This commit just adds some #include directives and "using namespace" directives.
(The "using namespace mozilla::dom;" are the important namespace directives
here, though I've included "using namespace mozilla;" alongside it as well, for
completeness/consistency.)
Differential Revision: https://phabricator.services.mozilla.com/D1647
AccessibleCaretManager uses scroll and reflow observers to detect when
to update the position of carets. However, it currently only registers
the observers on the leaf docshell, so only changes in the innermost
iframe are detected; that is, it fails to update caret position when an
ancestor iframe is scrolled. This patch makes it register observers on
all ancestor docshells so that changes in ancestor iframes are detected
as well.
MozReview-Commit-ID: bwiSjj8936
In the era of B2G, we wanted to hide the carets during scrolling, and
PostScrollState was designed to avoid carets flicking during momentum
scrolling.
These days, we no longer hide carets during scrolling, so PostScrollState
can be removed to make the code simpler and easier to maintain.
MozReview-Commit-ID: Bf6ZgYVlt1q
The accessible caret manager is owned by the event hub, that is owned by the
shell.
All the callers of methods that call FlushLayout on the AccessibleCaretManager
should hold an external reference to the event hub.
Flushing pending notifications can run arbitrary script, that can call Destroy()
on the pres shell (and thus tear down the accessible caret event hub, and the
manager with him).
I don't know why before my change this wasn't crashing badly, but the code as it
was just doesn't look sound to me at all either (maybe I'm misunderstanding
something and I should just revert that patch and give up on having nice
invariants during our flushes..., but I don't think it's the case).
This also adds some sanity-checking that we don't die under our flush.
MozReview-Commit-ID: 4s0UT0fD3TI
Fix warning: use '= default' to define a trivial destructor
I added #include "nsDocShell.h" because I got an error "incomplete type
'nsDocShell' used in nested name specifier." After that, some files like
nsCanvasFrame.cpp fails to compile because they include
AccessibleCaretEventHub.h. Hence the moz.build changes.
MozReview-Commit-ID: BYZx7txvkSn
Override OnScrollPositionChanged() in ScrollState because we want to update
carets during scrolling in subframes without APZ.
Due to the observation in bug 1273045 comment 8, we do not distinguish
PositionChangedResult::NotChanged and PositionChangedResult::Changed.
Instead, we always update caret even if its position is not changed.
To avoid excessive CaretStateChangedEvents are dispatched in
OnScrollPositionChanged(), we add IsScrollStarted to distinguish whether
OnScrollStart() is called or not.
MozReview-Commit-ID: KNi9Mct4dSk
When the user switches back to a tab which has the selection highlight on
the document, call UpdateCarets() to bring the carets back.
MozReview-Commit-ID: LxNoNRl4FHZ
On Windows, the contextmenu event is fired when the finger is lifted after a
long-press. However, there are various bits of code, such as the AccessibleCaret
or potential fixes for bug 1147335, which would benefit from knowing when the
long-press gesture was detected. By moving eMouseLongTap event up we can satisfy
that need. An alternative approach considered was to fire the eMouseLongTap
before the contextmenu on all platforms unconditionally, but that makes it harder
to implement platform-specific text selection behaviour the way we want. In
particular we would have to add an extra message or notification for non-Windows
platforms that initiated text selection if the contextmenu event was not
consumed.
MozReview-Commit-ID: 2lmwxmmGrVD
On desktop, the context menu is shown when the user lifts their finger after
a long press, but only if the eMouseLongTap event is not cancelled. So by
not cancelling it, we allow both the text selection and the context menu.
On Android, the context menu takes priority over text selection, so this
has no effect (i.e. if the context menu is shown, then the AccessibleCaret
code never even gets the eMouseLongTap event). Also on Android nothing
else relies on the cancellation of the eMouseLongTap event, so this change
is a no-op.
MozReview-Commit-ID: peFzB2afha
In my original design, I treat eTouchCancel to be like eTouchEnd for
ending a caret dragging procedure.
However when pointer events is enabled, it sents an eTouchCancel event
after the eTouchStart event whose primary usage is to be converted to
pointer events, which then cancels the normal caret dragging procedure.
Moreover, when pointer events is disabled, we don't get eTouchCancel
during a normal caret dragging scenario, so we don't really need to
handle eTouchCancel anyway.
MozReview-Commit-ID: GKju2Tp0q3Q
When enabling "dom.w3c_pointer_events.enabled", we might get a
eTouchCancel event without any touch data. That is, aEvent->touches is
an empty array. We need to make sure it's non-empty before accessing
aEvent->touches[0].
MozReview-Commit-ID: BQUsrJjHHEl
Also I removed the 'explicit' keywords from the constructor since they have no
argument so nothing can be implicited converted to them.
MozReview-Commit-ID: GrFcqO0Uf1o
Also I removed the 'explicit' keywords from the constructor since they have no
argument so nothing can be implicited converted to them.
MozReview-Commit-ID: GrFcqO0Uf1o
After calling FlushLayout(), PresShell::Destroy() might be called and we
should consider PresShell and other resources will be no longer valid.
