void UnoControl::setDesignMode calls setVisible on the XWindow
so it seems sensible that FmXGridPeer::setDesignMode should
also do so. The desirable side effect is that moving the window
in design mode then is moving a un-shown window so nothing
is invalidated so the paint loop doesn't happen.

Change-Id: Ic5a4ba62590372bec15a29b993b73d6dfb94a8a9
(cherry picked from commit a7816853)
Reviewed-on: https://gerrit.libreoffice.org/19906Reviewed-by: Lionel Elie Mamane <lionel@mamane.lu>
Tested-by: Jenkins <ci@libreoffice.org>
(cherry picked from commit a08a5ff7)

Change-Id: I937ed893dfd9919bf48b8fd38a1b83c524200532
(cherry picked from commit 6128c10f)
Reviewed-on: https://gerrit.libreoffice.org/19911Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Michael Stahl <mstahl@redhat.com>
(cherry picked from commit 82c0f2eb)

The missing convertMm100ToTwip() calls mispositioned the line shapes
inside the groupshape in case of using drawingML inside DOCX files.

(cherry picked from commit 11129d89)

Conflicts:
	sw/qa/extras/ooxmlimport/ooxmlimport.cxx

Change-Id: I0c9d652de43d779f9073a8bfe22866ce4f31d0fa
Reviewed-on: https://gerrit.libreoffice.org/19946Reviewed-by: Caolán McNamara <caolanm@redhat.com>
Tested-by: Caolán McNamara <caolanm@redhat.com>
(cherry picked from commit ca8cdb9a)

Remove listener we add, and don't crash on callback for disposed dialog.

Reviewed-on: https://gerrit.libreoffice.org/19936Reviewed-by: Caolán McNamara <caolanm@redhat.com>
Tested-by: Caolán McNamara <caolanm@redhat.com>
(cherry picked from commit 45c6d948)

Conflicts:
	svx/source/dialog/imapdlg.cxx

Change-Id: Ia6a1bddd212cb28a0331469e8e87324346d02841

(cherry picked from commit d5016b01)

Change-Id: Ief31d8153fdbf91cdd29df5ac7801bd88a98542e
Reviewed-on: https://gerrit.libreoffice.org/19955Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Caolán McNamara <caolanm@redhat.com>
Tested-by: Caolán McNamara <caolanm@redhat.com>
(cherry picked from commit d11d0915)

implementation can indeed be empty when cppu::writeSharedLibComponentInfo calls
getEnvironmentFromModule(mod, curEnv, "", ""), which in the final if-branch
calls getEnvironment with implementation = "".  Happens when adding desktop/qa's
test-active.oxt.  Regression introduced with
ae3a0c8d "Add .component <implementation
constructor='...' feature."

(cherry picked from commit 97464be9)
Conflicts:
	cppuhelper/source/shlib.cxx

Change-Id: Ia70958e4aa18d378ff35063bdf6187f6e073b765
Reviewed-on: https://gerrit.libreoffice.org/19956Reviewed-by: Caolán McNamara <caolanm@redhat.com>
Tested-by: Caolán McNamara <caolanm@redhat.com>
(cherry picked from commit 2466c85c)

Change-Id: Ief5c2effbddebe29fc7a4d38ed0af4fed72de40e
(cherry picked from commit e08fbe4c)
Reviewed-on: https://gerrit.libreoffice.org/19915Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Caolán McNamara <caolanm@redhat.com>
Tested-by: Caolán McNamara <caolanm@redhat.com>
Reviewed-by: Michael Stahl <mstahl@redhat.com>
(cherry picked from commit b4533424)

for gtk3. Probably also a fairly dumb idea for gtk2, but with the
extra unblocking opportunities it probably just "works out ok" in
that model.

