Prior to this the Android embedder code would extract the icudtl.dat asset out
of the APK and write it to local disk during the first startup of the app.
This change will make that work unnecessary and eliminate the risk of ICU
failures due to errors in the extraction process.
Behavior (visual) changes should be very minor. Things that are to be expected:
* A few things were not color managed correctly by the transform canvas (color emoji, some color filters). Those will be handled correctly with the tagged surfaces (although we're always transforming to sRGB, so nothing should change until we target a wider gamut).
* Image filtering will happen in the source color space, rather than the destination. Very minor.
* The transform canvas did caching of images in the destination color space. Now, the conversion happens at draw time. If there are performance issues, images can be pre-converted to the destination with makeColorSpace().
This change adds explicit validation of dynamic patches in all places
where they're used, instead of only validating it in some places which
wasn't as reliable because some of the code paths were missed.
This change also moves utility functions that deal with validating
patches from ResourceExtractor to ResourceUpdater, to make them
available as API for other places in code that need this validation.
This fixes potential race condition when patch gets downloaded on top
of zip file that's currently in active use by resource extractor and/or
asset manager. This change is necessary since download can happen in
the background while normal application operations are in progress.
This is a no-op change, except for fixing a bug where download task
reference wasn't cleared after download was completed.
This change also removes call to output stream flush(), which is not
necessary according to Java spec.
The rest of the change deals with requiring the code to work directly
with ResourceUpdater object instead of having FlutterMain be a facade
that forwards some of ResourceUpdater's methods. This simplifies the
other (more essential) upcoming changes that will be landing in the
followings few PRs.
This change introduces manifest properties that control when dynamic patches are downloaded and installed in the application lifecycle.
Application developer can choose whether between install on restart, install on resume, or immediate forced install of dynamic patches.
Minor refactoring of dynamic patching code.
- Changes naming of manifest properties to be consistent with documentation.
- Moves methods from inner class to outer class to make them more reusable.
This replicates similar logic found in ApkAssetProvider that locates assets by their short name in a sudbirectory inside of an archive file, instead of only at the root of archive.
Unique patch numbers are mainly useful for canary and A-B testing, but otherwise complicate things and can now be omitted.
Also, always append .zip to patch download URL. This obviates the need to configure redirects on the server when not using patch numbers.
After bumping the targetSdk of Flutter apps to 28, the keyboard doesn't open anymore when inputting text on an Android Pie devices. The problem is caused by a change on the way the focus in handled: when the call `mImm.showSoftInput(view, 0);` is made, the currently focused view is the `DecorView` from the `Activity` hosting Flutter but the passed view reference (`view` argument) is the `FlutterView`. As the `InputMethodManager` checks if both views are the same before showing the keyboard, it never appears because the `FlutterView` never has focus.
As Flutter doesn't have any input views as far as the Android side is concerned, the focus should always stay on the `FlutterView` itself and thus, this PR changes the `TextInputPlugin` to focus on the `FlutterView` as soon as it's created, fixing the keyboard issue.
Co-authored-by: Igor Borges <igorborges12@gmail.com>