Before this patch, AccessibleCaretManager and AccessibleCaret(s) are
deallocated in PresShell::Destroy(). However FlushLayout() are all
invoked in AccessibleCaretManager, we need to keep manager alive to
clean up after PresShell::Destroy().
This patch makes AccessibleCaretManager live after PresShell::Destroy(),
and use IsTerminated() to check whether PreShell is vaild after each
FlushLayout() calls.
Note that event though AccessibleCaretEventHub will be unref in
PresShell::Destroy(), all the callers to AccessibleCaretEventHub's
public methods already add a ref to AccessibleCaretEventHub. So we don't
need to worry about AccessibleCaretEventHub and AccessibleCaretManager
die immediately after PresShell::Destroy().
MozReview-Commit-ID: DDpXZ7v3zyo
After calling FlushLayout(), PresShell::Destroy() might be called and we
should consider PresShell and other resources will be no longer valid.
Before this patch, AccessibleCaretManager and AccessibleCaret(s) are
deallocated in PresShell::Destroy(). However FlushLayout() are all
invoked in AccessibleCaretManager, we need to keep manager alive to
clean up after PresShell::Destroy().
This patch makes AccessibleCaretManager live after PresShell::Destroy(),
and use IsTerminated() to check whether PreShell is vaild after each
FlushLayout() calls.
Note that event though AccessibleCaretEventHub will be unref in
PresShell::Destroy(), all the callers to AccessibleCaretEventHub's
public methods already add a ref to AccessibleCaretEventHub. So we don't
need to worry about AccessibleCaretEventHub and AccessibleCaretManager
die immediately after PresShell::Destroy().
MozReview-Commit-ID: DDpXZ7v3zyo
This is a regression by "Bug 1121468 - Go to NoActionState after
receiving release on LongTapState."
When receiving a scroll event in LongTapState, i.e. apz starts, we
should call OnScrollStart() and move to the ScrollState.
When constructing AccessibleCaretEventHub, bypass and store the
nsIPresShell pointer. This is a preparation for later patches.
Also, remove 'virtual' for Init() and Terminate() since the mock object
in gtest does not need to override them.
The bulk of this commit was generated with a script, executed at the top
level of a typical source code checkout. The only non-machine-generated
part was modifying MFBT's moz.build to reflect the new naming.
CLOSED TREE makes big refactorings like this a piece of cake.
# The main substitution.
find . -name '*.cpp' -o -name '*.cc' -o -name '*.h' -o -name '*.mm' -o -name '*.idl'| \
xargs perl -p -i -e '
s/nsRefPtr\.h/RefPtr\.h/g; # handle includes
s/nsRefPtr ?</RefPtr</g; # handle declarations and variables
'
# Handle a special friend declaration in gfx/layers/AtomicRefCountedWithFinalize.h.
perl -p -i -e 's/::nsRefPtr;/::RefPtr;/' gfx/layers/AtomicRefCountedWithFinalize.h
# Handle nsRefPtr.h itself, a couple places that define constructors
# from nsRefPtr, and code generators specially. We do this here, rather
# than indiscriminantly s/nsRefPtr/RefPtr/, because that would rename
# things like nsRefPtrHashtable.
perl -p -i -e 's/nsRefPtr/RefPtr/g' \
mfbt/nsRefPtr.h \
xpcom/glue/nsCOMPtr.h \
xpcom/base/OwningNonNull.h \
ipc/ipdl/ipdl/lower.py \
ipc/ipdl/ipdl/builtin.py \
dom/bindings/Codegen.py \
python/lldbutils/lldbutils/utils.py
# In our indiscriminate substitution above, we renamed
# nsRefPtrGetterAddRefs, the class behind getter_AddRefs. Fix that up.
find . -name '*.cpp' -o -name '*.h' -o -name '*.idl' | \
xargs perl -p -i -e 's/nsRefPtrGetterAddRefs/RefPtrGetterAddRefs/g'
if [ -d .git ]; then
git mv mfbt/nsRefPtr.h mfbt/RefPtr.h
else
hg mv mfbt/nsRefPtr.h mfbt/RefPtr.h
fi
All desktop platforms support APZ now. eWheelOperationStart and
eWheelOperationEnd are superseded by AsyncPanZoomStarted and
AsyncPanZoomStopped respectively.
This macro was used to implement Name() and Singleton() for state
classes. Instead of enforcing the state class to be a singleton, we can
implement singleton pattern in MOZ_IMPL_STATE_CLASS_GETTER directly.
In AccessibleCaretEventHub, it is not accurate to use APZ enabled to
determine whether to use long tap injector. On desktop browser, there's
no long tap events even if APZ is enabled. We should use a preference to
control that.
Since it's a fact the APZ on b2g has long tap events, we should use
preference to disable long tap injector so that when long tap events
stop dispatching to AccessibleCaretEventHub, we'll know immediately.
Delete SetUseAsyncPanZoom() usage in gtest since APZ is not related to
scrolling in AccessibleCaretEventHub.