Change-Id: I3d4716c691fc84a5e68c210fee51f9e31cfacc02
(cherry picked from commit f9dfd012)
Reviewed-on: https://gerrit.libreoffice.org/19929Reviewed-by: Adolfo Jayme Barrientos <fitojb@ubuntu.com>
Tested-by: Adolfo Jayme Barrientos <fitojb@ubuntu.com>
(cherry picked from commit 11d0bacd)

Change-Id: I16ddac82a76a1e665797e4be192d3642f2660d9f
(cherry picked from commit 82abd23f)
Reviewed-on: https://gerrit.libreoffice.org/19933Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Adolfo Jayme Barrientos <fitojb@ubuntu.com>
(cherry picked from commit 5a1d267a)

VclPtr related, dialog dtor is never called, so timer
is never stopped and timer refers to deleted parent
data

(cherry picked from commit 0943adb2)

Change-Id: I804ba082817802c0017e61c6b86c5d3694995ea8
Reviewed-on: https://gerrit.libreoffice.org/19926Reviewed-by: Maxim Monastirsky <momonasmon@gmail.com>
Tested-by: Maxim Monastirsky <momonasmon@gmail.com>
(cherry picked from commit 33df62ef)

The problem was always masked out the combination ALT+CTRL, this only
needed when press the ALTGR key, but the other ALT-Key (left side) is
valid.

Change-Id: I8a6f953f519627418fb48ee8d1d68b3cbf797250
Reviewed-on: https://gerrit.libreoffice.org/19923Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Caolán McNamara <caolanm@redhat.com>
Tested-by: Caolán McNamara <caolanm@redhat.com>
Reviewed-on: https://gerrit.libreoffice.org/19932
(cherry picked from commit 0ff2015b)

Change-Id: Id584edf091aebd95bb71d7efd2df6fdaa511c837
Reviewed-on: https://gerrit.libreoffice.org/19875Reviewed-by: Adolfo Jayme Barrientos <fitojb@ubuntu.com>
Tested-by: Adolfo Jayme Barrientos <fitojb@ubuntu.com>
(cherry picked from commit 131a1962)
Reviewed-on: https://gerrit.libreoffice.org/19920Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Caolán McNamara <caolanm@redhat.com>
Tested-by: Caolán McNamara <caolanm@redhat.com>
(cherry picked from commit ec0da57d)

Regression from commit 866a4436 (oox:
handle textboxes in ShapeContextHandler::endFastElement(), 2014-06-04),
the problem was that code in ShapeContextHandler::endFastElement()
assumed that all the XShape implementation it may see provides a TextBox
UNO property (which is specific to Writer) of type bool, but this is not
true.

Fix the problem by assuming false in case it has an other type.

Testcase can be a simple "does it load" one, as in the meantime
exceptions such as trying to read a bool from an uno::Any make the DOCX
XFilter::filter() implementation return false, unlike in the originally
reported LibreOffice 4.4 version that just silently discarded the
remaining content of the document.

Change-Id: I81781a7ec28d085ceee50d386b8eda023f83c29f
(cherry picked from commit 4cae3689)
Reviewed-on: https://gerrit.libreoffice.org/19905Reviewed-by: Michael Stahl <mstahl@redhat.com>
Tested-by: Michael Stahl <mstahl@redhat.com>
(cherry picked from commit 2c43db1e)

blendBitmap24 assumes bitmap is 2 lines high for doing
interpolation. For bitmaps of height 1, really nothing to interpolate
there. Fix cornercase by 'blending' between the very same line instead.

Reviewed-on: https://gerrit.libreoffice.org/19863Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Norbert Thiebaud <nthiebaud@gmail.com>

Cherry-picked from 99e3ab6e

Change-Id: I9b94000aa563e525d0bb2418346ad2c86af26df8
Reviewed-on: https://gerrit.libreoffice.org/19888Reviewed-by: Katarina Behrens <Katarina.Behrens@cib.de>
Tested-by: Katarina Behrens <Katarina.Behrens@cib.de>
(cherry picked from commit 5a86268e)

Change-Id: I85000851f82ea7cdc4b536683adbc8570de9af7e
(cherry picked from commit 396b5f41)

