Dependency version upgrades, not API 35 yet

[keyboardsurfer]

No description provided.

[riggaroo]

Looks like the top app bar changed and i dont know if we want that?

[riggaroo]

It seems to have removed the cog icons and search.

[keyboardsurfer]

The top app bar changes were introduced with the upgrade to API 35 and are only visible when running roborazzi. For this maintenance effort it's out of scope for me to dive into roborazzi and address this.

Instead I reverted back to API 35 for target and compile Sdk version.

@takahirom FYI

Changes addressed but can't be resolved due to force push on branch

[takahirom]

Thank you for reporting this 👀
I think SDK 35 doesn't provide padding for the ActionBar, and since HiltComponentActivity in NowInAndroid includes the ActionBar, it ends up overlapping with the content.

Roborazzi is not using view.draw(); instead, we capture window rendering. (I think this approach might involve some tradeoffs, such as fidelity and issues like this.)
Overlapping UI could also cause issues with other test APIs, such as clicking, so it might be good to avoid overlapping.

I believe simply setting windowActionBar = false could resolve this issue.

I'm checking if this approach works:
takahirom#4
takahirom#5
takahirom#6

[keyboardsurfer]

Thanks for checking this @takahirom. It seems like takahirom#6 is the approach we should take for Now in Android.
@dturner WDYT?

[dturner]

It'd be more conventional to set the theme in the manifest i.e. add android:theme="@android:style/Theme.Material.Light.NoActionBar
to the ui-test-hilt-manifest module's AndroidManifest.xml

[keyboardsurfer]

While @takahirom's approach seems to work for light theme, there are still issues around theming with API 35 when using roborazzi. This change is out of scope for this PR. So we're sticking with 34 for now.

[dturner]

Thanks for doing this. Good to merge once the build and instrumented tests pass.

[takahirom]

It seems that val composeTestRule = createComposeRule() has the same issue in a simple library.
I was able to reproduce it in an emulator test (instrumentation test) as well, so I'm thinking of filing this issue on Issue Tracker tomorrow or the day after.