* Handle startSync and loadMore
* Use SyncManager rather than MailService for periodic sync
and upload sync
* First of many CL's to disentangle sync from UI
* Note that the large majority of this CL is a refactoring
of IMAP specific code out of MessagingController and into
ImapService; MessagingController will eventually be
removed entirely from the app, as will much of Controller
Change-Id: I13546d0694479b33cf93c25920dedc1d38227f6c
* Currently, we store a generally incompatible packed string format
for to, from, bcc, cc, and reply-to columns
* Change this to UI-compatible, comma-separated RFC822 strings
* Upgrade database to use new format
Change-Id: I42c01d72b49531c4324d3956edf5ff362d5c9120
Linkify email addresses in expanded message details
Remove divider line in upper header on phone
Embolden sender name
Twiddle expanded details line spacing and padding to match Gmail
Add space char to collapsed details recipient enumerator (comma)
Bug: 5520692
Bug: 5520787
Change-Id: Iad4a265168d111aa01654a33a9a14f7968fc3181
* We need to call this out to the user (a surprised user is an
unhappy user)
* Required for fix to bug referenced below
Bug: 5533550
Change-Id: I64a870a703de7e02ce882aefc3636a1668021eb8
* We need to call this out to the user (a surprised user is an
unhappy user)
* Required for fix to bug referenced below
Bug: 5533550
Change-Id: I7da337057b9b28300ddd485283920e12c1ae53da
* Remove PolicyService APIs policiesRequired, policiesUpdated,
isSupported, clearUnsupportedPolicies, and isActiveAdmin
* Add PolicyService API setAccountPolicy, which is the sole
method by which security policies are promulgated
* Add protocolPoliciesEnabled and protocolPoliciesUnsupported
to the Policy class; these are packed, localized strings
indicating policies that the protocol itself have enabled
and/or cannot support (i.e. these are policies that are
unknown to the DPM, e.g. don't load attachments)
* Differentiate in security notifications between three kinds
of policy changes - changes that don't require user
intervention (e.g. reducing requirements), changes that
require user intervention (the legacy notification), and
changes that make the account unsyncable (e.g. the server
adding an unsupportable policy). Handle all possible policy
changes cleanly.
* Make security notifications per account (with multiple
accounts, notifications would get arbitrarily munged)
* Expose ALL enforced policies via the account settings
screen in two categories: policies enforced (including
both policies enforced by the DPM and policies enforced
by the protocol) and policies unsupported (note that these
can only be seen if policies are changed after an account
is created; we do not allow the creation of an account
when any required policies are unsupported). Add a
button that forces a sync attempt, for accounts that
are locked out, but whose policies have changed on
the server (this would otherwise require a reboot).
* Updated unit tests
Bug: 5398682
Bug: 5393724
Bug: 5379682
Change-Id: I4a3df823913a809874ed959d228177f0fc799281
We always switched to combined view when we tapped "Starred" in a
mailbox list. This was confusing since tapping to show the mailbox list
will take you to the mailbox list for the combined view, even if there's
only one account on the device! Other confusing things happened like
showing coloured account chips, even if there's only one account.
This fixes it so that we pass the account ID when building the
selection. Lots of wiring to make it happen, but very little change
overall.
Bug: 5388326
Change-Id: I1fe52f211cceca0c1b26581e200f3c7adcd0734a
When the socketfactory init code was moved, I forgot to re-add in the check
to skip hostname verification. This made "Trust all SSL certificates"
checkbox useless.
Bug: 5450563
Change-Id: Ie4cba749aaf8c0fd9f9c43f09ebf354c6600d4f0
* We weren't updating the account object passed in to setAccountPolicy,
which is problematic if the object is cached and used subsequently
in the code (it would have stale policy values)
Bug: 5245056
Change-Id: I279570dfd66c4a65d751f73e2e56578be186d339
* Rather than returning false, which can cause unexpected
results, e.g. locking out an account until reboot
* Partial fix for the referenced bug (there might be other
causes)
Bug: 5221119
Change-Id: I5b47093a3411deb6995624887197297323db0d2d
This is needed so Exchange can fire up Email activities.
No change other than the move.
Bug: 5122497
Change-Id: I6ee3f61654745fafd444314ecf75eb8ae6bbd01d
- hide sync options for mailboxes that aren't actually syncable
- hide search icon for any mailboxes that don't have any server
counterparts
Bug: 5086134
Change-Id: I60d5e5d1225a946464c5ea273c7d4b9299f433e5
- Since the Email and Exchange processes do not share UID, we need to
ensure the call to the KeyChain to request the certificate happens in
the Exchange process
- Misc UI fixes so it's not so ugly
Bug: 5117682
Change-Id: If80698850902e0178eb0998493b4cf4b89e2a15c
We seemed to not properly fallback to querying the database if the cache
did not contain a mailbox of a specific type. We may want to consider
caching mailboxes related to PIM data, but for now this is a quick fix.
