UiThread:
Fixes a common error from the market:
android.view.ViewRoot$CalledFromWrongThreadException: Only the original thread that created a view hierarchy can touch its views.
at android.view.ViewRoot.checkThread(ViewRoot.java:2802)
at android.view.ViewRoot.invalidateChild(ViewRoot.java:607)
at android.view.ViewRoot.invalidateChildInParent(ViewRoot.java:633)
at android.view.ViewGroup.invalidateChild(ViewGroup.java:2505)
at android.view.View.invalidate(View.java:5139)
at android.view.View.setFlags(View.java:4502)
at android.view.View.setVisibility(View.java:3030)
at
com.fsck.k9.activity.MessageList.hideBatchButtons(MessageList.java:2883)
at
com.fsck.k9.activity.MessageList.toggleBatchButtons(MessageList.java:2906)
at com.fsck.k9.activity.MessageList.access$500(MessageList.java:77)
at
com.fsck.k9.activity.MessageList$MessageListAdapter.pruneDirtyMessages(MessageList.java:2302)
at com.fsck.k9.activity.MessageList$1.run(MessageList.java:811)
where you're unlikely to even have access to them the first time you
walk through into the "Folders" preferences.
At the same time, move toward using a list preference widget, rather
than a custom activity.
- extracted handling of external intents to a new method
- only allow mailto URI for ACTION_VIEW and ACTION_SENDTO (as documented by AOSP)
- allow additional recipients via "to" parameter in mailto URI
- removed check for allowed MIME type when adding attachments since we're accepting all MIME types anyway
- show CC and BCC text fields when one of those recipient types was set by the intent
- use MIME type specified in intent for attachments (if present and doesn't contain a "*")
* mail-on-sd: (40 commits)
Added more comments to explain how the locking mecanism works for LocalStore
Fixed wrong method being called during experimental provider initialization (since provider isn't enabled, that didn't harm)
Add more comments about how the various StorageProviders work and how they're enabled
find src/com/fsck/ -name \*.java|xargs astyle --style=ansi --mode=java --indent-switches --indent=spaces=4 --convert-tabs
French localization for storage related settings
Remove unused SD card strings (replaced with storage indirection)
Merge mail-on-sd branch from trunk
Reset mail service on storage mount (even if no account uses the storage, to be improved)
find src/com/fsck/ -name \*.java|xargs astyle --style=ansi --mode=java --indent-switches --indent=spaces=4 --convert-tabs
Migraion -> Migration
move the Storage location preference into preferences rather than the wizard.
Made LocalStore log less verbose Added @Override compile checks
Added ACTION_SHUTDOWN broadcast receiver to properly initiate shutdown sequence (not yet implemented) and cancel any scheduled Intent
Be more consistent about which SQLiteDatabase variable is used (from instance variable to argument variable) to make code more refactoring-friendly (class is already big, code extraction should be easier if not referencing the instance variable).
Added transaction timing logging
Factorised storage lock/transaction handling code for regular operations.
Use DB transactions to batch modifications (makes code more robust / could improve performances)
Merge mail-on-sd branch from trunk
Update issue 888 Added DB close on unmount / DB open on mount
Update issue 888 Back to account list when underlying storage not available/unmounting in MessageView / MessageList
...
- Ported to the ColorPickerDialog API
- Translated the API and Internals from Indonesian to English
- Moved to use K-9 strings rather than custom strings.
In an ideal world, we should use ambilwarna as an android library
project, like it was intended.
updates on a thread rather than on the main ui thread. it results
in the list blinking with old data, but that's still a better user
experience than "frozen"
because a long SQL query takes more time before starting to return
results, K-9 was increasingly slow on large folders.
Doing two searches may take marginally more time but starts getting
results back to the UI much faster.
This doesn't work on initial sync, since the comparisons fail and you're
left with duplicates in the displayed mailbox
This reverts commit fa1c88bec348d0132acc60a320626bf0ca1170ec.
the list, use a less painful equality check than iteration.
This works because messageInfoHolders compare to each other using the
same key as message they contain.
Message retrieval:
- Use the given projection for the cursor result instead of always using a predefined projection. Extraction logic is now in dedicated classes for easy extension.
- Added BaseColumns._ID & BaseColumns._COUNT columns.
- Deprecated "id" column since it's not a reliable identifier (it's an incremential value generated at query() time)
- Moved ListPreference helper functions from FontSizeSettings to K9PreferenceActivity
- Used helper functions in Prefs
- Reordered preferences in Prefs to match the order in global_preferences.xml (display order)
- Removed the contact names cache (memory leak)
- Changed Contacts.searchByAddress() to Contacts.getNameForAddress(). Cursor is now immediately closed.
- Only try to resolve contact names when "Global settings" -> "Show contact name" is checked
- Never resolve contact names if number of recipients exceeds a threshold
Don't export the formatted date through the content provider: now exported as the raw Epoch value - date formatting is a UI concern. That makes MessageInfoHolder lighter.
Added semaphore to prevent third party content resolver from exhausting K-9 resources: it won't allow more than 1 cursor to be active at the same time. Concurrent threads (other content resolver invocations) will be suspended until the previous cursor gets released.
- Don't instantiate an empty Address array just to throw it away (less gabarge collection)
- Use a StringBuilder rather than a StringBuffer (faster as it isn't synchronized)
Status: Fixed
Extracted Intent.ACTION_SENDTO handling in a separate block to extract data from URI (as opposed to ACTION_SEND which uses Intent extras to pass data)
Removed dependency from K9 to MessageProvider (and added ability for components to register themselve for being notified when K9 instance is ready)
Moved Intent broadcast code from MessageProvider to K9
Made query() easily extensible using strategy pattern, the discriminant still being the Uri path
Fixed severe design flaw where the message provider would accumulate messages, causing an OutOfMemoryError even when the message provider isn't explicitely invoked.
query() is now synchronized with message retrieval
Added an option whether the registered name color is change or not
Avoid "Connection error" when MessageView is opened. The contacts.getName() might return null.
deleting by message_id. I believe this was our last serious leak in
our flash storage
deleting message headers was....just wrong. we've fixed other similar
bugs inherited from android 1.0 before.
