The only reason for `vfunc_paint` and `vfunc_pick` existing was to
implement a culling optimization. Although that optimization actually
made performance worse than none at all because it forced the painting
and picking cycles to spend more time calling into JavaScript.
Turns out we don't have to choose between native code and culling though.
Just reimplement the culling using native ClutterActor functions and we
get the benefits of both.
Performance on an i7-7700:
Moving the cursor over the icon grid:
Before: 70% CPU, 5.5ms per frame
After : 60% CPU, 4.5ms per frame
Scrolling the icon grid:
Before: 60% CPU, 4.4ms per frame
After : 50% CPU, 3.3ms per frame
Helps with https://gitlab.gnome.org/GNOME/gnome-shell/issues/174
The `reactive` property of icon actors was being restored multiple times
over the course of the pulse animation, all at slightly different times
as each icon finished animating at different times.
The problem is that toggling `reactive` on an `StWidget` incurs a style
change of the `insensitive` pseudo class, and style changes would quickly
queue relayouts incurring full stage reallocation. This occurred many times
during a pulse animation, limiting its smoothness and performance.
The solution is to not toggle the `reactive` property in the pulse
animation at all, which avoids incurring multiple full stage relayouts.
As a bonus, this means the icon under the cursor pulses with the correct
selection highlight, appearing more seamless and responsive.
https://gitlab.gnome.org/GNOME/gnome-shell/merge_requests/261
The `reactive` property of icon actors was being restored 24 times over
the course of the spring animation, all at slightly different times as
each icon finished animating at different times.
The problem is that toggling `reactive` on an `StWidget` incurs a style
change of the `insensitive` pseudo class, and style changes would quickly
queue relayouts incurring full stage reallocation. This occurred many times
during a spring animation hogging the CPU and limiting the frame rate.
The solution is defer and batch the cleanup for all icons until after the
last icon has finished animating. This way the CPU impact of the style
change and stage relayout isn't felt during the animation so the frame
rate remains higher and smoother. The overall CPU usage of the animation
is also reduced as the remaining relayouts are much more likely to be
grouped into a single frame.
Icon spring animation performance on an i7-7700:
Before: 83% CPU and 47 FPS
After : 78% CPU and 54 FPS
which is about a 22% increase in performance per clock (FPS/CPU).
https://gitlab.gnome.org/GNOME/gnome-shell/merge_requests/253
Removing Shell.GenericContainer from the IconGrid class was
challenging because it needs the "skip paint" API from it.
This API was added, too, as a workaround to the inability
to override vfuncs from GJS.
The overrides are largely copy-pasted and translated versions
of the Shell.GenericContainer code.
The IconGrid:key-focus-in signal was renamed to :child-focused
to avoid clashing with ClutterActor:key-focus-in.
In GridSearchResults, the internal IconGrid had it's y_expand
set to false, so it doesn't push other search elements (the
list results mainly) to the bottom of the screen.
Because skip paint wasn't and still isn't a GObject property,
rename it to _skipPaint to reflect that.
https://gitlab.gnome.org/GNOME/gnome-shell/merge_requests/153
As part of our quest to obsolete Shell.GenericContainer, IconGrid will
become a Clutter.Actor subclass. As the ::key-focus-in signal would
clash with Clutter.Actor::key-focus-in, rename it to ::child-focused.
https://gitlab.gnome.org/GNOME/gnome-shell/merge_requests/153
Pretty much like the previous patches, this extends St.Bin. The
most interesting aspect of this patch is that most of the sizing
routines of the icons is now delegated to the actors and layout
managers, removing quite a bunch of code.
The 'spacing' theme property is now redirected to StBoxLayout's
spacing property. Also adjust the Dash code to stop forcing a
potentially invalid width in the first icon too.
https://gitlab.gnome.org/GNOME/gnome-shell/merge_requests/153
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
We now cancel animations on override, however we also want to cancel
animations altogether on unmap (that is, when hiding the overview)
to avoid icons swarming into the void.
https://bugzilla.gnome.org/show_bug.cgi?id=736148
Until now we were waiting the animation to complete to allow the user to
make a new animation. This could bring some problems and annoy nervous
users.