Also fixes unit tests.
Bug: 5019661
Change-Id: Idcac0a6f15aa7f174890ae586f478fbb8f6e05b7
* Move AccountReconciler to the Email app (from EmailCommon)
* Ensure that Controller.deleteAccountSync() performs ALL actions
needed to clean up after an account deletion (delete attachment
files, reset policies, refresh the UI, etc.)
* Add reconcileAccounts() API to AccountService
* Remove accountDeleted() and restoreAccountsIfNeede() from the
AccountService API
* Remove unused callback
Bug: 4883073
Bug: 4767084
Change-Id: I43ffaf009db1a6f306bb0f2a74fb4dd3b2c4b966
There were two issues:
- the default account ID from the provider wasn't consistent: it was
using a snapshot from the cached data but that simply returns a Map, and
the values aren't guaranteed to be sorted (and indeed I saw that the
order was different on consecutive calls!)
- hitting app up from Compose always just kicked you out to the inbox
for the default account, but it probably should have used the account of
the compose screen, since it can be specified in the Intent
Bug: 5012008
Change-Id: Ic9a753b261e047790453bc1a9417bc0c6d2f87f9
This prevents things from always failing if the server requires a client
SSL certificate.
Note that the solution used to determine if a certificate request was
made for a given request is approximate; it is timestamp based and can
theoretically give a false positive. In practice, this is very unlikely,
since another cert request had to have happened around the same time,
AND the response must be a 401/403.
Change-Id: Ieb77cf91db3bd52ba4adf1fb07357fef7e204ba5
There were two places where account flags were being set on setup: one
was in manual setup when the user selected the account type
(pop/imap/eas), and another was when it was automatically determined
from XML. The two paths were inconsistent and one was setting bits like
SUPPORTS_SEARCH but the other wasn't.
This fixes being able to search in IMAP
Bug: 5011932
Change-Id: I50eada7485aaef4e2a85d25fa7eb9ac7ab03093a
* ImapFolder is currently very unsafe for use by multiple threads,
causing, among other things, the referenced bug
* Since ImapFolder is very lightweight, there's no particularly good
reason to be caching them anyway
* Rename isOpenForTest to isOpen
Bug: 4972084
Change-Id: I2bf17b9cfc8549a222e991f3e59abfd00a4d3afd
Renamed onPostExecute to onSuccess and made sure it won't called
if a task is cancelled in time.
Also removed isCancelled(). To implement it right we should make sure
that onPostExecute() isn't finished when setting mCancelled, but it's a bit
of a pain to implement right, and we don't really have to use it.
Change-Id: I3a0baf504506ffc4952a5553f7098a8415842fa3
This helps post runnable to the handler and cancel pending runnables
at once.
It'll be used for delay-call methods that initiate a fragment transaction,
and cancel then in onSaveInstanceState().
Change-Id: Ib8bdb0e676e756854ab067a27e5e0f397219a4b4
Supress "Body part ended prematurely" and "Unexpected end of headers detected".
I believe these happens because we feed partial messages to the parser.
We've kept these logs only "just in case", but I don't think it's ever been
usueful...
Change-Id: I29e5e48abf71612eed656ce6569246caf8a320bc
* Add protocolSearchInfo column to Message table; this can be used
to store information related to search results. For IMAP, we
store the serverId of the mailbox that the message lives in on
the server
* Add upgrade code for this column
* Change MessagingController to use the proper serverId for remote
operations, depending on whether the Message is a search result
or not
* Fix some smaller issues with earlier code
Change-Id: I0c7f1d89a4659b95701d02646c0e8426680e2f6a
1) Have CachedCursor implement CrossProcessCursor; still need to
figure out how this ever worked
2) Close cursor used internally in findMailboxOfType
Bug: 4869024
Change-Id: Id20d37b7b83e133aa4d5fe9293a42ae217024f01
* Guarantee that up to 16 Account (with HostAuths), and Policy rows
are always cached. Also, 6 commonly used Mailboxes per Account
(Inbox, Outbox, Drafts, Sent, Trash, and Search)
* Precache these rows when EmailProvider starts up
* Ensure that newly added, precachable rows are cached when created
* Clean up some inefficient/wrong caching code
* Fix a commonly called method in NotificationManager in which we
load a single Mailbox row using selection vs withAppendedId
* Confirm that we don't read from the database in typical use and
heavy message loading
* Add a special URI for finding mailbox by type (using the cache)
* Add special-case code for EmailContent.count(Account.CONTENT_URI)
which is used in a number of places (including on the UI thread)
and whose value is easily determined
* Add a special URI to get the default account id
* Confirm that all unit tests work
The goal here is to be able to load all Account, HostAuth, Policy,
and Mailbox objects (by id) without worrying about disk access.
