We've been dangling on the edge of unsafety, unnoticed, for a little while
about the reference count safety of ShellScreenshot. GJS owns the entire
reference count, so as soon as it goes out of scope it could die, causing
GJS to try and fetch the corresponding wrapper object for a stale pointer.
We haven't seen any crashes because of luck -- SpiderMonkey tries to group
together deallocations to limit GC pauses, and there isn't really a lot
of GC pressure in the duration that a screenshot happens, so we tend to
be mostly stable. But in the case that you create a lot of objects while
a screenshot is going on, by hammering the "Print Screen" button, for
example, you can destroy the GObject before the callback finishes.
https://bugzilla.gnome.org/show_bug.cgi?id=672775
StBoxLayout currently does not handle height-for-width children
correctly under some circumstances. As a work-around, hard-code
a label height of two lines of text, which should work for most
locales in the one place the widget is currently used.
https://bugzilla.gnome.org/show_bug.cgi?id=672543
We currently require users to tab away from the search entry before
search results can be navigated using arrow keys. For convenience,
support using arrow keys directly from the entry.
https://bugzilla.gnome.org/show_bug.cgi?id=663901
To avoid messing up St.Buttons' internal state with a pointer grab,
we wait for the pointer to leave the actor before starting the
drag operation manually. This works generally fine, but makes starting
a drag operation harder than necessary. To fix, enforce a reasonable
button state when starting the drag, rather than special-casing buttons
before the drag.
https://bugzilla.gnome.org/show_bug.cgi?id=637103
Currently compilation fails with -Werror, as we don't handle the
(newly introduced) smooth scroll events in switch statements; add
some basic support, which should make the compiler happy.
https://bugzilla.gnome.org/show_bug.cgi?id=672413
Currently, click and drop events are handled by each WorkspaceThumbnail
instance. With the introduction of the workspace cut and the request
to extend the reactive area of the workspace selector to the edge
of the monitor, it becomes more convenient to do all the event handling
inside ThumbnailsBox, even if this requires some manual layout computation.
https://bugzilla.gnome.org/show_bug.cgi?id=643319
Two small fixes which made _showNewStyleDialog() err out:
- g_key_file_load_from_data() expects a string as first
argument, but g_buffered_input_stream_peek_buffer()
returns an array of "data"
- g_key_file_load_from_data() is documented to allow -1 as
length parameter for \0-terminated strings, but the actual
type of the parameter is unsigned (d'uh)
https://bugzilla.gnome.org/show_bug.cgi?id=671556
Tweener uses a clutter timeline to manage all active animations
running at a given moment. The timeline is mopped up when no
animations are going any more.
Clutter requires timelines to have a finite duration, but since
animations can happen at any moment, no fixed duration can
accomodate the shell's needs.
To combat this problem, the tweener code picks a relatively
long duration: 1000 seconds. No string of animations should take
that long, so, in theory, that should be good enough.
Unfortunately, this tactic fails, in practice, when the user
suspends their machine, or VT switches. An animation can take
much longer than 1000 seconds (~16 minutes) to complete in those
cases. When the user resumes, or VT switches back the timeline
completes immediately (since it's already late) and tweener
never notices that the timeline stops ticking.
This commit changes the tweener timeline to automatically loop
back to 0 after completing, so that despite its fixed duration
property, it effectively never stops. Since the timeline loops,
its concept of elapsed time no longer increases monotonically,
so we now ignore it and track time ourselves with
GLib.get_monotonic_time().
This partially reverts commit
35764fa09e.
https://bugzilla.gnome.org/show_bug.cgi?id=653833
