Modeling focus change as an operation is awkward, because focus changes
logically happen during event processing, not layout. In particular, you
want to apply focus changes even if a widget is subsequently never laid
out.
Now that input.Source is concrete, it's much more straightforward to
offer focus changes as a command which can be queued through the
Source. A future change may similarly offer a command for directional
focus changes.
Signed-off-by: Elias Naur <mail@eliasnaur.com>
This change gets rid of the event.Queue interface by replacing it with
input.Source values. Source provides the interface to Router necessary
to implement interface widgets.
Signed-off-by: Elias Naur <mail@eliasnaur.com>
In the early days of Gio, FrameEvent was part of package app. It was
moved to package system to enable layout.NewContext be a convenient
short-hand for constructing a layout.
However, it seems the better design to leave FrameEvent (and Insets) in
package app, and move layout.NewContext there as well. More importantly,
the move allows us to replace the event.Queue interface with a concrete
type.
Signed-off-by: Elias Naur <mail@eliasnaur.com>
It was a design mistake to make profiling data available to programs.
Rather, profiling should either be a user-configurable debug overlay,
reported through runtime/trace, or both.
This change drops the io/profile package because we're about to overhaul
event routing.
Signed-off-by: Elias Naur <mail@eliasnaur.com>
This change removes the extra frame scheduled when events was delivered
during a frame. This extra frame was intended to paper over state changes
that happen later than the layout depending on it.
However, it is better for programs to never allow such state change skew,
and recent changes allows them to refresh and query state before layout.
This is an API change because programs may rely on the extra frames.
Those programs should ensure that state is updated before relying on it
in layout.
Signed-off-by: Elias Naur <mail@eliasnaur.com>
The double-negative DisabledOp is harder to understand than a
straightforward EnabledOp. Note that the absence of an EnabledOp
implies still means that the widget is enabled.
Signed-off-by: Elias Naur <mail@eliasnaur.com>
When storing a string in an interface value that escapes, Go has to heap
allocate space for the string header, as interface values can only store
pointers. In text-heavy applications, this can lead to hundreds of
allocations per frame due to semantic.LabelOp, the primary user of
string-typed references in ops.
Instead of allocating each string header individually, provide a slice
of strings to store string-typed references in, and store pointers into
this slice as the actual references. This only allocates when resizing
the slice's backing array, and averages out to no allocations, as the
backing array gets reused between calls to Ops.Reset.
We introduce two new functions, Write1String and Write2String, which
make use of this new slice for their last argument. We could've
automated this in the existing Write1 and Write2 methods, but that would
require type assertions on each call, and the vast majority of ops do
not make use of strings.
Signed-off-by: Dominik Honnef <dominik@honnef.co>
This commit updates the logic behind SemanticAt to use the same hit area
traversal as normal event routing, which should result in more accurate
results for screen readers trying to resolve widgets that might be partially
obscured by non-semantic content.
While here, I realized that the iteration of hit areas needed to stop at
the first matching semantic area, and I added that capability and updated
the ActionAt logic to leverage it as well.
Signed-off-by: Chris Waldon <christopher.waldon.dev@gmail.com>
When running ActionAt, the router used to only consider the topmost clip area, even
if that clip area had no input handlers attached whatsoever. This change updates the
logic for that test to use the same traversal as normal event handling, ensuring that
action inputs behave intuitively like any other pointer input area. Included is a test
catching the problematic behavior that prompted this change.
Signed-off-by: Chris Waldon <christopher.waldon.dev@gmail.com>
Putting a string in an interface value has to (normally) heap allocate
the string header and string. However, putting the address of a local
string variable in an interface value has the same effect, as this
causes the local variable to escape to the heap.
Signed-off-by: Dominik Honnef <dominik@honnef.co>
Now, Gio will send one system.StageEvent with system.StageInactive when
the window is not active. It is implemented on macOS and Windows.
This change is not fully backward compatible, if your code compares
the Stage (`stage < system.StageRunning`), you need to consider
the new system.StageInactive.
