Change-Id: Id9680c42f41a00932240f62ff51e54eda295223b
Reviewed-on: https://gerrit.libreoffice.org/15358Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Adolfo Jayme Barrientos <fitojb@ubuntu.com>

Change-Id: I1c680186271f20b6c01c06dbcc642bc1cf90865a

Change-Id: I1243f8407e8301e0e4662b8839ab53d29dcfe5ed

So the ScTableRefToken always holds some RPN "reference".

Change-Id: I69aa485b9b85286a3b272617781bb07198ea52c8

Change-Id: I142c0c39ab4bb3c1d099cba42491cc9928ab300c

Change-Id: Iaf76cf298dd2f9f9b8d9fb56ecb9e7b29fdc6ca1

Change-Id: If91d6724a07a40c28c0ebe64e94fa3166138072b

Change-Id: Id1b53f7f05305f25ac0b78d3d6c7dbe0fdf52bba

Change-Id: I76ae48783942a558c674c5dcb2869540a819bf49

The ridiculous algrorithm used for TEXT_DRAW_CENTERELLIPSIS will go faster
if we cut out most of the text at the beginning instead of one at a time.

(regression from 912ecaf5)

Change-Id: I9310dda1847222215bafe372e3efef9d365e1ad9

Change-Id: Id887e25f626917d4775f0df92935ecbf08b0455d

Change-Id: Ie0768c822d1271cacd230f8bd9a4c7987e1ebba3

Change-Id: I4613071807bdedeb2058af12f0f808bdffc0e5cd

Change-Id: Ibffcb06e1bfccc4ad604114813b0226395e2d791

Change-Id: I451ab0fb0e7ee1a6d20091caa9c1142b16796dc7

UNC: \\server.domain\file.odt is stored as file://server.domain/file.odt.
Windows local: C:\folder\file.odt is stored as file:///C:/folder/file.odt.
Linux: /home/user/file.odt is stored as file:///home/user/file.odt.

I cannot find any case where we only have file:// (but not a third slash) on
a local file.

Change-Id: Ib15b3380cb40f1cd37123be7a38426b74fbde5cf
Signed-off-by: Stephan Bergmann <sbergman@redhat.com>

Change-Id: Ie077fc273601607cba72099d650c6774f08240fb

...all uses of bytes::data() look like they're fine with std::vector<T>::data()
not necessarily returning nullptr when empty

Change-Id: I48c3877cfb9fb1a4a129bd1e072410a8bebc87b1

...used values were 0, 1, and SC_FORCEMODE_NONE = 0xff

Change-Id: Id6303b77a11d22c3d19cb22c3b5838cc7cec624f

Change-Id: I9ca57f62755b614c8b80c06907f923403c7f68df
Reviewed-on: https://gerrit.libreoffice.org/15338Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Adolfo Jayme Barrientos <fitojb@ubuntu.com>

Change-Id: I790cf788ff0043bfadf0fabd9a78166e5f05b26f
Reviewed-on: https://gerrit.libreoffice.org/15337Tested-by: Adolfo Jayme Barrientos <fitojb@ubuntu.com>
Reviewed-by: Adolfo Jayme Barrientos <fitojb@ubuntu.com>

Previously we just kept the stream around till the next embedded object
or document end and hoped that the stream wrapper is deleted before the
underlying stream goes away.

Change-Id: If2bdb1e664d803fc48d06204fe3aef72b279b5b6

There were two problems here:

1) The input stream is read multiple times, so plain naive XInputStream
is not enough, XSeekable is needed, as it is the case for the stream
provided by the OOXML tokenizer.

2) Seeking to the correct start position is not enough, code using the
stream will assume that seeking to the beginning of the stream allows
reading the correct data again, so provide a dedicated stream instead.

With this, handling of math equations created by Word <= 2007 and
embedded to RTF files can be edited finally.

Change-Id: Ic225e1e1060f8bdd5651a21e68970620c9ac6b68

Change-Id: I771c83d0ed8a06a2eeec57a40a80092b537d9b84

Change-Id: Iec84588daf5555d1207827efb0f26d8354ad462b

The transparent frame background is painted opaque because
OutputDevice::DrawTransparent() records a MetaTransparentAction
instead of returning early.

Note that master and 4.4 use drawinglayer to paint backgrounds
so the problem is only visible in 4.3.

(regression from 36b59f2efc2bfb2c9256c39eb5687808deb)

Change-Id: Ide7a076b72123097a9fe099b96d36cda7f7bb082

Change-Id: I30536ecd8a03a19a647d247e7c6ece3d6f1695c1

Avoids this mess:
checking whether to use gnutls... checking which redland to use... internal

Change-Id: I1950755fb5331b6158b99a2741b15d34a32dcd38
Reviewed-on: https://gerrit.libreoffice.org/15333Reviewed-by: David Tardon <dtardon@redhat.com>
Tested-by: David Tardon <dtardon@redhat.com>

Also, be consistent in spacing, indentation and alignment. Inconsistency is
infuritating.

In this case, I decided to drop the attempts to align the names of "methods",
and their parameter lists. They were not all aligned anyway, and some of those
that tried to be in fact were off by one column. Also, consistently use a
blank line between the method pointers,

If you feel strongly about this and *want* such inter-line alignment, by all
means re-introduce it, but at least try to be consistent then, and do it for
all methods.

Change-Id: I382ca0eaf8533ebc3f829ebd978a9676ebcefaf0

Change-Id: Ie8dafb4d5dfd2ab09d71bc509f91e5c2c483bfb8

Project: translations  8514f4f02022d70c8221eb2f2b47a7f18de1475e

Updated Slovenian translation

Change-Id: I7cd4ad4cbab7189893ef2d4321443362af7dea21

pDoc isn't null here

Change-Id: Icf705f2bc67cb08a1920aea5965eb6f0e4786ada

pre recent refactor pBox was assigned NULL
on the now early return pLines->empty() path

Change-Id: I6ecb31afc7a70fd5e11ff474145fd0e73dc33cef

This reverts commit add2ca7b.

It causes too much breakage on windows. It is better to start again from
scratch.

error C2487: member of dll interface class may not be declared with dll
interface

Change-Id: I467670d483112ddbe8e608a7884a9c4e168e470d

error C2487: member of dll interface class may not be declared with dll
interface

Change-Id: I00d2a30490ab3ae5fa574e398fb6327089b8e770

Change-Id: I37a1d16f76cad6b8ff5727ecd09d718e15519de9

Change-Id: Ibfb4a60bfdce303264ef1826f943672695cdf274

error C2487: member of dll interface class may not be declared with dll
interface

Change-Id: I63898789eb9c9ee767a5c72e31156e23d50c81fb

error C2487: 'Read' : member of dll interface class may not be declared
with dll interface

Change-Id: Iabee3995f612116c3b6bd5c881ed617c0393addf