Flag some actors that are good candidates for caching in texture memory
(what Clutter calls "offscreen redirect"), thereby mostly eliminating
their repaint overhead.
This isn't exactly groundbreaking, it's how you're meant to use
OpenGL in the first place. But the difficulty is in the design of
Clutter which has some peculiarities making universal caching
inefficient at the moment:
* Repainting an offscreen actor is measurably slower than repainting
the same actor if it was uncached. But only by less than 100%,
so if an actor can avoid changing every frame then caching is usually
more efficient over that timeframe.
* The cached painting from a container typically includes its children,
so you can't cache containers whose children are usually animating at
full frame rate. That results in a performance loss.
This could be remedied in future by Clutter explicitly separating a
container's background painting from its child painting and always
caching the background (as StWidget tries to in some cases already).
So this commit selects just a few areas where caching has been verified
to be beneficial, and many use cases now see their CPU usage halved:
One small window active...... 10% -> 7% (-30%)
...under a panel menu........ 23% -> 9% (-61%)
One maximized window active.. 12% -> 9% (-25%)
...under a panel menu........ 23% -> 11% (-52%)
...under a shell dialog...... 22% -> 12% (-45%)
...in activities overview.... 32% -> 17% (-47%)
(on an i7-7700)
Also a couple of bugs are fixed by this:
https://bugzilla.gnome.org/show_bug.cgi?id=792634https://bugzilla.gnome.org/show_bug.cgi?id=792633
When not using arrow notation with anonymous functions, we use Lang.bind()
to bind `this` to named callbacks. However since ES5, this functionality
is already provided by Function.prototype.bind() - in fact, Lang.bind()
itself uses it when no extra arguments are specified. Just use the built-in
function directly where possible, and use arrow notation in the few places
where we pass additional arguments.
https://gitlab.gnome.org/GNOME/gnome-shell/merge_requests/23
If a source actor is set, use that for determining the arrow side (i.e.
whether the BoxPointer widget should expand in a certain direction).
This is better because it ensures that the popup is displayed on the
same monitor as the widget it originates from.
Without this, entering text with a vertically aligned input method
close to the bottom of a monitor would expand the BoxPointer downwards
on the monitor beneath it, instead of upwards, which is what one would
expect.
https://bugzilla.gnome.org/show_bug.cgi?id=786886
Any symbols (including class properties) that should be visible
outside the module it's defined in need to be defined as global.
For now gjs still allows the access for 'const', but get rid of
the warnings spill now by changing it.
https://bugzilla.gnome.org/show_bug.cgi?id=785084
This is done for properly drawing popup menu when arrow rise is 0 (in
case of background menu).
Normally, the menu with arrow rise set to 0 is drawn properly having
all four corners rounded. But when the source(click/arrowOrigin) is
near screen's edges, one of the corners (depending on source's position
and arrow alignment) is drawn right angled.
This happens because the rounded corners are skipped and right angled
arrow is drawn when arrow origin is close to the edges.(That's why when
arrow-rise is 0, it forms right angled corner).
So, a few condition checks are made to ensure that right angled corner
is not drawn.
https://bugzilla.gnome.org/show_bug.cgi?id=699608
You would think we would already do something like this, but apparently
lots of code was calling hide() without checking if the box pointer was
already visible, causing it to queue a full tween. The biggest win was
with ibusCandidatePopup.js, which called hide() on every DBus message.
This increases the performance for me to enter the overview by a tiny
bit. The remaining time is spent updating the frequent apps / all apps
display.
https://bugzilla.gnome.org/show_bug.cgi?id=712727
We must reduce the forWidth in the call to get_preferred_height()
with the border width, otherwise we might request a smaller height
that we actually need and overflow.
https://bugzilla.gnome.org/show_bug.cgi?id=696564
We need to adjust the offset of close buttons, in case the box
pointer has the arrow at the top. To do so, extend close buttons
to hook into a boxpointer (since that's the common use for them)
and automatically adjust their position.
https://bugzilla.gnome.org/show_bug.cgi?id=707842
The popup of the FolderView is now contained inside
the parent view, solving the overflow of apps with a ScrollView.
