Quick answer: Rotating a device is a surface area for bugs: layouts must reflow without overlapping, the safe area moves, and on Android a rotation can recreate the activity and wipe state if you are not careful. Test rotating during every screen, mid-action, and during loading, confirming the game keeps its state and stays playable. Capture orientation in every report so you can reproduce the exact rotate that broke things.
A screen rotation looks trivial but it triggers one of the most disruptive events a mobile game can face. The viewport dimensions swap, the layout must reflow, the safe area insets move to new edges, and on Android the system may destroy and recreate your activity, taking unsaved state with it. Players rotate without thinking: to read a wide map, to type with a bigger keyboard, or just because they shifted on the couch. If your game freezes, scrambles its UI, or silently resets progress when that happens, it reads as broken. This post walks through QA for orientation changes: rotation handling, layout reflow, safe area, and preserving state through the transition.
Decide your orientation policy and test it
The first question is what orientations you actually support, and the bug is usually a mismatch between what you declared and what the game handles. Some games lock to landscape, some to portrait, some support both. Whatever you choose, the declared policy in the app manifest must match reality. Test by rotating the device in every state and confirming the game either rotates cleanly or stays locked as intended, with no half-rotated frame where the UI is sideways for a moment before snapping back.
Locked orientation is not a free pass, because the system can still deliver configuration changes for other reasons, and a foldable device changes aspect ratio without a classic rotation at all. Test on a foldable or a tablet in split-screen if you support them, since those resize the window without a ninety-degree turn and break the same code paths that assume a fixed size. Treat any window dimension change as the event to handle, not just the literal rotation, so your game survives the full range of resize events modern devices produce.
Verify layout reflow without overlap
When the viewport flips from tall to wide, every UI element anchored to an edge or sized as a fraction of the screen has to recompute. Bugs show up as buttons that overlap, text that clips, HUD elements that fly off-screen, or touch targets that end up under a notch that just moved. Test reflow on every screen, not only the main menu, because the in-game HUD, pause menu, inventory, and dialog overlays each have their own layout that may have been tuned in only one orientation.
Aspect ratio extremes are where reflow breaks worst. A layout that works at a typical sixteen-by-nine ratio can fall apart on a tall twenty-by-nine phone or a squarish tablet, and rotating moves you between those extremes. Test the narrowest and widest aspect ratios you support in both orientations and confirm nothing essential is cut off or unreachable. Pay attention to the safe area, since rotating moves the notch and the system gesture zones to different edges, and a control that was safe in portrait can land in a dead zone in landscape.
Preserve state through the rotation
The most damaging orientation bug is losing state. On Android a configuration change recreates the activity by default, and if the engine or game does not save and restore properly, the player can lose their current level, an open menu, or a half-finished form. Test rotating during active gameplay, mid-level, with a dialog open, and partway through entering text, then confirm everything is exactly where it was. A rotation that kicks the player back to a menu or clears their input is a serious defect even if nothing visibly crashes.
Loading and transition moments are the riskiest time to rotate. Test rotating while a level is loading, during a scene transition, and while a network request is in flight, because these are the states most likely to be mid-mutation when the resize event arrives. Rotating during a save operation can corrupt data if the save is interrupted by an activity recreation. Confirm the game either completes or cleanly resumes these operations across a rotate, and that no progress is silently dropped at exactly the moment the player was not looking.
Watch for crashes and rapid rotation
Orientation changes are a classic crash trigger because they tear down and rebuild resources. A texture, render target, or surface tied to the old dimensions may be referenced after it is invalidated, producing a crash on the first frame after rotation. Test the rotate on every screen and watch for crashes specifically right after the dimensions change, since that is the window where stale resource references bite. Low-memory situations make this worse, because the recreated activity may fail to reallocate what it needs.
Rapid and repeated rotation exposes race conditions a single clean rotate hides. Flip the device back and forth quickly several times and confirm the game does not deadlock, leak resources, or end up in a state that does not match the current orientation. Test rotating while the previous rotation is still animating or settling, which is how players actually behave when they are unsure which orientation they want. These stress cases catch the timing bugs that a careful one-rotation test pass will always miss.
Setting it up with Bugnet
Orientation crashes and state-loss reports are hard to reproduce because they depend on the exact moment and device the player rotated. Bugnet's crash reporting captures the stack trace and device context, including the current orientation and resolution, so a crash on rotate arrives with the frame data you need instead of a player saying it crashed when they turned the phone. Adding orientation as a custom field on reports lets you see at a glance whether a cluster of crashes all happened in landscape or right after a rotate.
Because the same orientation bug hits many players doing the same thing, Bugnet's occurrence grouping folds the duplicate crashes into one issue with a count and a shared stack trace. That makes a rotate-triggered crash obvious as a single high-frequency issue rather than scattered noise. Filtering reports by orientation or device class confirms whether the problem is universal or specific to a foldable or a particular aspect ratio, which tells you whether the bug is in your reflow logic or your resource lifecycle.
Make rotation a standing test case
Rotation is easy to forget because most testing happens in a single comfortable orientation. Add an explicit rotate step to every test case: on each screen, rotate and confirm layout, state, and stability, then rotate back. Include the stress cases, rapid flips and rotating during loads, so race conditions surface in QA rather than in the field. A written rotation checklist that a tester runs on every screen will catch the overlapping HUD and the lost-state bug long before a player files a frustrated report.
Re-run the rotation pass whenever you add a screen, change a layout, or support new device shapes like foldables, because each of those reintroduces the same family of bugs. Keep your device list current as aspect ratios get taller and resizable windows become more common. Orientation handling is a place where a little discipline pays off enormously: the games that survive a rotation gracefully feel solid, and the ones that scramble or reset feel fragile in a way players remember and reviewers mention.
A rotation reflows layout, moves the safe area, and can wipe state or crash. Test rotating on every screen, mid-action, and under stress, not just the menu.