Instead of that, destroy clones on new animations triggers and
create a new animation with the new direction.
https://bugzilla.gnome.org/show_bug.cgi?id=736148
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
Currently we use the same amount of total delay divided by all items,
but that makes the items animate slow if the amount of items is small.
To avoid that, use a smaller total amount delay for an amount of items
smaller than four.
https://bugzilla.gnome.org/show_bug.cgi?id=737017
The spring animation has to use clones to escape any clip set on the
parent, as it mainly occurs outside the parent. The same does not apply
to the pulse animation, which is very much in place - in fact, if the
parent is clipped (for instance a scrolled app folder), not having the
clip applied to the animated icons is indeed wrong.
Just animate the original grid actors instead, which gives the expected
result.
https://bugzilla.gnome.org/show_bug.cgi?id=736885
Commit 9e5704b introduced a regression due to a sloppy mistake not
making actors reactive on animation out.
Fix that making actors reactive when animating out as well.
Currently we consider the last item animating being the last actor of
the list, but that's wrong since the last item animating depends on the
distance between the source actor and the original position of the
animated actor.
Take that into account to check the last item of the animation.
Although the actor are with opacity 0, they were reactive, so the if the
user quickly make a click on the grid while animating it can launch an
app accidentally.
To avoid that make the actors non reactive while animating.
Following design mockups, animate the icons on AllView, FrequentView,
Dash and Search to zoom out when opening a new window of the app or when
the app is not running and the user execute it.
https://bugzilla.gnome.org/show_bug.cgi?id=734726
Following design decision, we want to animate AllView and FrequentView
when opening and closing with a swarm spring form.
This involves a few changes needed to allow that, since from some time
now, we are animating page changes in viewSelector, using only a fade
transition. However now we want to let appDisplay and iconGrid apply its
own animation.
For that we special case the change to and from apps page on
viewSelector to let appDisplay to animate its own items, using and API
on appDisplay which at the same time uses an API on iconGrid.
Thanks Florian Müllner for the debugging work
https://bugzilla.gnome.org/show_bug.cgi?id=734726
Add a new animation to folder view based on designers mockups that
emulates pulsating icons.
The code on iconGrid is though to work well for the upcoming patches to
animate AllView and FrequentView.
https://bugzilla.gnome.org/show_bug.cgi?id=734726
The scale passed to _updateChildrenScale is exclusively computed from
properties, so it can be moved into the function itself - as the scale
now becomes a mere detail, rename to a more appropriate _updateIconSizes
at the same time.
Currently, our logic for page panning isn't great. If the user starts a
pan upwards and hesitates over a new page, we'll go to the *next* page
on release, since the difference is greater, but the velocity wound down
to 0.
Instead of trying to treat it like page down or scrolls, simply do the
math to find the page where the user scrolled to.
This is unfortunately broken for fast swipes, since the user doesn't get
far enough into the new page to make a difference. I'm getting the
impression we'll need a gesture recognizer for this, though, however
crude. Simple hacks I tried, like a velocity multiplier, didn't work
properly.
https://bugzilla.gnome.org/show_bug.cgi?id=729064
When we create a result actor, cache it, so it can be used for
subsearches of the same initial. For now, to keep memory usage
and the stage graph relatively clean, don't persist the actors
across searches, but maybe we should do this in the future.
This also means that we don't query getResultMetas for items
that we've seen in the same initial search.
https://bugzilla.gnome.org/show_bug.cgi?id=704912
While this is good style anyway, after the latest appDisplay changes
the first call to get_preferred_height() happens before we properly
compute those properties, resulting in a size request of NaN that
triggers a Clutter warning.
