Question

I am implementing a dashboard plus action bar UI, like in the Twitter app:

Each button on the dashboard takes the user to a different activity. A few of these activities are more important than the others, and I could imagine the user switching between them via the dashboard reasonably often.

I feel like I have two options:

1. Keep an activity cycle going using intent flags, so that when the user goes back to the dashboard it just pushes the dashboard activity to the top of the stack. Then when the user returns to another activity, it pushes that one to the top of the stack. No activity would be destroyed until the OS does it to gain back memory, which would be fine.

2. Let the activities be destroyed when the user goes back to the dashboard, then recreated later.

Which option is better in terms of performance and best practices? I like option 1 but am not sure if I’m abusing the purpose of those intent flags. And if I do go with option 1, should I also override what the back button does so that finish() isn’t called?

Answer 1

Personally I like the first option better. This way, you would easily remember the state of the other activities when the user returns to them.

For example, if in a child activity the user scrolls some list, then goes back to the dashboard, and then returns back to the child activity, the scroll position will be where he left it off.

Regarding memory, I don’t think it’s an issue. Let’s take a tab component for example (which is a parallel navigation controller to your dashboard). With a tab control, all the child activities (the tab activities) are never destroyed as well.

If memory does become an issue, I would combine your two ideas. For the less important activities I would implement approach 2 (destroy them on back), and for the more important activities (where state is important for the user for example), I would implement approach 1.