When scanning the entire data range in an attempt to determine where
the first data row is, we shouldn't stop at the first column.  We should
scan all data columns to make sure we pick the lowest value row
position among all columns.

Change-Id: I693ec169857b181c6b4237efcba3fa6a46eaa531

fixes circular vcl<->xo dependency

Change-Id: I5a9eb1f48044a2d43901b40f695ea2a93c70388e

This reverts commit 7bce8c2e.

It printed help of the plain gerrit tool which always confused me :-)

Change-Id: Ied81688ada4b536486d998709230fc24fa3c9b62

they are not longer forced from configure

Change-Id: I8ea37b7233138de2900caf231762eefec0ba6563

This complements change a9fc1456
The Translation-compensation of that fix didn't work for flipped shapes.
Also, vertically flipped shapes made problems.
Also 360 °C = 0°C (drawingml.hxx)
I didn't dare to fix it earlier because there is some problem with the selection frames.
But that is an import bug, not caused by some export stuff!
This import bug makes the manipulation rectangles around the shapes flipped & turning in the wrong direction, perhaps due to
CroftCro$oft clockwisification!

Change-Id: I05aa036d31cd5d9a8b194897b3d92ec12d6d2c61
Signed-off-by: Lennard <Wasserthal@nefkom.net>
Reviewed-on: https://gerrit.libreoffice.org/1250Reviewed-by: Petr Mladek <pmladek@suse.cz>
Tested-by: Petr Mladek <pmladek@suse.cz>

Change-Id: I044033ca41dea75dc438bc3989d067fc8e70b9fc

As far as I can tell, there is not circular dependency, but make
complains, and only when invoked from toplevel, not from tail_build.
Looks like gbuild problem, but do a hackish change to make
unbreak clean compilation, for now.

Change-Id: I445ba343f9eaa988c60c288bf5fc1c5d1c7b67a5

gb_Side is set either from toplevel Makefile or from any module Makefile,
but not when starting build from tail_build

Change-Id: I4df6eb221f5b778ce3c0234251f8dbbb56645028

...where for whatever reason (TODO: investigate) they keep failing at least with
my local Mac OS X 10.7 build at
toolkit.AccessibleFixedText::com::sun::star::accessibility::XAccessibleText:

LOG> getCharacterBounds(-1)
LOG> Expected exception
LOG> getCharacterBounds(8)
LOG> Expected exception
LOG> getCharacterBounds(0)
LOG> Text at this place: F
LOG> Character bounds outside component
LOG> Character rect: 1, 3, 6, 11
LOG> Component rect: 0, 22, 100, 12
LOG> getCharacterBounds(1)
LOG> Text at this place: x
LOG> Character bounds outside component
LOG> Character rect: 6, 6, 10, 8
LOG> Component rect: 0, 22, 100, 12
LOG> getCharacterBounds(2)
LOG> Text at this place: e
LOG> Character bounds outside component
LOG> Character rect: 15, 6, 7, 8
LOG> Component rect: 0, 22, 100, 12
LOG> getCharacterBounds(3)
LOG> Text at this place: d
LOG> Character bounds outside component
LOG> Character rect: 22, 3, 7, 11
LOG> Component rect: 0, 22, 100, 12
LOG> getCharacterBounds(4)
LOG> Text at this place: T
LOG> Character bounds outside component
LOG> Character rect: 30, 3, 8, 11
LOG> Component rect: 0, 22, 100, 12
LOG> getCharacterBounds(5)
LOG> Text at this place: e
LOG> Character bounds outside component
LOG> Character rect: 39, 6, 7, 8
LOG> Component rect: 0, 22, 100, 12
LOG> getCharacterBounds(6)
LOG> Text at this place: x
LOG> Character bounds outside component
LOG> Character rect: 45, 6, 10, 8
LOG> Component rect: 0, 22, 100, 12

Change-Id: I690ffc773b53e0f19536605e3cb89280bc98c19a

...it now matches the duration of another nearby timeout in the code, and it
happened to cause a parallel "make check" to sometimes fail on a slow MacBook.

