stuartmorgan
stuartmorgan
> try to call this [flutter/flutter#99155 (comment)](https://github.com/flutter/flutter/issues/99155#issuecomment-1052023743) in `onBackgroundServiceStart`. The error message is about `plugins.flutter.io/path_provider_ios`, which means the Dart registration has successfully run in this scenario.
Alternately you could wrap the Win32 APIs with a DI-able object, and then test that you get the expected calls when you simulate various return values (using a mock version...
Here's a very simple example of that pattern in one of the Windows plugins: - Wrapper [header](https://github.com/flutter/plugins/blob/main/packages/url_launcher/url_launcher_windows/windows/system_apis.h) and [real implementation](https://github.com/flutter/plugins/blob/main/packages/url_launcher/url_launcher_windows/windows/system_apis.cpp) - [Test implementation](https://github.com/flutter/plugins/blob/main/packages/url_launcher/url_launcher_windows/windows/test/url_launcher_windows_test.cpp#L28-L43) and example [test configuration](https://github.com/flutter/plugins/blob/main/packages/url_launcher/url_launcher_windows/windows/test/url_launcher_windows_test.cpp#L87-L97)
> And yes, `create` still needs to be called. There's actually a regression I just found where we're double-calling `create` now for HC in some cases. I'm fixing it as...
Given: > Ok, this was a bit too rushed. Crash is still there even with `controller.create()` line removed. it sounds like the original issue here won't be fixed by my...
> Since we already shipped it and people are starting to remove `create` as a workaround, I'm going to have to handle it internally by dropping duplicate `create`'s, and then...
If it's happening on 3.3, then the double-create issue I fixed was apparently unrelated, so un-assigning. @AAverin Have you tried switching to `initExpensiveAndroidView` as suggested [here](https://github.com/flutter/flutter/issues/107297#issuecomment-1204633350)? (Since the crash here...
@bparrishMines @dnfield Same here; still needs tests, but I would appreciate feedback. I have concerns about the fundamentals of the create/resize logic code here for TLHC (i.e., I suspect there...
> This also fixes a 3.0 regression found during manual testing, where `create` is sometimes called multiple times: once by the plugin author, but then again by internal code that...
This still needs widgets tests for the new switching widget, but I think it would be worth reviewing now to see if that approach is in fact what we want...