixels compatibility
Benjamin Otte [Wed, 18 Jan 2017 03:12:15 +0000 (04:12 +0100)]
css: Remove deprecated number-as-pixels compatibility

Previously, for compatibility with GTK 3.0, we allowed specifying
numbers without units and interpreted them as pixels, even when the CSS
specification didn't.

Remove that now that we can break API.