READ ME
Version 1
getpreference is used as the primary storage for persistence. Session keys such as IMG_LAST and IS_FROM_CAMERA are used as session persistence, in case the activity gets destroyed on rotation.
Version 2
This version includes all of the UI-shells for future implementation. We start by enabling the ActionBar which for now does not contain any menu options, but does present the title. On the main activity, we have sliding tabs that present three panes. The first is the start pane which will allow users to access the necessary options to start an exercise activity. History pane is left blank for now. The Setting pane will allow users to set global app fields.
Edge cases:
- Privacy setting persists
- Orientation of gallery-picked image persists
- Units preferred persists
- Input Type and Activity Type persists
- List view is presented correctly upon rotation
Version 3
This version includes the sqlite implementation with usage of AsyncLoader for history fragment view and the loading of individual exercise entry. A Runnable was also used to delete entries.
Edge Cases:
- Units are converted
- Units from the preferences are used when inserting
- Units from the preferences are used when displaying
- end to end unit conversion
Version 4
This version includes the Map Location Manager, MapView, NotificationManager, and Services. (worked with Richard Cornew)
Edge Cases:
- Map is redrawn on onCreate
- LocationService sends new locations via Message Handler
- GPS and Automatic saves correctly in the same format as Manual entry
- Notification leads back to Map Activity correctly.
Version 5
This version includes the automatic activity type detection. (work w/ Richard Cornew)
It's an extension to version 4, the automatic activity features uses the same MapExerciseActivity to implement everything. With sensor detection, we make sure to keep it in the LocationService. We used an advance data structure -- linkedhashmap as a LRU cache to keep the N most recent sensor information, which we will use to classify the current exercise.
- Note that we wait for 10 seconds before classifying your exercise -- often times, we stop before we record our exercises.
Version 6
Peer-coded with Richard Cornew.
information passed to the server is in the format of JSON String -- the server will decode the json string
DEMO provided by GCMDemo is taken into inspiration. We use query.do and refactored it into sync.do to fullfil or post and get method requests.
For deletion, we also use the same mechanisms: however, we included GCM to registration and message sending! Upon deletion message, the app will delete the ID passed back via GCM
For GCM receive, we check if the fragment is visible or not. If visible, we reload the LoadManager again to reload the database into history fragment every time. That we can get the latest entries.
Major changes on the app site is on HistoryFragment.java