Change-Id: Iddcc22f62ea134ea797292daa02e47042d52d0d1
Reviewed-on: https://gerrit.libreoffice.org/16710Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Samuel Mehrbrodt <s.mehrbrodt@gmail.com>

Change-Id: Ic6415423f46aaee7ba90239a617c318cf92ae222
Reviewed-on: https://gerrit.libreoffice.org/16711Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Samuel Mehrbrodt <s.mehrbrodt@gmail.com>

Change-Id: Id9296115f30858e7fd470a199e59343a96d7deec
Reviewed-on: https://gerrit.libreoffice.org/16712Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Samuel Mehrbrodt <s.mehrbrodt@gmail.com>

regression from

Change-Id: Ie12713778112d1c6ddc9ed817af3998a3dcd9ba8
crashtest: don't crash on layout of ooo70429-[1|2]

Change-Id: I8b0a6c3323a4ffbe6a1ba09e5cff9ddd92ed49f1

This reverts commit 4a0cb642.
MacOSx only (more exactly CoreText only)
The handling of DXArray was buggy and need to be re-worked
so this revert to the previous case where DXArray are esssentially
ignored at the char level and only kept at the fragment size level
and used to redistribute space in a fragment

This create some issue too but on more esoteric case than
the one noticed in tdf#88941

Change-Id: Icb33d0600267c3136f3a6b15de8ecd739898ce04
Reviewed-on: https://gerrit.libreoffice.org/16747Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Norbert Thiebaud <nthiebaud@gmail.com>

... the maMarkStack, which causes the order of calls to
startFastElement()/endFastElement() to differ from the order of the tags
that are written into the output.

This is an attempt to improve the assertions, but if an assertion fails
it's generally not obvious where the problem actually is since the
unpredictable order may cause the problem to be detected and reported
much later than its root cause.

Let's see if this finds any new problems in export testing.

Change-Id: I97699cc8ef9b18ea9f4f221d5210134feecf0336

This reverts the changes in FastSaxParserImpl from commit
16e8ffbd and instead fixes the problem
of inserting string properties with empty value locally in
OOXMLDocPropHandler.

This change was not wrong in any obvious way, but it turns out there is
one doc rhbz583386-4.docx that, when imported with this change and
exported to DOCX again, results in a non-well-formed document because of
some weird SDT stuff.

That problem is rather baffling, but unfortunately the
DocxAttributeOutput usage of FastSaxSerializer::mark() makes the DOCX
export rather un-debuggable, so avoid that problem by reverting the import
change for now.

Change-Id: I0d874cbfe82d4f15d58b50116dda152341bdf7b0

Change-Id: I458f8e8665373d3e5ab9ed20e39783f4d34ca128

Release the accelerator, and cleanup global references.

Change-Id: Id9931527f1b6d89067bc1e520e521ccfa6e28197
Reviewed-on: https://gerrit.libreoffice.org/16740Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Michael Meeks <michael.meeks@collabora.com>
Tested-by: Michael Meeks <michael.meeks@collabora.com>

- crashed due to out-of-range index with empty vector
- kill pruposeless manual index fiddling for good
- add trivial test

Change-Id: Ic224c70f432d2b26718e05aa5a2742461cfb7bec

- the previous code tried to avoid breakage from changing an
  implementation detail
- this turns out to be harder than expected in a Python2/3 portable way
- also it would mean the intentional drive-by documentation of the
  expected implementation name will at least not bit rot now

Change-Id: I5972d0e79ef64abe30573182c5e23268e4830ade

The bool simplification from 3ac5138c
accidentally changed the logic of the return statement

Change-Id: Id6075356184e340dc970d2e59d1fb1d1a173d0a7
Reviewed-on: https://gerrit.libreoffice.org/16724Reviewed-by: Michael Stahl <mstahl@redhat.com>
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Caolán McNamara <caolanm@redhat.com>
Tested-by: Caolán McNamara <caolanm@redhat.com>

Change-Id: Id545f3447472a8e6ee137a6630ee14b45a45d673

Change-Id: I43397f2b0caa31c334f87acf11fc96df2e051ad8

Exporting linked textboxes to docx format is in terrible condition.
Spacing, textboxes instead of frames, duplicate links and orphaned
shapes lying around, not to mention being being unlinked in MSWord...
This fix resolves this situation slightly: what LO saves can be
re-imported and re-saved without breaking the links - an incremental
improvement.