There will still be a single disk read for uncommon Mailbox reads,
but this should be considered acceptable.
Change-Id: Ibc9aa7acc73185e360b0b6f3053b90a985e97210
TODO: Unit tests
Added "Insert quick response" button to MessageCompose's action bar. Clicking
it opens dialog with available quick responses. Selecting one of the responses
will insert it into message body at the current cursor location. Also added
menu in account preferences to create, edit, and delete quick responses.
Change-Id: I85f3f6b36801cf112ec9d7c31135a917456173d7
* Broke up synchronizeMailboxGeneric into three pieces; it's still
horrible, but this at least stops my eyes from bleeding
* Remove unused method/tests from Folder interface
Change-Id: Ib4d979536be657137cf70ca535cf429d707be41b
When the KeyStore fails to give us back a certificate for any reason (it
was removed from the keystore perhaps), propagate the error back up.
Change-Id: I4f0ef783c1665589cc8ccb43d95da43a297a3e9a
This introduces an exception which needs to be thrown from a KeyManager
when it tries to establish a connection with a server requesting a
certificate.
Change-Id: I06dfad7789ed5d320b630e7e4380e15da42a48df
This sends the bit to the DPM. Separate changes have been/will be made
to change the provision parser and support it in the DPM.
Bug: 4185316
Change-Id: I44872ceb095a28539b047a0641cc499c7186a9b3
This introduces the ability for clients (i.e. the exchange service) to
register "special connection types" that use a client certificate stored
in the system keystore. The alias is encoded into the URI scheme for
those clients, and the socket factory used for those connections will
use the approprate KeyManager.
Lots of TODO's, including bubbling a lot of this up to the higher level
and wiring the UI to actually set the alias in the HostAuth table.
Change-Id: If5e1901c5b58731fdabd3e6b6da7198134b512d2
* Also, a convenience method to test the base flag
* Follow-up CL will add these flags for Exchange
Change-Id: I38dc1d19e89805989e674c17919db238b4913a1c
* Remove per-store limitations
* Use constants for VISIBLE_WINDOW, rather than having the
potential for differences between Stores
Change-Id: Idd5e0874bba6e3390e4f093bcb03f4b1bb399c11
Some email servers require client certificates to be presented to
establish an SSL connection. While this certificate will be maintained
by the system key store, we need to store the "alias" of the certificate
stored in that system store.
Wiring up to use the actual alias will be done in future CL's. It is
currently unused.
Change-Id: I8d1290151342daea9ceb0df8a4088405b44faa81
The last touched time will track the last time a message within that mailbox
was read. This will be used for the recent list.
Change-Id: I97a5fda52fd09b416fc3278a11a87b807da05c9c
The way the config activity communicates with the widget will be through
shared preferences. We now read / write shared preferences for widget
configuration. One step closer to the configuration activity...
Change-Id: I7c54259d84ad8d304a61652af5b3edff4c7d67db
The new widget UX allows for a single display mode for the widget. This can
be configured when the widget is added or at a later time during widget
re-configuration.
We don't have the configuration activity (yet). We first need to restructure
the widget to take a single account / mailbox combination. Hooking up the
configuration activity will occur in a future CL.
Change-Id: I38a5796c44938a6abd0d2bb50ac77241cc86a497
After choosing the account, we need to select a mailbox for the shortcut.
In order to replace the fragment, we cannot hardcode it in the XML. Instead,
we need to dynamically create it.
Also, restructure the fragment class and rely on the class to "do the right
thing"
Change-Id: I752ad5bbdf2484332ec2b73852cae74a5d2092fa
- Added Message.NO_MESSAGE
- Renamed PSEUDO_ACCOUNT_ID_NONE to NO_ACCOUNT
- Removed PARENT_KEY_NONE and use NO_MAILBOX instead
- For starters, cleaned up the UI controllers to use them.
Change-Id: I6cfd87ece2fced8e9f7c76d034c4d1dbf9e4db10
- Moved the method to EmailCommon.
- Use *_SELECTION for magic mailboxes
(meaning we now use subqueries for magic mailbox selections, rather than
building the mailbox ID list by ourselves)
Change-Id: I3ebf6af62fd912fea6faea0f75e05fc61c87af3b
Since the notification controller now operates exclusively using database
observers, there's no reason for the exchange service to call the
notifyNewMessages() service API.
Change-Id: Iaa7e2f5eae786162eab23b02b03ce6d1e8a738e9
We will suspend notifications whenever we display the message list for an
account (including "combined inbox"). As soon as the message list is paused,
notifications will be resumed.
Change-Id: I481a0f59ce68f89c32210d862d0267f3f334063b
* This is a serious bug dating back to the first Honeycomb release
* It was possible that a newly created Message could not yet be
committed to the database when the AttachmentDownloadService
tries to download one of that message's attachments.
