Since MOZ_NATIVE_DEVICES builds against play-services-{basement,base,cast},
some ad-hoc de-duplication is necessary.
--HG--
extra : commitid : 2jNIgZpLUq2
extra : source : 0957d3435ac22765d7868cb3c7db1e0787836bc3
These flags are not intended to be feature specific. On day one, we
intend to support a single GCM-backed feature -- Push Notifications --
but the set of GCM-consuming features is potentially large (e.g.,
possibly Firefox Sync tickles and Send Tab to Device alerts). Such
features can and will have their own build flags.
Note that GCM sender IDs are not sensitive -- see link in code
comment. Since this is something custom branding will almost always
want to set, I don't want to bury the default value in confvars.sh.
--HG--
extra : commitid : HBHDBk2tC6S
extra : source : d4e39587c87c16564eeac6d1f8568814b63e5325
The handled cases are:
* The user is not signed in (sync setup is shown)
* The user is signed in but has no other synced devices (toast is shown)
--HG--
extra : commitid : LsCwJgg4Zkn
extra : rebase_source : a5551ecbe992968eaf7055c9444f4c49ce5e95c0
This makes space to do more in the upcoming patches.
--HG--
extra : commitid : 4zbRVGonbfQ
extra : rebase_source : 50ad23724b31df3338ac36c3d178976cddb2047e
The set operation is:
* known engines are enabled by default;
* unchecked engines from fxa-content-server are disabled;
* the "forms" engine is special: it's always the same as "history".
--HG--
extra : commitid : Lbu8A5CH8R9
extra : rebase_source : 74e921492787f69fb36cd5c5e9da61dfd1e2ffc5
The only wrinkle is that we don't always have an Activity context, so
we always start a new task. We might want to revisit that decision.
--HG--
extra : commitid : Liun7Xv30B5
extra : rebase_source : c5afc439ba4fe818e4fda15b2756f510cb76d639
This patch bumps the browser.db version to 25 and makes
tabs.client_guid a foreign key to client.guid. It also replaces the
non-standard "rowid" with "_id".
Together, these disambiguate client_guid and guid and fix non-local
client deletion.
--HG--
extra : commitid : 5dSVFfdTJRK
extra : rebase_source : e72fe51e1afd1a94bbc96efcfaa3a302de468043
Right now, when the Gecko side commits its composition, we notify the
Java side and the Java side commits its composition as well. However,
some keyboards have trouble dealing with us manually committing the
composition. To better support these keyboards, this patch makes it so
that even if the Gecko composition is committed, the Java composition
is still kept. Because the Gecko composition on Android is used only for
display purposes, it doesn't matter too much when the Gecko composition
is out-of-sync with the Java composition.
Add a test to testInputConnection that reproduces this bug. Because this
bug involves JS running during editing, another input field is used on
the test page. Additional code was added to testInputConnection,
GeckoInputConnection, and GeckoViewComponent in order to reproduce the
bug in the new input field.
The menu code is hard to change (I filed bug 1216328) so I opted
to save time by adjusting the padding in the image with an inset
drawable. It'd be better to adjust it statically in the menu code.
--HG--
extra : commitid : D47jDAhHbJV
extra : rebase_source : 9025c9a6e6e7b0010f2f2f3c6311786bc8b2d9ab
The approach is the same as that used on Desktop. The function name
calls out the "prePath" and "milliseconds" since the first is likely
to be overlooked by Fennec consumers, and the second is notably
different than Desktop (where the database timestamps are microseconds).
--HG--
extra : commitid : 2A7Eux7zPkE
extra : rebase_source : 1ae224da0e2a7836bda84a4371e89afa84f2bf2a
Before we inherited from GeckoBase, which did not have the colorAccent
attribute.
--HG--
extra : commitid : LK0WCHwaH6x
extra : rebase_source : caeaf49dd9f7e6f05e3b8cfede997eeb2a75a805
On newer devices there's no fading edge but on Gingerbread you can see
an ugly, misplaced fading edge effect.
--HG--
extra : commitid : 16zS0rofxHM
extra : rebase_source : f3bed7e22971204cc8f1697e8adc28fb4ff54635
extra : amend_source : 0d418d43d827bc541f0ce50fc63e455587540c56
After switching to a full screen tabs tray there's no chrome to drag anymore.
--HG--
extra : commitid : CirRiLLp5Ry
extra : rebase_source : 64fcd4ffd9ae61788e38bed175b6d98701aca3ff
This creates feature parity with the tab view of the grid tabs tray.
Previously the ImageButton created an unpleasant effect when the List row was pressed (due to
duplicateParentState). Rather than losing the selection state by removing duplicateParentState,
this patch will effectively remove the focus state from the close button. This ressembles the
behavior of the close button in the grid version of the tabs tray.
--HG--
extra : commitid : Boe3L8qOvJ1
extra : rebase_source : 008d7532becde2afa272449074d80aec74d50f17
This will create feature parity with the look & feel of tabs in the grid view.
--HG--
extra : commitid : 3eOQy5B93tt
extra : rebase_source : b2079bb0cf7fbba0b82010c22b59b58e8dbd9905
This patch is using the same value we use in the tabs tray grid (tabs_layout_item_view.xml).
--HG--
extra : commitid : I4YLFKAccy1
extra : rebase_source : 21cee94b2ca474228437368ea9219f26670080a7
Right now we call disposeNative on GeckoEditable in the
nsWindow::Natives destructor. However, we may still have pending native
calls in the event queue at that point, and these events will cause
exceptions when handled. This patch makes GeckoEditable call
disposeNative, after ensuring there's no pending calls.
This patch adds a separate close() call to nsWindow, and let the
GeckoView decide whether to make that call or not. This lets us use the
static version of disposeNative. If nsWindow is destroyed in the
meantime, we still want to call disposeNative, which would only be
possible using the static version of disposeNative.