Change-Id: I8f0aef39eeed88a2b3dfc673a565fb1d8f4713b0
Reviewed-on: https://gerrit.libreoffice.org/16516Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Miklos Vajna <vmiklos@collabora.co.uk>

Based on a patch by Henry Castro.

Change-Id: Ia7aca20feb8f6095adf7dfe510ed78b1e9882740

With double-buffering enabled and typing quickly, one saw blinks in the
text sometimes. The reason was that PaintDesktop() painted directly
outside SwViewShell::Paint().

The problem is more visible with VCL_DOUBLEBUFFERING_AVOID_PAINT=1,
loading a simple document + pressing a key draw an unexpected white
rectangle.

Given that InvalidateWindows() already calls PaintDesktop() inside
Paint() implicitly (via SwViewShell::Paint()), fix the problem by just
avoiding the DLPrePaint2() + PaintDesktop() + DLPostPaint2() calls.

Change-Id: Ib8ce26a0bdd526bf85d362f0e865dd61d29f6d11

The original work of tdf#55430 tries to select an object under
another one on the second click, but these images are unselectable
so this fails. Red Hat has a whole new shiny bunch of templates which
have such images in their masters.

Check if the object is selectable before continuing

Change-Id: I182abaf50e8bb1084c5819dc9e1ffd8b386a9e93

Change-Id: If3b58c02b289224938b1dfd8074f9802ea6d2474

The basegfx changes appear to break Windows builds.

This reverts commit 3b32c589.

Change-Id: I99a87da22d877bc1386552a8978b0bf5e8e7f51f

The inactive pageindicators had the same grey like the active ones, I correct
this and changed the dark grey into light grey.

Change-Id: Iea810d466504ceb80750c45358a01f8b8d3c8b9e
Reviewed-on: https://gerrit.libreoffice.org/16621Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Yousuf Philips <philipz85@hotmail.com>

(and note the operator == vs. = typo)

Change-Id: I9c25149967ed9b80f5e8f014de0dfee2e8f485d8

Change-Id: I313adfb72272a631299efb3d2464873853a126be

Change-Id: Ic4058c133a676646bc0c5d1e52207471bb78a1cb

(and note the operator == vs. = typo)

Change-Id: I4567927f3f702f8e74b168becce4002e84d5c7a0

Change-Id: I43710d057043a258ec9ab94bca0703434098e6c9

Change-Id: Ib615ff7eb12232b0b83664af134744d96754ffd8

Change-Id: Ie453a7c97a762932884c5c41b13607a170368107

Change-Id: I72bdd6627a191a65d5c3b479c100e30442a8bb9e

it introduces other problems instead.

This reverts commit 05806f46.

With this, if the sw window is small enough that the page shadow is
supposed to be under the sidebar, we no longer paint the shadow over the
sidebar.

In the non-double-buffered case the output is a window, so
SdrPaintWindow::PreparePreRenderDevice() does its own buffering, and
clipping is taken care of when that buffer is copied to the window in
SdrPreRenderDevice::OutputPreRenderDevice().

This wasn't a problem before, as either the output was a non-screen
device or UI elements were painted (like the sidebar), but never both at
the same time.

Change-Id: Ic02c814fc4deeea86c245c74bd69656533d9f835

Change-Id: I88c2d7de243660ba727f057d6a2d67726b41b982

to restore location correctly

if SwFEShell::EndDrag is tweaked to not group the anchor change
undo and the draw size/pos change undo together and then you
undo each step in the ui one at a time this works fine.

The first undo occurs, then the layout kicks in on idle and calls MakeObjPos
on the drawing item and updates it for the new anchor pos

Then on triggering size undo the anchor is correct so the new
pos is set correctly.

If they are bundled together however, the layout at idle doesn't occur
until after the anchor and size have been set, so the adjustment is
too late

Change-Id: I5402d851fcae7f240cc6e9a0e15159115bb663ce

Change-Id: Ia5d63f7153e4b02acc5e206739316264d6d1184e

Works with the non-system bzip2; the --with-system case might still need some
love - but then again, distros want to have the online update switched off
anyway...

Change-Id: I26a27aec07a8f1aac22a4d14a38fb5b833550ea7

Change-Id: Id7547390c49c0bcf672eb1a9863fc236518bfa49

Change-Id: Id69448d893f7492e7321c6c6f84d6cccc9024220

Avoids leftover artefacts of the blinking text insertion cursor. Did not seem
to have any ill effects.

Change-Id: I8a1e7413034eacb3014a4ca902cf5e72bda492a4