This patch also creates non-virtual methods,
EditorBase::BeginTransactionInternal(), EditorBase::EndTransactionInternal(),
TransactionManager::BeginBatchInternal() and
TransactionManager::EndBatchInternal().
Although, this could be replaced with API to use PlaceholderTransaction,
we should investigate it when we have much time.
Differential Revision: https://phabricator.services.mozilla.com/D2989
--HG--
extra : moz-landing-system : lando
Calling EditorBase::EnableUndoRedo() without argument means that editor supports
unlimited undo/redo stack. AutoDisableUndo class calls it without argument
when it needs to restore undo/redo feature.
However, <input type="text"> and <textarea> limits number of maximum
transactions up to 1,000, perhaps for footprint. So, AutoDisableUndo should
store the last number of maximum transactions before disabling undo/redo from
the constructor.
MozReview-Commit-ID: CoI6ZXyTd3X
--HG--
extra : rebase_source : e2b9af17e5857dcc0a6781e254e45fdb790c9a9e
nsITransactionListener::AddListener() and
nsITransactionListener::RemoveListener() are of course virtual methods.
Additionally, they are safe to call without grabbing the TransactionManager's
instance with local variable. Therefore, if EditorBase has methods to
add/remove transaction listener to/from its transaction manager, we don't
need EditorBase::GetTransactionManager() anymore.
So, this patch adds AddTransactionListener() and RemoveTransactionListener() to
EditorBase and TransactionManager, and remove
EditorBase::GetTransactionManager() (not nsIEditor's one).
MozReview-Commit-ID: FkPa1YgfagD
--HG--
extra : rebase_source : 8ef7796136804e3d7604c5ca5417fb3379606b93
nsIEditor::EnableUndo() and nsITransactionManager::Clear(),
nsITransactionManager::SetMaxTransactionCount() are called a lot but they are
virtual and some of or all of them are called once. There should be each
non-virtual method to do what each root caller wants. Therefore, this patch
adds EditorBase::EnableUndoRedo(), EditorBase::DisableUndoRedo(),
EditorBase::ClearUndoRedo(), TransactionManager::EnableUndoRedo(),
TransactionManager::DisableUndoRedo() and TransactionManager::ClearUndoRedo().
Note that this patch makes TransactionManager won't clear mUndoStack nor
mRedoStack if mDoStack is not empty. This is checked only by
TransactionManager::SetMaxTransactionCount() but according to the comment,
TransactionManager::Clear(), TransactionManager::UndoStack() and
TransactionManager::RedoStack() should check it too.
MozReview-Commit-ID: 6qBZOQNwdhw
--HG--
extra : rebase_source : 3249137f7acca0b4698713ab732774140bcc27e8
EditorBase::Undo() and EditorBase::Redo() implement only undo/redo function.
TextEditor::Undo() and TextEditor::Redo() call them with calling some
notification methods. However, this causes redundant AutoRules instance
creation and doesn't make sense to separate them under current design.
Therefore this patch merges them into TextEditor. Unfortunately, they are
XPCOM methods but we cannot implement proper overloads of non-virtual since
they are already minimized design. Fortunately, reducing only one virtual
call per undo/redo isn't so effective. So, let's keep simpler design.
Additionally, this patch makes them stop committing composition because
Chrome does not commit composition even if "undo"/"redo" is requested with
execCommand() during composition and it doesn't make sense to try to
undo/redo after committing composition since first undoable transaction
becomes the committing composition and committing composition causes
removing all transactions from redo transaction queue.
MozReview-Commit-ID: 78qlV2I9Lzk
--HG--
extra : rebase_source : 545c787d47fe02bf7e085be9d3ae028816750e69
Now, both TransactionManager.h and TransactionStack.h are exposed. So,
TransactionManager::GetNumberOfUndoItems() and
TransactionManager::GetNumberOfRedoItems() can be rewritten with non-virtual
inline methods because they just return mUndoStack.GetSize() and
mRedoStack.GetSize(). Then, we can implement EditorBase::NumbeOfUndoItems(),
EditorBase::NumberOfRedoItems(), EditorBase::CanUndo() and
EditorBase::CanRedo() as inline methods.
MozReview-Commit-ID: 3CJd0VrlvFY
--HG--
extra : rebase_source : 6848d80a395f1c161e10bfb50d15bd63de288095
Then, all classes in editor/txmgr is now in mozilla namespace and all
headers which are included by other directory are now exposed. So,
we can remote local includes from other directories now.
MozReview-Commit-ID: Kdb1c4Hp9Sy
--HG--
rename : editor/txmgr/nsTransactionStack.cpp => editor/txmgr/TransactionStack.cpp
rename : editor/txmgr/nsTransactionStack.h => editor/txmgr/TransactionStack.h
extra : rebase_source : 75035e9e618680af9188820595ab050b612a456f
Its return value is never used, and most implementations return nullptr anyway.
