Apparently causes problems for some people.

Unfortunately the scp2 entries are "non-standard" so i've had to add
another variable to let the installer find it.

Change-Id: I7a01160369ed8dc401625c06760008341afcee67

This reverts commit 0cf61137.

This fixes the saving / loading of RTL setting of the table, and changes
the UI behavior when you click the LTR/RTL buttons with table border selected
a bit.

Change-Id: I1abcebdad3cdb72307535f82d973eb52418f2787

Change-Id: I756ac431dfcd6a4d6a365fc38b465f4b588ddd79

Change-Id: Ifc883acd91108981a190f35b19648ec3a747db7b

I got scrolling to respond to the JNI scroll method call.  (This is not yet
the gesture I want to implement (scroll by pixels), but at least it’s a
start.)

At first I tried to do some hacking in the C++ part, then found the problem
was in the Java parameters, which set the mouse whell delta too big.  When
passed the value 1 or -1 to the Y value of AppSupport.scroll(int x, int y),
the scrolling is reasonable.

Change-Id: Id8fa0ada1a035760b7b05bf21325d0e51ef28fdc

Change-Id: I3da23e991a70e12bdfefeda5f5e7669ab111cfb4

Change-Id: I9895ca15680c01da17a59cb7da51478d5f61657e

Change-Id: If20e645d8d708ebe17e2d9455c896c94592993bb

Change-Id: Ida59dcca7c9a0305aecddaa68f4e585321144458

Need to save out the VBA project storage to ensure we don't get any
nasty warnings trying to open ppt files containing old controls

Change-Id: Ic2ff3ac5863ee3b901b1ee48f7d45e82dcb27ff9

tab stop positions of styles

(cherry picked from commit 7b20978d)

Change-Id: I068a8dced069c4676623723f16ae4d85997f086d

Change-Id: I1f40bba348a3c0be7a892b56a9bf1b7da0b4435f

Can be identical to the one for version 21.

Change-Id: Ic974dd9ed06947e6be1c04598d3a7e715795e202

Change-Id: I4e3b56fd8c088983a8a01eb6ccf88691d6fd93ef

(no fractional second)

Change-Id: I8bf448783b24fceae9a750f40887d6296999b203

fractional second allowed in timestamp. But not in time. Don't ask why.

Change-Id: I5415889ab4cf4835b8b0db7363b7d9eb0a332393

Change-Id: I0aba2344afde94226cfbac5b3221e827e6774b3a

std::ostringstream::width is not sticky

Change-Id: I32d77bec68506b7691a4f86dadb24e62fdc13d42

Change-Id: I72a27a05c9e3e49b17817c2566fbd7f02b7835c5

Change-Id: Icbe5ac8e20b5a05fdba4ed7b8a287c3d9e94d2da

Change-Id: I3c0282d731082c32558a527110762dffad01b44b
Reviewed-on: https://gerrit.libreoffice.org/4558Reviewed-by: Markus Mohrhard <markus.mohrhard@googlemail.com>
Tested-by: Markus Mohrhard <markus.mohrhard@googlemail.com>

With JDK upgrade to 1.7.x focus gain/lost events were dropped and thus not
discoverable by AT tools.

The problem has nothing to do with different packaging of JAB (2.0.3 is
included with JDK 1.7u6 and higher).

The reason for the event droping was a new check in
DefaultKeyboardFocusManager.java that was introduced in JDK 1.7.x:

@@ -294,18 +271,10 @@ public class DefaultKeyboardFocusManager
        extends KeyboardFocusManager {

-      if (!(newFocusedWindow.isFocusableWindow()
-               && newFocusedWindow.isVisible()
-               && newFocusedWindow.isDisplayable()))
-         {
-             // we can not accept focus on such window, so reject it.
-             restoreFocus(we);
-             break;
-         }

With this check in place it is assumed that Component.isDisplayable() must
return true, for event to be dispatched. But particularly
`Dialog.isDisplayable()` didn't and thus all tab key events (focus traversal)
were dropped.

Many thanks to Jamie and Stuart for helping out with reproducing and tracking that down!

Change-Id: I4876773199ca06a6e73213401692e8275a1f3b22

Change-Id: I7a7e13104e7256da9a82e16e14ddfa04e35e7f8a
Reviewed-on: https://gerrit.libreoffice.org/4525Tested-by: Eike Rathke <erack@redhat.com>
Reviewed-by: Eike Rathke <erack@redhat.com>

Raptor already sets up all 4 error handlers in xmlSAXHandler so why it
would need the global ones in addition to that is a mystery anyway.
Messing with libxml2's globals can only cause trouble.