* ADS, when it sees that the message (apparently) doesn't
exist, deletes the Attachment (it appears to be orphaned)
* The effect is that the user never sees one of the attachments
in a message.
* This bug has been reported externally
* The fix is simply to check for the message's existence before
deciding to delete it (this check will always work properly)
Bug: 4409692
Change-Id: I106ed2fe88d2435ad7a462fced5cb307c2559fd6
The notification controller now observes changes to the account database and
adds or removes message observers as appropriate.
Change-Id: I1670fcfd6ce744030199b86708a6ada55b239a84
We were sort of using observers to maintain the new message notifications.
However, other parts of the code would poke into the notification controller
to set things such as a list of newly added message IDs. Now, we rely
exclusively on db observers to manage notifications.
As a side effect of this, we now set the notification text correctly to be
the most recently _added_ message. This may be different than the most recently
sent message [since there may be a non-negligable delta between when the
message was sent and when it was received].
NOTE this still suffers from an outstanding bug where we continue to get
notifications when the Eamil UX is visible. That and monitoring changes to the
account table will be addressed in future CLs.
Change-Id: I4c68273716cc685574a1ca71e5d634f53fe0d882
* Use EmailProvider to backup/restore into a backup database
* Remove all of the old AccountBackupRestore code
* Get rid of the legacy Account class and all of the Preferences
crap that referenced it
* Remove corresponding tests
Change-Id: I2de75aafdacc87246174303961e58547303f641e
Now attachments are actually stored in an explicit list, instead of
being inferred from the state of the UI. This makes it possible to
switch states and restore attachments, and test.
Change-Id: I8c5f80f17f8c9e78d880ac4a1ac6ae22c2ec0579
If we receive new messages, we may display a notification to the user. If
those same messages are read elsewhere (i.e. via a web client), we will
remove the notification.
Change-Id: Iba09afe01942e0deaac8210fd6f9b315b1c8c93f
* Before the Email/Exchange split, EmailProvider was guaranteed not to return a null
cursor; this is no longer true
* We now throw a new RuntimeException (ProviderUnavailableException) when calling
commonly-used utility methods (e.g. restoreXWithId) and the cursor as returned from
EmailProvider is null (which implies that the provider isn't available).
Change-Id: I152d13bff0129c23586cd570d23c87d69cfce372
* Update sync window options to include "Automatic" and "All"
* Make the default (for new accounts) "Automatic"
Change-Id: Icbc696f55abcfad79e66079ebb414ca50778dcaa
* Replace crazy (and soon to be "full") bit fields stored in an account's
securityFlags with a row in a newly created Policy table (thus, fully
expandable)
* Update code from database version 17 to 18; adds Policy table, a
policyKey row in Account, and a revised trigger that deletes Policy
information for deleted Accounts
* Update old PolicySet unit tests to work against the new Policy class
* Add test for the conversion of securityFlags to Policy
* Tested in a variety of scenarios; appears to be functionally equivalent
Change-Id: I1505ee75230d6a0d3c2b62a46326f39c2c7f9eb5
- "save draft" no longer closes the message
- ensure consistent state if there are successive saves
- misc changes to the way a message is loaded if there is a pending save
Bug: 3072398
Change-Id: I9cd01319772293e4d410020ab27603821a95ec9f
- Now we always use a fragment as a callback, rather than assuming the parent
activity implements it.
- Use a generics trick to make sure the callback fragments really implements
Callback.
(Might be abuse of a language feature, but it's at least safer than runtime
check...)
Bug 4314669
Bug 4345496
Change-Id: If4048d456b298784097e202cffab170177ac7b2d
not making any real code changes:
* removed deprecated, unused methods
* remove 'throws' clauses when that exception is never thrown
* renamed method Controller#moveMessage()-->moveMessages()
Change-Id: Ifd006f760f0c19283e94a11a45c71295c8da35f7
With the recent changes to hierarchical folders, the move-to dialog is
quite unusable if you have multiple child folders with the same name.
While waiting for UX to decide on the exact display, make a few quick
changes to display the fully-qualified pathname instead of just the
child folder name.
Change-Id: Id5c1cc98364fbf7a82a05ac30e944507c7d16320
We now create folder hierarcies for IMAP. This also includes a nifty SQL
statement that will get your existing database into shape.
Change-Id: If07a0632e9b250cf0c33c3e16bfba5816beab94c
The logic is simplified by creating database rows for new mailboxes down
in Imapstore. This means that the difference between local and remote folder
lists are mailboxes that need to be deleted.
Note -- this is still not the final CL. We probably update the database too
frequently and the column values aren't updated to support nested folders.
Change-Id: Ifbe4e0cf74ba81e5b6156b452ab72c56c35235ab
