As tests/cases/examples/image_rendering.60 itself demonstrates, the png file
should be loaded into the CPU side image cache once, but the two textures
should be cached independently.
cc #465
Like CSS image-rendering it has "smooth" and "pixelated" options.
Only OpenGL has been tested right now, have not tested WASM or Qt.
Right now the first instance of a @image-url() will set the scaling
for that specific image. The same image used from memory on the otherhand
can all have a different scaling property.
That's all it is nowadays, it's a wrapper around Rc<Window>. It's not an
alias because we need to also "wrap" it to C++ via cbindgen, but that's
about it.
This requires the image size query to be window independent and go
through the backend instead.
This implies minor changes for the Qt backend and bigger ones for the GL
backend:
* There exists now a thread-local IMAGE_CACHE, which is used by the
backend's image_size() function as well as by the renderer for
loading CPU side images.
* The image remain as decoded images in there (including SVG tree)
and the window now has a texture_cache, which holds CachedImage
with ImageData::Texture.
* Rendering an image item therefore fetches the CPU side image,
calls upload_to_gpu() on it, which creates a new Rc<CachedImage>
and that's stored in the texture_cache.
* The texture cache continues to be pruned when hiding the window.
Split the vertical and horizontal pass into different property cache
This will allow to implement "height for with"
This patch does not port the Rust or C++ binding yet
We don't need a property reference anymore, a plain image reference is fine AFAICS.
Dirtyness of the source is tracked on the caller side of layouting_info()
or in the draw_image_impl anyway.
Have a function first called before the children, and then the main function
called after the children if they did not accept the event. This will allow
processing the Flickable gesture properly
Remove the pos parameter to the render functions and instead let
the item renderer apply the transformation on the rendere (femtovg
canvas or QPainter).
So `draw_*` functions in the backend now always operate in item local
coordinates.
Don't use the item's rendering cache to determine the image size, as
that's soley for rendering. The Qt backend doesn't use the item cache
and the GL backend neither after this change. Instead both backends have
a cache for decoded images.
The GL backend uses the item graphics cache for the image size function,
which uses a PropertyTracker. That tracker must have the Image's source
included in its dependencies, to avoid that when loading a HTML image
for example, the cache isn't invalidated when the source is changed
before the HTML image was loaded async. That's why the get() call on the
source property must happen from within the PropertyTracker's callback.
For all other items the default is the empty size.
This also required three fixes for the HTML image loading:
* The upload_pending property value was inverted, it needs to start
out as true (pending yes) and be set to false when it finished
loading (not pending anymore)
* Mark the upload_pending property as dirty before scheduling the
redraw, in case it's sync
* Pass the item rendering cache to the image_size function to ensure
that the Rc<CachedImage> is not only weak inside the image_cache
of the GLRendererData but also strong on the item.
This means going back to an immutable renderer reference and interior
mutability for the canvas and the gpu cache. This is because while
traversing the item tree for rendering we may end up destroying
other items due to the lazyness of the models.
This is the start of a different rendering abstraction, the result of a
separate discussion.
The objective is to provide a better backend encapsulation, that will
make it easier to implement other backends and let the backend decide
what to cache and what not. This is done by passing the struct with the
Property<> fields directly to the backend, so that it can decide what
the cache and what to use every time.