Also, solved a lot of bugs in popup/FolderView calculation
of position and size.
https://bugzilla.gnome.org/show_bug.cgi?id=706081
Commit d6cace32 introduced a typo in the left/right arrow side
calculation code that causes in most scenarios (where the monitor
width is greater then the height) to not flip the box when it doesn't
fit inside the monitor.
https://bugzilla.gnome.org/show_bug.cgi?id=703403
The arrow origin must be in the coordinate space of the box pointer,
but it cannot be calculated when the box pointer is shown, because it's
not allocated yet, so pass the source to the boxpointer and adjust
the arrow at the next paint.
https://bugzilla.gnome.org/show_bug.cgi?id=694264
Due to limitations and bugs in SpiderMonkey's GC, wrapper objects
for cairo contexts and similar may not get cleaned up immediately
after repainting, leading to leaking memory. Explicitly disposing
of such objects after they're not needed can clean up large portions
of memory for cairo surfaces.
https://bugzilla.gnome.org/show_bug.cgi?id=685513
As we may be flipping the box pointer in response to re-allocation,
like the addition of a new actor to the boxpointer, we can't queue
a re-layout while in a re-layout, so defer.
https://bugzilla.gnome.org/show_bug.cgi?id=690608
Depending on the current state of arrowSide, the box allocation may be
wrong; e.g. if the user requested a TOP, but we flipped to a BOTTOM, the
next request would look to the y2 value of the flipped BOTTOM, which is
wrong.
Instead, use the origin, plus the calculated preferred size of the box.
https://bugzilla.gnome.org/show_bug.cgi?id=690608
Make sure we re-allocate after we flip sides, to ensure that
padding around the child actor is updated correctly. Additionally,
ensure that we flip after we setPosition, as we won't get re-allocated
auotmatically by just changing the position.
https://bugzilla.gnome.org/show_bug.cgi?id=690608
Given that it's modifying state, there's no reason it can't be
using it instead. This makes it easier to use, so we aren't passing
in a bunch of instance variables every time.
https://bugzilla.gnome.org/show_bug.cgi?id=690608
Currently close() is a no-op when the menu has already been closed.
However, repeated calls could pass different animation parameters.
For instance in the user menu, we try to hide the menu immediately
before locking the screen, to avoid the popup jumping across the
screen while fading out - as we do this from the corresponding
item's activate handler, the closing is still animated if the menu's
own handler (which requests a full animation) is run first.
Fix this by changing close() to overwrite ongoing animations before
bailing out early.
https://bugzilla.gnome.org/show_bug.cgi?id=686484
If the arrow's origin is so close to the edge that the arrow will not
be isosceles, we try to compensate as follows:
- We skip the rounded corner and settle for a right angled arrow as
as shown below.
|\_____
|
|
- If the arrow was going to be acute angled, we move the position of
the box to maintain the arrow's accuracy.
https://bugzilla.gnome.org/show_bug.cgi?id=680077
This makes sense if we want to follow what the following comment says:
// We also want to keep it onscreen, and separated from the
// edge by the same distance as the main part of the box is
// separated from its sourceActor
Using a magic number violates the "separated from the edge ...
separated from its sourceActor" part.
https://bugzilla.gnome.org/show_bug.cgi?id=682343
The hide animation causes hover notifications for the actors inside the
boxpointer. PopupBaseMenuItems, in particular, grab the keyboard focus on
hover notifications to enable keyboard navigation on menus. This, in turn,
breaks modal dialogs' keyboard navigation since key focus is taken away from a
just created dialog when the menu is hiding.