We need the callback to be able implement the formula bar

Change-Id: I1c78ab0b9ed9304c0465a9993a7101f8efb91052

Conflicts:
	include/LibreOfficeKit/LibreOfficeKitEnums.h
	libreofficekit/source/gtk/lokdocview.cxx

Conflicts:
	libreofficekit/qa/gtktiledviewer/gtktiledviewer.cxx
	libreofficekit/source/gtk/lokdocview.cxx

We need to know the client's view level to correctly handle the mouse
events in calc. PaintTile() set a zoom level that corresponds to the
requested tiles and previously postMouseEvent would call SetZoom(1,1).
Now we can make use of knowing the client's view level and call
SetZoom() with the correct parameters

Change-Id: I34b5afcdcc06a671a8ac92c03e87404e42adf4cd

Conflicts:
	sc/source/ui/unoobj/docuno.cxx

Conflicts:
	libreofficekit/source/gtk/lokdocview.cxx
	libreofficekit/source/gtk/tilebuffer.hxx

And no need to copy it to be able to dump it.

Change-Id: I5c0782c489a5d7c24173a81cc82efd195c4aa56e
(cherry picked from commit 44c7272f)

Fails without e.g. the last hunk of commit
1ba9d7fd (sw lok: forward key events to
annotation window if necessary, 2015-11-12).

Change-Id: I7f39530881f6141fea956b751aa57eb2bdcadcb2
(cherry picked from commit 1118689e)

LOK is Linux-only at the moment, don't bother with disabling each and
every unit test on Mac/Windows for now.

Change-Id: I2ff1ed47251c16ec6a8d43138789480d95ea720e
(cherry picked from commit 18dfcbb1)

As long as we don't tweak the map mode of the comment widgets
permanently we also have to disable the selection engine's timer, as
that would emit events without the correct map mode: so disable that for
the LOK case for now.

(cherry picked from commit b3dc0d7c)

Conflicts:
	sw/inc/SidebarWin.hxx
	sw/source/uibase/docvw/SidebarTxtControl.hxx

Change-Id: If377ff2f064c30feb473f153f9d5b29b8ace7113

With this, pressing ctrl-alt-c to create a new comment has the correct
cursor position.

Change-Id: Icb8d708dab015d8ffa9bcfe28de66238a75b50bc
(cherry picked from commit d78432e9)

SwPostItMgr::registerLibreOfficeKitCallback() already took care of
informing existing outliners, this commit gives a callback to newly
registered ones as well.

(cherry picked from commit dacc616f)

Conflicts:
	sw/inc/drawdoc.hxx
	sw/source/core/draw/drawdoc.cxx

Change-Id: I660dcb54231a9d404bf80b4284003d119dae6a5c

So that it's possible to drag the text selection start/end handles in
comment text when there is an existing selection.

(cherry picked from commit e320d070)

Conflicts:
	sw/source/uibase/docvw/SidebarWin.cxx

Change-Id: I3acc4770928d4f385f0ca09a2484a9e112409907

With this, newly typed characters show up instantly in comments, not
only after changing the zoom level.

(cherry picked from commit 4cbbaf57)

Conflicts:
	sw/source/uibase/docvw/SidebarTxtControl.hxx

Change-Id: I1470db1ec03cc415917375f1f95434cf0944e559

LOK always works in absolute twips (origo being the top left corner of
SwEditWin), so not only the callbacks have to translate relative twips
to absolute ones, but the opposite have to be done for mouse event
coordinates.

With this, clicking at a random position inside a comment places the
blinking cursor at a reasonable position, not always at 0,0.

(cherry picked from commit 57972554)

Conflicts:
	sw/source/uibase/docvw/SidebarWin.cxx

Change-Id: Ic8d20f177acd9e1908acf17698c53a1470bd4aec

Change-Id: Ic052544b2835181652732b8de9eaf79572a9db6e
(cherry picked from commit 022c716f)

With this, selections of Writer comment text show up at the correct
position, not at the top left corner of the window.