Change-Id: If8bddb82a05c46637fdc2fcc3373411fcb37a466

Change-Id: I29dcc26eb13222ef4348ab3281a19b2e6f8c46f2

Change-Id: I61731f472405e46828204ffb1cacf1c9e6a9b5bf

Change-Id: Ib933d150d5d82383d23619857b2797a63be28626

Change-Id: I574c523d37d54a05aa79d2ccf52e0727e1a1d744

There are still places that should be improved a bit, but this works.

Change-Id: Ieb7947a294ec95b6fd8cec2e8c4bc731e2594c42

Change-Id: I347790e1ca2ee6a27303d3644aa1459ff76ba4b0

I'm not quite sure why the code has to do it this complicated way,
but at least it's just in one (ok, two) places. Additionally, it
seems all those checks are needless paranoia, so just assert.

Change-Id: I9f0d4ecc5aec6995eb66ae553a4bd92cc5450b86

Although it seems reading works even when opened just for write ...

Change-Id: I57039e7b72dd466e1b087c81377b2b2dcd26fb5c

So that SvXMLImport::GetSourceStorage() returns something useful.

Change-Id: Iecb2044336d45eefcb2b0ca5cfda5bc173d4e71e

Change-Id: Id41200667089d4f42f7999060fd97545bb069788

http://lists.freedesktop.org/archives/libreoffice/2012-September/038163.html

Change-Id: I41a195180411800f72b8e251b5c19553a51cde44

Otherwise e.g. informing about new fonts when loading a document
with embedded fonts doesn't work. I'm not quite sure why this
should be limited to only printers or pdf, so this may be temporary
until a problem shows up. There'a also a warning about refcount
in ImplFontCache::Invalidate(), which seems to be bogus, because
invalidating caches shouldn't have effect on anything except for speed,
but with the way font handling in vcl is designed I wouldn't be
surprised if ImplFontCache wasn't actually a cache.

http://lists.freedesktop.org/archives/libreoffice/2012-September/038163.html

Change-Id: Ib68c0c10a0f5340a532c5c59396ef2cfbc588b5f

Change-Id: If3607886050c4e2e649e5678cc39f290b122f537

Not that I know why the directory had to be renamed in the first place.

Change-Id: I3e4d07591876636ae4930240ac1acc68522fb94f

Sign also all the dylibs and frameworks in the bundle.

Change-Id: I7f67b9d7eda0204b24e2ea2ef44a53fb8db0f8aa

Change-Id: I7d63af2ddb67104daaddcb5f111e26269b9b5030

Change-Id: I519e4cfb7821644f243e0c4770f74f6ee6657168

Change-Id: If819c60a4d1fb3af2f4f87e0e4acfd00ecad95d3

Otherwise the build --without-system-stdlibs fails; This option is used
for the universal Linux build

Change-Id: I0f7d87bb161d50748fee05f9c3f6a4b87b788aa4
Reviewed-on: https://gerrit.libreoffice.org/1938Reviewed-by: David Tardon <dtardon@redhat.com>
Tested-by: David Tardon <dtardon@redhat.com>

Change-Id: I4531ab82a52e3da41f12409185f7c0730381aa23

Change-Id: Ida6b3820989acaa2f1bac43aa40276eaa12c8aae

Normally such a page break is surrounded by text, and text() already
calls checkNeedPap(), but in case the whole paragraph contains a page
break only, we need it here as well.

Change-Id: I3961272d77c743e0e6da2d5052e57cec5f5937e7

After all, if somebody else wants to distribute builds that require
something newer, they are allowed to.

gdb 7.5.1 complains about ambiguous base class "XInterface" with cast()
so try dynamic_cast() instead.

Change-Id: I99a21eea7251e663deee395f6a13ca0ed041e8c9

With the previous commit it appears to work now.

Change-Id: Ifa2c69c6e8617461ec2098ecc959e99a4812565e