Since input events aren't useful while menus are animating we just prevent
them from propagating.
https://bugzilla.gnome.org/show_bug.cgi?id=662493
The last patch in the sequence. Every place that was previously
setting prototype has been ported to Lang.Class, to make code more
concise and allow for better toString().
https://bugzilla.gnome.org/show_bug.cgi?id=664436
Currently BoxPointer/Menus always point to the center of the
associated source actor. This is generally what we want, but
add some API to adjust that behavior for the cases where it
isn't.
https://bugzilla.gnome.org/show_bug.cgi?id=659274
js2-mode is no longer developed and we recommend js-mode these days,
so switch the modelines to specify that, and make them consistent
across all files.
https://bugzilla.gnome.org/show_bug.cgi?id=660358
A boxPointer should be able to be attached to any actor, not just ones on the
primary monitor. Assume that the sourceActor doesn't straddle monitors, and
constrain the boxPointer to the monitor the sourceActor is on.
https://bugzilla.gnome.org/show_bug.cgi?id=659861
The specs call for a 2 pixel gap between the panel and its menus,
though we need to specify this as 4 pixels, since it's relative to the
bottom of the icon/title, not the bottom of the panel (up until now,
the point of the menu arrow was actually overlapping the menu's
highlight underline).
Also, move the gap specification into the CSS, since it makes more
sense there.
https://bugzilla.gnome.org/show_bug.cgi?id=655627
Remove ShellGlobal's monitor-related methods, and have
Main.layoutManager provide that information instead. Move
Main._relayout() to LayoutManager, and have other objects connect to
the layout manager's 'monitors-changed' signal to know when the screen
geometry has changed.
https://bugzilla.gnome.org/show_bug.cgi?id=636963
Instead of setting the x/y position of the box pointer, which results
in a long change of workarounds for limitations of the Clutter
layout system, set the anchor point instead, which takes the
positioning out of the layout system.
The position is computed as a combination of the position computed
from the allocation and the box pointer's size, and an offset that
we tween when animating showing and hiding the box pointer.
https://bugzilla.gnome.org/show_bug.cgi?id=645744
Using ClutterActor.get_transformed_size() can produce bugs if we
happen to position the box pointer when the source actor has a
relayout queued. Use our newly added reliable utility function
instead.
https://bugzilla.gnome.org/show_bug.cgi?id=645744
If the BoxPointer changes size (eg, when opening the "More" section of
the network menu), reposition it to make sure it's still aligned
correctly and still completely on-screen.
This is not the right fix for this problem (and causes the menu to be
drawn in the wrong position for one frame). The right fix would
involve a ClutterConstraint, but that would be more invasive, and can
happen post-3.0.0.
https://bugzilla.gnome.org/show_bug.cgi?id=645647
Point the arrow to the center of the sourceActor's content box, rather
than its allocation, in case it has asymmetric padding (as the
rightmost message tray summary item does).
https://bugzilla.gnome.org/show_bug.cgi?id=641728
Don't do the "slide" effect when moving from one menu to another; only
do it when opening the first menu, or closing a menu without opening
another one.
https://bugzilla.gnome.org/show_bug.cgi?id=634755
In some circumstances, a boxpointer would draw itself with the arrow
partially overlapping the rounded corner, causing things to not line
up correctly. Don't do that.
And while we're at it, don't draw the pointer very very close to the
corner either, since it looks odd if the corner flows directly into
the arrow.
https://bugzilla.gnome.org/show_bug.cgi?id=635393
Although within St itself there are situations where the semantics of
these functions (return TRUE or FALSE and return the actual value in
an out parameter) is useful, it's mostly just annoying at the
application level, where you generally know that the CSS property is
going to specified, and there is no especially sane fallback if it's
not.
So rename the current methods to lookup_color, lookup_double, and
lookup_length, and add new get_color, get_double, and get_length
methods that don't take an "inherit" parameter, and return their
values directly. (Well, except for get_color, due to the lack of (out
caller-allocates) in gjs.)
And update the code to use either the old or new methods as appropriate.
https://bugzilla.gnome.org/show_bug.cgi?id=632590
