This can be reproduced by deleting the last item of the printer queue in the
printer demo.
It is a regression showing up because we now emit the MouseExit event after
the mouse grab as released.
The problem is that we upgrade the weak item, and call geometry() on it.
Calling geometry will re-evaluate the layout cache which will re-evaluate
the model which will result in the component being removed and the cache
entry having less item than expected.
It is ok to simply return 0. for these layout location since the item will
disapear anyway.
Move the sixtyfps::namespace entry into a dedicated C++ integration overview
page. Also duplicate and specialize the instantiation and model bits, which
differ between the compiled code and the interpreter.
Finally, fix the generated C++ docs to not mention that there's a constructor,
instead we generate a constructor function.
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.
Doxygen would still generate xml for private_api::ArrayModel and would
try to resolve a reference to iself, which is not possible because we've
excluded the namespace:
```
stderr: (!) Critical error while generating the file for [/home/vagrant/sixtyfps/target/cppdocs/api/classsixtyfps_1_1Model.rst].Traceback (most recent call last):
File "/home/vagrant/.local/share/virtualenvs/docs-4wjFCyEr/lib/python3.8/site-packages/exhale/graph.py", line 2591, in generateSingleNodeRST
gen_file.write("{0}\n".format(node.baseOrDerivedListString(
File "/home/vagrant/.local/share/virtualenvs/docs-4wjFCyEr/lib/python3.8/site-packages/exhale/graph.py", line 393, in baseOrDerivedListString
link=nodeByRefid[refid].link_name
KeyError: 'classsixtyfps_1_1private__api_1_1ArrayModel'
```
Making the entire class as \private works around it.
Oddly, this isn't needed for IntModel, perhaps because it's not a template?
Move those two classes into the private_api namespace, which is excluded
from the API reference documentation.
For generate code the explicit qualification of Property<T> is changed,
for the cbindgen generated item types the private_api namespace is
pulled into the cbindgen_private namespace.
A few problem:
- the horizontal and vertical property were swapped
- The implementation of the "preferred_xxx" property was not materialized properly
because the `bindings` were borrowed in the materialize_fake_properties pass
- Since the Window has a stretch factor of 0., the preferred size of the inner layout
was not taken into account when merging the LayoutInfo. I believe taking the
maximum preferred size is the right solution when merging them.
The item tree is intended to be initialized with pointers to the item
vtables, which are implemented using data relocation records that
resolve to the symbols exported by the sixtyfps_cpp shared library.
On windows, according to
https://docs.microsoft.com/en-us/cpp/c-language/rules-and-limitations-for-dllimport-dllexport?view=msvc-160
such data relocations are not supported, so this patch implements the
fallback through getter functions.
Mark some functions as inline to avoid generating symbols that would
produce duplicate errors when linking multiple .o files that include the
SixtyFPS headers.
* sixtyfps_timer_start needs to *take* the timer id out of the Rust
timer to avoid that the subsequent drop stops the timer again
* For the Qt event loop, call `timer_event()` once before entering
QCoreApplication::exec(), to schedule any timers that were started
beforehand.
* Added a way to quit the event loop gently, in order to use that
from the C++ unit test.
The problem is that the backend is using the `meta_property_listener` to know
if the layout must be updated, but the C++ code would not register a dependency
NOTE about the test: The test doesn't really test that it works because
the test backend don't have a meta_property_listener and apply the layout
inconditionally in sixtyfps_send_mouse_click