Signed-off-by: inkeliz <inkeliz@inkeliz.com>
Before this change, inverse transformations of pointer positions would
stack up, leading to incorrect positions when an enter or leave event
was delivered to multiple areas.
Signed-off-by: Dominik Honnef <dominik@honnef.co>
This reverts commit cd0c9dab9f. It turns out
that Enter/Leave is important for cancelling press-then-release-outside
for clickables.
Signed-off-by: Elias Naur <mail@eliasnaur.com>
When a key.InputOp is focused, a key.Event is matched to it and its ancestors.
If there is no focus, every handler is matched.
This change always matches to every handler, after checking the focus and
its ancestors.
Signed-off-by: Elias Naur <mail@eliasnaur.com>
The app.Window.Perform(ActionMove) is the wrong abstraction for
initiating a move gesture: Windows needs to know the move gesture
area at pointer move, and macOS needs to know the pointer button
down event that triggers the move gesture. This change replaces
Perform(ActionMove) with a new system.ActionInputOp that marks an
area movable.
Signed-off-by: Elias Naur <mail@eliasnaur.com>
Allowing clients to initiate resize gestures is a waste: macOS
doesn't support them, and the only reason we added them was to
implement client-side decorations for Wayland. Now all desktop
platforms implement resize gestures as needed, and we no longer
need the system.Action actions.
Signed-off-by: Elias Naur <mail@eliasnaur.com>
When a key.InputOp is focused, keypresses that it does not explicitly
include in its key set should check for ancestor clip areas that are
interested in them. Previously this check only included ancestors of
the final clip area in the hit tree, and could fail to find ancestors
of the focused key.InputOp because they were in a different branch.
This commit also adds a test to lock in the new behavior.
This can likely be made more efficient by adding a rapid way to map
from the focused key tag to its index in the hit tree. I wasn't sure
whether the complexity was warranted, but I'm happy to do that if
it's desired.
Signed-off-by: Chris Waldon <christopher.waldon.dev@gmail.com>
There are no public API that uses f32.Rectangle anymore. Move Rectangle
to an internal package for internal use.
Signed-off-by: Elias Naur <mail@eliasnaur.com>
The unit.Value is a struct and thus more inconvenient to use than its
underlying float32 type. In addition, most uses don't need a general
value, but rather a specific unit given by the context. This change
replaces unit.Value with two float32 units, Dp and Sp. It also changes
variables and parameters of unit.Value to a specific unit type matching
the context. That is, unit.Dp everywhere except for text sizes which are
in Sp.
Switching to typed float32s has multiple advantages
- They can be constants:
const touchSlop = unit.Dp(16)
- Casting untyped constants is no longer necessary:
insets := layout.UniformInset(16)
- Calculation with values is natural:
func (s ScrollbarStyle) Width() unit.Dp {
return s.Indicator.MinorWidth + s.Track.MinorPadding + s.Track.MinorPadding
}
The main API change is that calls to gtx.Px must be replaced with either
gtx.Dp or gtx.Sp depending on the unit.
Idea by Christophe Meessen.
Signed-off-by: Elias Naur <mail@eliasnaur.com>
Like the change to op.Offset before this, clip.RRect and UniformRRect
is usually used with integer coordinates. Change to integer coordinates
to eliminate many useless conversions to float32.
Signed-off-by: Elias Naur <mail@eliasnaur.com>
If the currently focused handler don't want the key event, try every
other handler, from top to bottom. This change requires widgets to
only react when focused.
Fixes: https://todo.sr.ht/~eliasnaur/gio/406
Signed-off-by: Elias Naur <mail@eliasnaur.com>
Instead of cmpletely replacing the IME snippet for every update, expand
the old range if there is overlap. This change avoids never-ending
restarts of the IME on Android where snippets are expanded in two
calls, one for expanding before the selection and one for exanding after
the selection.
Signed-off-by: Elias Naur <mail@eliasnaur.com>
Elias Naur