Similar to adapting the spacing dynamically to the available
space we already do, scale down icon sizes if the grid is too
small to fit the requested minimum number of rows/columns.
https://bugzilla.gnome.org/show_bug.cgi?id=706081
IconGrid has never really been a general purpose container, but has
always been used in conjunction with BaseIcon. IconGrid will soon
gain the ability to adjust the item size dynamically to adapt to the
available space, which will require that we can make some more
assumptions about the items added to the grid (namely: we need
access to BaseIcon's setIconSize() method).
So change addItem() to take an object instead, which should have
an actor and a (BaseIcon) icon property.
Based on a patch by Carlos Soriano.
https://bugzilla.gnome.org/show_bug.cgi?id=706081
Add methods to open/close extra space for n rows. The app picker
will use those to make AppFolder popups appear inline with the
main grid rather than on top of it.
https://bugzilla.gnome.org/show_bug.cgi?id=706081
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
Add a property to also add the calculated spacing
around the grid.
This will allow FolderView to be aligned with the
main grid without cutting off any of the surrounding
boxPointer decorations or the close button
https://bugzilla.gnome.org/show_bug.cgi?id=706081
When we adapt the grid to different display sizes,
we don't want the number of displayed items to get
too small. In the future we will scale down icons to
make sure that the grid fits add least minRows
x minColumns items, but for now we only take the
properties into account when calculating the dynamic spacing.
https://bugzilla.gnome.org/show_bug.cgi?id=706081
Organize applications in AllView by pages using the new PaginatedIconGrid
added previously. Pagination is generally a better pattern for collections
than scrolling, as it better suits spacial memory.
Hook into AppDisplay's allocation function to communicate the available
size to the different views before child allocations - this is only
required by the paginated view (as pages must be computed before
calling get_preferred_height/get_preferred_width), but doing it for
all views will guarantee that their dynamic spacing calculation is
based on the same values.
https://bugzilla.gnome.org/show_bug.cgi?id=706081
Since the parameter of the function is the width, reflect that in
the function name. Also, since we are counting columns, not only
children for each row, reflect that in the function name also.
https://bugzilla.gnome.org/show_bug.cgi?id=706081
The new PaginatedIconGrid class acts as a container for pages.
So the new class provides the container behaviour and some
useful functions like positions of pages, number of pages, etc.
But, it doesn't add indicators of the pages and doesn't manage
the scroll of the pages, neither any management of the pages
like in which page currently it is, etc.
https://bugzilla.gnome.org/show_bug.cgi?id=706081
Recreating icons on every style change -- like hover, can have
disasterous effects. Not only is the quick creation/destruction of
the actors bad, but adding/removing actors at runtime queues many
relayouts, which makes the whole system slower as a lot of unnecessary
reallocations are figured out.
While an optimization was here before, it was broken because it
broke high-contrast themes. Connect explicitly to the texture cache
to know when the icon theme has changed, instead of removing a valuable
optimization.
https://bugzilla.gnome.org/show_bug.cgi?id=672941
We explicitly include NoDisplay applications in the ShellAppSystem because
we want app tracking for them, but we explicitly filter NoDisplay applications
out when showing them to the user because we don't want to show them to the
user. We also based our "All" apps view on a flattened list of apps. While
we did check for NoDisplay on the app item itself, we didn't check against
its parents. Refactor the app display view to not use a separate flat list
of applications, but instead a concatenation of all the applications in all
the loaded categories.
https://bugzilla.gnome.org/show_bug.cgi?id=658176
clutter_actor_get_children requires making a temporary GSList from
a linked list structure, and then creating a JS Array from that GSList.