Change-Id: I2935efe5c4cd75d48cc4ecdeaa8437e91b48349e

Project: help  f92fe458ddf0b7a79a11f426eff8f271639c6ada

at the paragraph

- some minor reformatting and cleanup of older comments

(cherry picked from commit e4578fcc)

Conflicts:
	sw/source/core/text/txttab.cxx
	sw/source/filter/ww8/ww8par.cxx

Change-Id: I345a20c78f26be60de3eb01ac6f5303de46c2fc1

Change-Id: Id1028f87a6dded7d2c27fe8d793342574d06e3f9
Reviewed-on: https://gerrit.libreoffice.org/4528Reviewed-by: Michael Stahl <mstahl@redhat.com>
Tested-by: Michael Stahl <mstahl@redhat.com>

Change-Id: Id90ef9f43578d952fd8d7872d4fd2fd622d07dbd
Reviewed-on: https://gerrit.libreoffice.org/4529Reviewed-by: Michael Stahl <mstahl@redhat.com>
Tested-by: Michael Stahl <mstahl@redhat.com>

Change-Id: Icc70cc02d1b3b154fc1cab0419d89b437e2a366f
Reviewed-on: https://gerrit.libreoffice.org/4530Reviewed-by: Michael Stahl <mstahl@redhat.com>
Tested-by: Michael Stahl <mstahl@redhat.com>

see origin f1ecd5ff, f9d08dd5c9af8f8128371540e
& follow up 0a686a9b

Change-Id: I1a10de6984f7ed78efdf6f32bfca14b433b24405

Change-Id: I0ca0edf4fc8f19adb1859eb4a679fee6480ea295

There's a document which has coordsize="6094,0" path="m444,424r6094,e" ,
where the 0 is somewhat questionable (and the spec doesn't seem to say
anything about it), but given it's a horizontal line, vertical coordinate
space size shouldn't matter, so treat it as 1, making the space visible
and avoiding division by 0 in our code.

Change-Id: I53725cd73a699b8e6d3c96e8fc39c276d082bc0c

a) make sure backcolor is skipped when libreoffice backcolor
is default. Libreoffice default is 0 ( but if we write that out mso
reads it as black )
b) don't take the label default backcolor in exchange for the
libreoffice default. Libreoffice default backcolor for label is white,
mso's default is rather different ( and never seems to be used )
c) re-enable compObj export ( seemed this was disabled unintentionally
except for buttons )

Change-Id: Ie68267809cee2e92bd06e88e698f17a10eeeb6ad

Reported by: Du Jing
Patch by: Jianyuan Li
Review by: sunying

There were multiple problems here:

- \f had a special handler for the listlevel destination, \af had not
- in case of multiple \f or \af, the first one is used for the bullet
  font
- in case only \af is used for the bullet font, its value should be used
  for \f as well

Change-Id: I6631504c1aa9f2e0792a3469a5fdce5b7bd49518

change for all text

- group intrinsic actions of language change for all text into one Undo action
- secure <SwRootFrm::CalcRects(..)> - catch NULL pointer
- correct <CursorGuard> - really restore the cursor

(cherry picked from commit 4207db47)

Conflicts:
	svl/inc/svl/undo.hxx
	svl/source/undo/undo.cxx
	sw/inc/IDocumentUndoRedo.hxx
	sw/source/core/edit/edundo.cxx
	sw/source/core/inc/UndoManager.hxx
	sw/source/core/layout/trvlfrm.cxx
	sw/source/core/undo/docundo.cxx

Change-Id: I972988bbc21a519e0956ff196aa93a46287b9a2e

... always so don't need to check for it.

Change-Id: I4bbc2a44a9c3d4ebde766a64a6ab78f7ead9c7af

All the comments of the form //-------- are removed from /svx/source/accessibilty/AccessibleControlShape.cxx"

Change-Id: I7ae9f656939e2512ce503a5ba103d765594f20ac
Reviewed-on: https://gerrit.libreoffice.org/4543Reviewed-by: Miklos Vajna <vmiklos@suse.cz>
Tested-by: Miklos Vajna <vmiklos@suse.cz>

The option should be used also at link time (as e.g. the gcc man page says),
but cairo doesn't do so, and since -flto makes clang generate .o with LLVM
bytecode, plain ld chokes on it.

Change-Id: Ia5ce9a98afa4f88b8d0943a5971450d7009af913