napi and esbuild are not just dev dependencies. Since we distribute source packages, they are in fact
run-time dependencies, as
"npm install" will run "npm run build" and that
requires those two.
If a branch that always return has a "void value" and the side that
doesn't return has a value, we need to synthetize a default value
so the struct is complete, even if that value is not used.
This makes it possible to fall back to software rendering.
This change also removes the use of the Qt backend on Windows and macOS, since
the native style on those platforms
resolves to fluent/cupertino and doesn't require Qt anymore.
This fixes a couple of bug:
- Bug #3740 happens because `Repeater::model_set_row_data` did not use
the inner.offset to get the instance (So that's the `val == 106` part
of the test)
- But I went ahead and also tested what happenned if you changed the
model from the "outside" using the `model[i] = `, and that was not
implemented, hence the move of the code from
`Repeater::model_set_row_data` to `RepeaterTracker::row_changed`,
That does need Pin though, so Pin was added everywhere
- C++ is not affected by bug #3740, because because the C++ listview
don't do the "allocate only visible" optimization. But the
"val == 1106" part of the test would fail so this patch also moces
the update from `model_set_row_data` to `row_changed`.
But following that we don't set the state as Dirty in `row_changed`,
the write_to_model test started filling because the `row_added`
function was missing an update of the index on every further items
The change in the interpreter prevent a borrow_mut from causing trouble.
The Value really don't need to be in a RefCell anyway
Fix#3740
The rules require comparing if tiles are equal, which used to be a string comparison as we just
converted the image to a path. With ImageData this doesn't
work anymore, so this patch proposes an optional path
property that makes the code also a tad bit more readable.
If you have a window like so:
```
component W inherits Window {
width: 200px; // or some other bindings
}
```
Before this patch, it will be converted by the compiler to something like
```
component W inherits Window {
width: 200px; // or some other bindings
min-width: width; // (not actual property, but part of the layout_info)
max-width: width;
}
```
When the window is on the screen, the platform backend will set the max
with and min width on the window manager window to the value from the
layout info.
But slint will also set the width and the height of the WindowItem to
the actual value. This will break the binding for width if any, and
will also cause the min and max with do be updated, which is wrong.
We haven't had much problem with that before, but with the
ComponentContainer, this becomes a problem as we want to set the width
and height of the inner from the outer by adding a two way binding,
which cause a binding loop at runtime.
The behavior change is that if you have a fixed window size and use that
on a MCU or platform that has a different size, the window will be
cropped or padded but will no longer be resized
Share tsconfig.json so that ava runs the typescript compiler also on the tests.
Similary, the syntax check now also covers the tests.
Sadly, the exception that the Slint compiler throws is not type declared,
so we have to use any.
Similarly, the object share returned by loadFile() is unknown
in terms of properties, callbacks, etc.
Saves 1M in the binary
We use the browser to render the images anyway
Note: i-slint-core still bring image with png and jpeg codecs, as well
as svg deps. We just need the image-decoders feature to load the builtin
images, of these there are only svg.
In the ci and on workstations the steps are the same:
1. Install (without building, which would build release)
2. Build debug
3. Run tests
Previously only on Windows the tests were run in debug, because
the "npm install" step in the nodejs test driver would overwrite a
previously created debug build. On windows they are separate DLLs, and
thus co-exist. With this patch we always build debug.
We currently use the debug representation of Span to know if they are
next touching eachother as there are no stable way currently.
But that doesn't work with rust analyzer as it doesn't have a
representation we can test.
Also Span::source_text is not implemented with rust analyzer.
So just default to identifier with `-` touching instead of not touching
as this happens more often.
One can still use `(foo)-(bar)` when one really need minus
CC #685