For simple cases like the number of children, use clutter_actor_get_n_children.
https://bugzilla.gnome.org/show_bug.cgi?id=677426
When displaying children, we need to make sure that we use the rowLimit
property that we pass to iconGrid, rather than assuming that it's
always MAX_SEARCH_RESULTS_ROWS in case some subclass (in an extension maybe)
wants to do something different.
https://bugzilla.gnome.org/show_bug.cgi?id=675527
Commit 26580f8f reintroduced an optimization on style changes to avoid
creating icons unconditionally. As this breaks icon theme changes (for
instance when toggling "High Contrast" in the universal access menu),
remove it again.
https://bugzilla.gnome.org/show_bug.cgi?id=672941
IconGrid items used to be square, so a single size value made
perfect sense. However, as contact search uses rectangular items,
using a single size is wrong - the allocated height ends up twice
the size of the visible height, which is particularly visible if
another provider displays results below contact results.
https://bugzilla.gnome.org/show_bug.cgi?id=670148
If both spacing and -shell-grid-item-size are 0, as they would be with nothing
setting them, we enter an infinite loop where we try to compute the layout.
Avoid the situation entirely by defaulting -shell-grid-item-size to a sane
value instead of 0.
https://bugzilla.gnome.org/show_bug.cgi?id=662747
Forcing the icon size will distort it unnecessarily, and will
in any case not work if showing an animation (which is a ClutterGroup).
Instead, set the size on the bin, and make it align its child
if needed.
https://bugzilla.gnome.org/show_bug.cgi?id=666606
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
All classes that have at least one other derived class (and thus
benefit from the framework) have been now ported. These includes
NMDevice, SearchProvider, AltTab.SwitcherList, and some other
stuff around.
https://bugzilla.gnome.org/show_bug.cgi?id=664436
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
Currently the icon texture is only updated on style changes when
the icon size is set from CSS and differs from the previously used
icon size.
As the style change may have been triggered by an icon theme change,
textures that are created for themed icons should always be recreated;
given that this is the case for most uses (with the exception of
file thumbnails), recreate the icon texture unconditionally to avoid
complexity.
https://bugzilla.gnome.org/show_bug.cgi?id=643738
Mainly due to StTextureCache's way of handling fallback icons, an
implementation of BaseIcon.createIcon() may return an icon smaller
than the requested size. Given that BaseIcon is not used for isolated
elements, but rather for groups of related items (App view, Dash,
Search Results, ...), having some elements end up with the wrong
size is more annoying than having some elements end up ugly due to
scaling, so explicitly enforce the requested icon size.
https://bugzilla.gnome.org/show_bug.cgi?id=642043
- Center the icon texture in the area allocated for it, and always give
the nominal size - avoid off-by-one scaling if the parent allocated
a little less or more size than we wanted.
- Use Math.floor() when centering horizontally to avoid allocation
at a half pixel.
https://bugzilla.gnome.org/show_bug.cgi?id=642124
Currently there is a serious problem with ellipsization in various
parts of the overview. While wrapping the label or giving it more
space may be appropriate approaches for the application view, neither
works very well for the dash - possibly the best option there is to
not show the label at all.
So add a constructor parameter to BaseIcon to allow hiding the
label.
https://bugzilla.gnome.org/show_bug.cgi?id=636156
As all children were considered for the grid's layout, hidden items showed up as
empty space. Instead, exclude hidden children from the layout, so that the grid is
only made up of visible items.
https://bugzilla.gnome.org/show_bug.cgi?id=631537
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
Both current design and new mockups use square icons. Currently this
is implemented by setting a fixed width and height in the CSS, which
does not work for the growing/shrinking items in the new sidebar.
By moving the assumption of square items into the code, the size will
adjust according to icon size, spacing, padding and font size if not
explicitly overwritten in the CSS.
https://bugzilla.gnome.org/show_bug.cgi?id=625887
The current design features a sidebar where the icon size shrinks
automatically to fit more items. Add a setSizeManually parameter
to BaseIcon, which disables sizes read from CSS and allows to use
BaseIcon.setIconSize() to support this.
https://bugzilla.gnome.org/show_bug.cgi?id=625887
All mockups now use a representation for documents/places similar to
the one used for applications. Rename AppIcon to BaseIcon and move its
code together with WellGrid out of appDisplay to stress their general
usefulness.
https://bugzilla.gnome.org/show_bug.cgi?id=625887