Change-Id: If865503d9a02a27730e382d65c42c706dd533a93
(cherry picked from commit 22e97c13)

With this, it is possible to click inside a comment (and get a blinking
cursor inside a comment), and also possible to use the arrow keys to
native around and still get correct blinking cursor position.

(cherry picked from commit 5fb91dfd)

Conflicts:
	sw/inc/SidebarWin.hxx
	sw/source/uibase/docvw/SidebarTxtControl.hxx
	sw/source/uibase/docvw/SidebarWin.cxx

Change-Id: I29eb1e60e4e571151f0b18bec8cf765ea09af09f

Writer comments are separate widgets, but we want to have coordinates in
absolutes twips, so give Writer a chance to inform us about the delta.

For now only do this in case the map unit is twips, as Impress sets the
origin already.

Change-Id: Idf340944165e44b0888c00965d6be7798712ff76
(cherry picked from commit e988ab84)

This is primarily for tiled editing that doesn't know about multiple
cursors, so if both the postit cursor and the main cursor is shown for a
short period of time, then the main one is hidden, then gtktiledviewer
still shows the main one.

OTOH if we can avoid showing the cursor in general just to hide it a bit
later in general, let's not do that.

With this, the "cursor is shown at the comment anchor, not inside the
anchor when clicking on the postit" bug disappears.

Change-Id: I2383292c5f84604dc8b126510b0797b8426920ae
(cherry picked from commit 1c482cb5)

And to allow proper reaction by the annotation windows, inform them when
a LOK callback is registered.

With this, it's possible to modify the contents of annotations via LOK.

(cherry picked from commit 1ba9d7fd)

Conflicts:
	sw/inc/PostItMgr.hxx
	sw/inc/SidebarWin.hxx
	sw/source/core/view/viewsh.cxx
	sw/source/uibase/docvw/SidebarTxtControl.hxx
	sw/source/uibase/docvw/edtwin.cxx

Change-Id: I4489941512197880940e20cbaeb0b47a7a6f26fc

LOK sends all mouse events to SwEditWin, so add initial hit testing in
its mouse handler to forward the mouse events to the right VCL widget.

(cherry picked from commit b7ecf627)

Conflicts:
	sw/inc/PostItMgr.hxx
	sw/source/uibase/docvw/SidebarWin.cxx
	sw/source/uibase/docvw/edtwin.cxx

Change-Id: I67e8e19f47156261fd7c7eafd4e63f743e0c4ce9

So SwSidebarWin member functions don't get twip values when they call
GetPosPixel().

Change-Id: Ied4ff7f49d0320766b045d78f731900af92d37ef
(cherry picked from commit c4ef30ea)

It is only needed when tiled rendering.

(cherry picked from commit 88f48668)

Conflicts:
	sw/source/uibase/docvw/SidebarWin.cxx

Change-Id: Ie1668f5f3d4d17abc212e2262a6c155dcb855d2e

With this, e.g. a line above a note (with width of 180 px) is also 180
px, not 12 px.

(cherry picked from commit 09640807)

Change-Id: I7c4eeda1bc904242dc298013411b9671ba0f2149

The map mode is in general disabled during tiled rendering, so mouse
positions can be sent in using twips, but here we have to temporarily
enable it, otherwise the width will be returned always in pixels.

With this, the gray background of the sidebar has the proper width, not
e.g. fifteenth of the expected value (using default zoom).

Change-Id: I4380ee0ba6bcda97cf71735161dbdc826e7a2532
(cherry picked from commit 6c3dbdbc)

Change-Id: Ie682cfe97b8c48bc66575b89875ca617dbb8ab4e
(cherry picked from commit b3b9f2c5)

(cherry picked from commit 9b504ad3)

Conflicts:
	sw/inc/docufld.hxx

Change-Id: If15ca89f2b7c54441f076eb233c4e1414dac25c0

Change-Id: I16e51e074704026a45471e7a08c3b96846d44053
(cherry picked from commit 5f687598)