If pSVData->maWinData.mpFirstFrame is null in
Application::InitAppFontData() then we need to create a new WorkWindow.

Change-Id: I6665ff6525321554c8fd4943ec623978ce35f168
Reviewed-on: https://gerrit.libreoffice.org/9272Reviewed-by: Tor Lillqvist <tml@collabora.com>
Tested-by: Tor Lillqvist <tml@collabora.com>

Change-Id: Id047044f86ca358c40a87bebde84453787ccec6d

Lossless optimisation of Galaxy PNG files. Saved ~20% overall.

QA- Inspect icons with Galaxy theme selected, using all aspects of the office suite. There should be no visible changes.

Change-Id: I41f17e06cea7acd729749a069fc0ab4f8fbc9630
Signed-off-by: Andrew <dent.ace@gmail.com>

Change-Id: I1bb883502ccd4c6fb5bdfd93f2fa5a04de0b2701

For ordinary documents the SwTxtNode::MakeFrm() would set the
m_bFormatIgnoreStart/End flags of the RSID-only hints, but the document
loaded by Compare Document (and Merge Document) is special because it
is loaded, there are editing operations on it, but it has no layout,
so the assertions about these flags trigger.

(regression from 6db39dbd)

Change-Id: I8a6e02b68f22e609640adbe93ff194e4081d9856

Change-Id: I08698b06d0485d2cd3573af665e0621b42d37349

Change-Id: Idf1460e5207e0797390593535f2b2c080609a77b

Change-Id: I35ccc5ba619f7e861fca2c5e78de86d88480017d

Change-Id: I77044f6f9e99ddaa64cf69ba22205d6536a90b1b

Change-Id: I2c031ffc70e7ddacde52f120513f9dd5ff1a4c50

Change-Id: I8edbf48bc981b4636dd55b5f4cd2b314d387486d

Change-Id: Ia8964b4b7e38d7bba9417695f571858683c2fd5b

Change-Id: I9fb6da4d448923fd5b8822a72d6945a0c0b5f40d

Change-Id: I53112795f301679429afb8da4e01c77e4d2fe820

Change-Id: I1d07802f2671c460dbe145c4d5178aff272ca091

Change-Id: I806058555f58a3f03153372e581fae6d8cb0670b

Change-Id: Ifcce7f0d75cb595add30a4403e5c42348043137b

Change-Id: I1efe68a3b63ddd4b1583a29dd8d71d217bfdd9a8

Change-Id: Id9ac29ec88a98e626a1ecfb364e474b2a11a8004

Change-Id: Id9d014d01d4c2da2d2c421893ba49a4b2fddacc0

Change-Id: Iedee81b4c0474a1bbda0b2366c246f6b6b433a10

Change-Id: I6abb408c91c9d6ea2d60c328be221c7608a7906c

Change-Id: I51b3e9bccbe54686fbe869ce5ba669d000530572

Change-Id: Iacf4373221b93a97e030b09501cc91589f2fdff7

Change-Id: I118771b78d83441202c15d3b08b9055ba5e63623

Change-Id: I8a829ee5d36281098f2788a704b203e6b777d371

Change-Id: If4c218c471b34e2f0c8f083131bc24f1ac9d48e3

Change-Id: I11aa7ecfa52054085aaca70b165a831b5ec8577b

Project: dictionaries  d478664ac0a8e8997d8d17a76a39d2704f7a254d

It would be possible to modify the condition in
SwChapterField::ChangeExpansion() to check for rTxtNd.GetNumRule() as
well and make the bugdoc work again, but do we really want to use
"chapter reference" to refer to plain numbering?  Perhaps it would be
better to add a "numbering reference" field type; without investigating
what Word actually does here and what the actual problem is that was
supposedly fixed here (which is not possible given there's no reproducer
document) it's best for now to just revert it.

(regression from commit 63198037)

Change-Id: I7eb9ea46e42729c53c350ff450d49a1eb84ee8eb

Change-Id: I9bfbde0f5abad6d6f297fc18e24f6b20be914ce4

Change-Id: I76375280ee4726c1d497858165f38f4e8b4224cb
Reviewed-on: https://gerrit.libreoffice.org/9268Reviewed-by: Caolán McNamara <caolanm@redhat.com>
Tested-by: Caolán McNamara <caolanm@redhat.com>

Sure, not a lot of code, but best to not even compile it, to catch
eventual usage attempts.

Change-Id: I7462ea9a5fbe6080fcec2f424d1a1b80ad0afe21

They won't be used for now, but just to enforce portability of the
libgltf code a bit.

Change-Id: Ie5d2b9125eb0f4a2e432dcb8b55d5f494a5ed34d

Change-Id: I7060ae18a98e0793a936834b398f1e232d336672

Change-Id: I0dacaa9fb08062aba758613bd25ebcf7039f36e0

Change-Id: I5a918a6ae4159dd561e46116f612037e194a4847

Change-Id: I66d7c8444c1a89c21ea2ce5e42fb3cd97fd39b7c

The code here counted the numerical XOR of the two pointer values, but
most probably the logical XOR was intended (in case only we or only they
have a node index, then we are not equal).

Change-Id: I261b8c4b498eb860fa0e0fa72ba5a5e0758064e6

...so declare them as such.  This avoids the recurring mistake of storing such
IDs as sal_uInt32, truncating in 64 bit environments, causing RemoveUserEvent to
potentially not remove the event, it thus firing "too late" and probably causing
a crash.

While at it, consolidate the trivially unnecessary overloads of both
Application::PostUserEvent and Window::PostUserEvent.  And in each of them, it
looks like deleting the mpLink member was missing from the failure branch.

Change-Id: Iab13afbb06e12ac15dec6a6b5b85a7e402a3c654