All GPU APIs except OpenGL ES 2 can generate mipmaps for textures.
This trades 33% more GPU memory use for improved rendering quality
and speed for downscaled images.
Signed-off-by: Elias Naur <mail@eliasnaur.com>
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 that change, Gio could crash when the WebGL context was lost
unexpectedly. Now, Gio will properly handle such situation and
recreate the buffers/resources when context is restored and will
wait until context is recovered.
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 commit switches the priority of EGL and Vulkan so that EGL is always
tried first. This is because our EGL backend performs significantly better
than the Vulkan one, and we want the most performant experience to be the
default.
Our hypothesis is that the EGL backend is benefitting from lots of optimization
within the OpenGL driver that Vulkan drivers expect applications to perform
themselves. Rather than invest a bunch of time optimizing the Vulkan backend
right now, this change lets us focus on other priorities.
Signed-off-by: Chris Waldon <christopher.waldon.dev@gmail.com>
Bump golang/x/exp/shiny to use the specific module,
instead of using a single all-encompassing module.
This significantly reduces the number of packages in go.sum.
Unfortunately, this requires us to bump the go.mod language
version to 1.18, otherwise using `go mod tidy` and similar
tools will complain about incompatibility with module lookup
in Go 1.16. The code should still compile with 1.17.
Bump github.com/gioui/uax, which gets rid one additional unneeded
dependency and should also reduce the binary size.
Signed-off-by: Egon Elbre <egonelbre@gmail.com>
This commit adds methods to widget.Scrollbar that enable consuming
code to check if the scroll indicator is processing a drag gesture
or if the scroll track is currently being hovered. These accessors
enable scrollbar style types to have enough information to hide the
scroll indicator when it isn't needed, whereas currently they cannot
differentiate between a scrollbar indicator that is being dragged
but hasn't moved since the last frame and a scrollbar indicator that
is not being dragged.
Signed-off-by: Chris Waldon <christopher.waldon.dev@gmail.com>
Window.decorations.height is supposed to be a constant during the
lifetime of the window, unlike w.decorations.Config.decoHeight that
varies depending on the decorations state (fallback or custom).
This change makes that so, fixing a problem where the fallback
decorations would fail to offset client content after a maximize
or minimize.
Signed-off-by: Elias Naur <mail@eliasnaur.com>
Before this change, the middle alignment would align according to
the widest child. This change aligns according to the widest child
or minimum constraint, whichever is largest.
Signed-off-by: Elias Naur <mail@eliasnaur.com>
Before this change, an IME text edit would always have its newlines
replaced with spaces. However, for Editors where Submit is enabled
we want newlines to result in SubmitEvents.
Signed-off-by: Elias Naur <mail@eliasnaur.com>
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>
Before that change, on Android, was impossible to overlay GioView with
a custom view. This change adds FrameLayout and renders GioView into
that, allowing to use addView from Android API.
Fixes: https://todo.sr.ht/~eliasnaur/gio/427
Signed-off-by: Inkeliz <inkeliz@inkeliz.com>
The Nix version of the macOS toolchain has difficulties compiling
Objective-C modules; disable modules instead of figuring out why.
It also doesn't include any frameworks automatically; add them explicitly.
While here, move suppression of OpenGL deprecation to a GL-specific
file.
Signed-off-by: Elias Naur <mail@eliasnaur.com>
We used to track the pressed pointer buttons through the global function
[NSEvent pressedMouseButtons]. However, it's possible that at the time a pointer
press event is delivered, the pointer button is up again. To ensure a consistent
view of the pointer press state, track it through the buttonNumber property on
delivered events.
Signed-off-by: Elias Naur <mail@eliasnaur.com>
Previously, Gio doesn't reclaim the focus when they lose that to a
parent window. In such a case, the child window can steal
keyboard focus, and Gio will never recover it.
Now, Gio will recover the focus when clicked.
Signed-off-by: Inkeliz <inkeliz@inkeliz.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>
Pass through a fallback window decoration height to the Wayland backend,
so that it can account for it when determining surface size.
Fixes: https://todo.sr.ht/~eliasnaur/gio/435
Signed-off-by: Elias Naur <mail@eliasnaur.com>
When a window is destroyed, it is no longer valid to call its wakeup
method.
Thanks to Jack Mordaunt for identifying the race.
Signed-off-by: Elias Naur <mail@eliasnaur.com>
This commit adds a simple linear-history undo/redo mechanism to
widget.Editor bound to Short-(Shift)-Z as well as tests for this
new feature.
Notes on the implementation:
- using a slice to hold the history does mean that we incur
allocations as the user types, but I hope that the Go slice
growth heuristic means that the number of times we pay this
penalty is very small. We also never shrink the slice in this
implementation, which ensures that undoing work and then making
additional modifications is very efficient, but could be framed
as a memory leak.
- this implementation creates a new history element every time
we call replace(). This means that, on desktop, it's essentially
one per rune of input. Users likely want to be able to undo larger
units of change, so a future improvement could be to coalesce
changes so long as the selection doesn't change between them.
- I think it's possible to store only one of the Apply/Reverse
change contents in the history slice, but it's significantly
more complicated. To implement this, you'd need to add a field
indicating if the modification represented a forward or backward
change, and then rewrite the modification's content as you performed
undo/redo operations.For the time being, I'm not sure it's worth
this complexity.
- Future work could introduce a limit to the number of history
entries stored. If we did this, we should also change the
data structure for storing history. Enforcing such a limit
using a simple slice like this would be extremely inefficient.
Perhaps a ring buffer or a linked list would make more sense?
- Applications will likely want to be able to manipulate undo
history in the future. We may wish to export undo() and redo()
from the editor. Applications will also likely want a mechanism
to save the undo history to disk and restore it (implementing
persistent undo). I'm not sure what the most suitable API for
that is yet, so I decided not to try to tackle it yet.
Signed-off-by: Chris Waldon <christopher.waldon.dev@gmail.com>
By keeping all the information in a single map, we avoid multiple
lookups and can switch between frames more easily.
Before ~35us, after ~20us for adding 50 new+old and switching
the frame.
Signed-off-by: Egon Elbre <egonelbre@gmail.com>
Using clean struct creation creates a lot of temporary variables in
assembly. Inline the assignments, which generates less code.
Signed-off-by: Egon Elbre <egonelbre@gmail.com>
Size and NumRefs are always used together, so consolidate info to
a single table to avoid two separate lookups.
Signed-off-by: Egon Elbre <egonelbre@gmail.com>
splitTransform func was creating multiple copies of f32.Affine2D
due to not having access to the internal and passing around non-pointer.
This makes splitTransform from ~8ns to .Split ~2ns.
It seems that the non-pointer receiver was ~0.6ns slower in this case.
Signed-off-by: Egon Elbre <egonelbre@gmail.com>
Elias Naur