MozReview-Commit-ID: 8rxC053mmE8
--HG--
extra : rebase_source : 61a0b8b1373396182efd27d3c01b96e5e5541364
Giving '0' (literal zero) to RefPtr is now ambiguous, as both
RefPtr(decltype(nullptr)) and RefPtr(T*) could be used.
In any case, our coding standards mandate the use of 'nullptr' for pointers.
So I'm changing all zeroes into nullptr's where necessary.
MozReview-Commit-ID: A458A4e9for
--HG--
extra : rebase_source : bc4107ce1767329e3dddfe34c55ba79202192e06
The .cpp files under editor/txmgr have a lot of unnecessary empty lines and they make reading the code more difficult because fewer lines can be displayed in screen. This patch removes such unnecessary lines and C like variable declarations at start of some methods because they also waste lines and make them unclear where is the first user of them.
MozReview-Commit-ID: GXGkdoLJ4Jw
--HG--
extra : rebase_source : 449e191553cff8c875565802d930abe2a6dba4a6
For conforming to our coding rules, |result| of nsresult variants should be renamed to |rv|.
MozReview-Commit-ID: Bk8CyLAnvXQ
--HG--
extra : rebase_source : dd3dc34a032f22abf3fd7f85556b47ffbeec55b2
Currently, editor code uses following style (or similar style) in a lot of places:
if (foo)
{
}
else
{
}
This patch fixes this as conforming to our coding rules, i.e., it becomes:
if (foo) {
} else {
}
Additionally, this fixes other odd control statements in the files which include above issue because it's difficult to find following issues with searching the files:
* if (foo) bar;
* if (foo) { bar; }
* if (foo)
bar;
Finally, if it becomes much simpler than current code, this patch rewrites existing code with "early return style". But this case is only a few places because this is risky.
MozReview-Commit-ID: 2Gs26goWXrF
--HG--
extra : rebase_source : 603f9003a3566b3203bdeb27dc73ac33502d2853
This removes the unnecessary setting of c-basic-offset from all
python-mode files.
This was automatically generated using
perl -pi -e 's/; *c-basic-offset: *[0-9]+//'
... on the affected files.
The bulk of these files are moz.build files but there a few others as
well.
MozReview-Commit-ID: 2pPf3DEiZqx
--HG--
extra : rebase_source : 0a7dcac80b924174a2c429b093791148ea6ac204
Using std::deque here causes problems for libc++ builds; TestTXMgr on
OSX 10.6 opt times out when libc++'d std::deque is used. Running the
test locally shows that the test process consumes gigabytes (!) of
memory and is thus reduced to swapping, rather than making any progress.
libc++'s std::deque also appears to be slightly slower in said test that
even OSX libstdc++'s std::deque. (Admittedly, this test is artificial.)
Let's sidestep the slowness of libc++'s std::deque by rewriting
nsTransactionStack to use nsDeque rather than std::deque. Not only does
this change enable OSX 10.6 tests to pass, it also makes TestTXMgr
significantly faster in opt builds: TestTXMgr is anywhere from 25-60%
faster (depending on the platform) than when using std::deque from
libstdc++ or libc++.
Checking to see whether the stack is empty is a reasonably common
operation. We can make the code clearer and more efficient (no need to
refcount to check the emptiness of the stack) in several cases.
We unnecessarily refcount in several places when pushing items onto an
nsTransactionStack. Adding an already_AddRefed overload will enable
avoid those unnecessary refcounts.
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
--HG--
rename : mfbt/nsRefPtr.h => mfbt/RefPtr.h
The patch removes 455 occurrences of FAIL_ON_WARNINGS from moz.build files, and
adds 78 instances of ALLOW_COMPILER_WARNINGS. About half of those 78 are in
code we control and which should be removable with a little effort.
--HG--
extra : rebase_source : 82e3387abfbd5f1471e953961d301d3d97ed2973
They are kept around for the sake of the standalone glue, which is used
for e.g. webapprt, which doesn't have direct access to jemalloc, and thus
still needs a wrapper to go through the xpcom function list and get to
jemalloc from there.
There are, sadly, many combinations of linkage in use throughout the tree.
The main differentiator, though, is between program/libraries related to
Gecko or not. Kind of. Some need mozglue, some don't. Some need dependent
linkage, some standalone.
Anyways, these new templates remove the need to manually define the
right dependencies against xpcomglue, nspr, mozalloc and mozglue
in most cases.
Places that build programs and were resetting MOZ_GLUE_PROGRAM_LDFLAGS
or that build libraries and were resetting MOZ_GLUE_LDFLAGS can now
just not use those Gecko-specific templates.
Ehsan Akhgari
Masayuki Nakano
Chris Peterson
Marco Castelluccio
Gerald Squelart
Florian Quèze
Andi-Bogdan Postelnicu
Benjamin Smedberg
Michael Layzell
Tom Tromey
Jonathan Watt
Nathan Froyd
Sylvestre Ledru
Birunthan Mohanathas
Nicholas Nethercote
Mike Hommey
Andrea Marchesini
Trevor Saunders