Why do we need the DLLIMPORT macros in a header only class?

Change-Id: I0da4e9a3fd9b8db77e7298d4d592fd29eff1f709

Change-Id: I3de9afa7b5d975487fd03ae2a9c5393143abff9c

Change-Id: I8d63d942c7cb7876b864a9720e9a8adbc7345172

Change-Id: I9667611135300155ee87cca1ad444ad8a6a9aeb1

This reverts commit 10654c03.

Change-Id: I37d8752fa6a5b532c8fce3175bf42d3c8c93b3a8
Reviewed-on: https://gerrit.libreoffice.org/18360Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Eike Rathke <erack@redhat.com>
Tested-by: Eike Rathke <erack@redhat.com>

Change-Id: I4c978a1d2f2a63eae08e30a2f349b445b4a98ecc
Reviewed-on: https://gerrit.libreoffice.org/18347Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Eike Rathke <erack@redhat.com>
Tested-by: Eike Rathke <erack@redhat.com>

Change-Id: Idbd3b5a6bf0008409cf1b4f3764e1b34f4284de9
Reviewed-on: https://gerrit.libreoffice.org/18361Reviewed-by: Eike Rathke <erack@redhat.com>
Tested-by: Eike Rathke <erack@redhat.com>

Change-Id: I78c0f32498ae559604f78706db85141bbebf89ad
Reviewed-on: https://gerrit.libreoffice.org/18359Reviewed-by: Eike Rathke <erack@redhat.com>
Tested-by: Eike Rathke <erack@redhat.com>

Change-Id: I7474d6ac1c7128ca5b22478b2abb3799c083e99c
Reviewed-on: https://gerrit.libreoffice.org/18348Reviewed-by: Eike Rathke <erack@redhat.com>
Tested-by: Eike Rathke <erack@redhat.com>

Change-Id: Ifd6928bafc59abc252f29af81f26f6362e9fc933

Change-Id: Iab0a53abd723f0309f40742636315079a4b2c532

Change-Id: I60ec3a95bf7d631fefe061335bf0955216df14b9

After the document is created, an event is dispatched on the main loop
that calls SfxPickList::Notify(), which modifies document properties.

It tries to prevent setting the document to modified by calling
SfxObjectShell::EnableSetModified(false), but Writer cunningly outwits
it by simply having its own independent(?) modified flag that is set
unconditionally in DocumentStatisticsManager::DocInfoChgd().

Let's assume that if the modified flag shouldn't be modified in
SfxObjectShell, it shouldn't be modified in DocumentStatisticsManager.

Somehow in 4.4 and 4.3 the same thing was going on, but it didn't result
in a visibly enabled Save icon in the UI, but with 5.0 it does - cannot
easily bisect why that changed due to tdf#91383.

Change-Id: Id30fd831eb29910c9fb44ed3031bf8da23586bea

There are even more and it is causing a number of performance issues.
This just fixes the most obvious place.

Change-Id: Id85c13b5b7f4bb0bc077e62c39258067e17c0094

The code comment mentions correctly that we want to show all significant
digits but a value of -1 means that the last digit before the decimal
point is rounded away.

Change-Id: Id91a6076e37629502281c5dda426018f93d1e465

Change-Id: I5f7f993f2cf0604ec62cea3460b651c07a7b4383

... when deleting the temporarily inserted styles in
SwDocStyleSheet::FillStyleSheet().

Change-Id: Id4abc067ce10b41486f659350267c7e3933db472

...the code was like this ever since 8ab086b6
"initial import"

Change-Id: I808296a280c01f02adeba1018b9f9bc6977971d3

"return bDone;" was commented out ever since
8ab086b6 "initial import"

Change-Id: I604c7b49623e730cbbe486563f1a91f713ff6810

Change-Id: Id325d05060e427ad46acd45d0ec3f0e9159cea9f

Change-Id: I4a61595294c24a609a5952ce72f9f88524969784

Change-Id: I4a7199633d0d940ae009ac144989da7d6c9262ec
Reviewed-on: https://gerrit.libreoffice.org/18378Reviewed-by: Miklos Vajna <vmiklos@collabora.co.uk>
Tested-by: Jenkins <ci@libreoffice.org>

...ever since its inception with fccfef48
"INTEGRATION: CWS impresshydra"

Change-Id: I098b4434bfb1ed2c10f1476c21f1427fe2292b60

Change-Id: I91ba41cb052c38aa3b047cf079090b01bbe93b39

...the odd setting of iconSize (to the same value inside the if as well as
unconditionally afterwards) was introduced in
17a5c821 "INTEGRATION: CWS vcl10: #109868# don't
use ridiculously large icon sizes," together with the olwm comment, and olwm-
specific code further down below that was removed again with
b118d56c "Remove Olwm support and Windowmaker
gravity handling;" so effectively revert the code to the state before
17a5c821

Change-Id: Ic68f6b5b44f48292a7a6983935427e529d8a5002

... which means the internal column names should already had it. An
unsuccessful cell lookup fallback doesn't necessarily fulfill a warning
condition.

Change-Id: I973b8edff83fcb47a34cc35fd9bf5163882f7742

Change-Id: I843b2cb12783549033fc81f82751704859139960

Change-Id: I9c5d9c53785828b3535c960f826ccd3720b2e4c3

...ever since c3607b3b "INTEGRATION: CWS
presfixes01"

Change-Id: I29107b0e55d16a8a88e25c7b929eeb40d56686f0

If the style does not actually have any font items, as is the case for
frame, page and number styles in Writer, the maFont will be
default-initialized and the maPixelSize incorrect.
Avoid using these variables.

Change-Id: I084cd8faa90a3d53310ceec55e8dae3af3dd586c

SwDocStyleSheet::FillStyleSheet() already takes care to remove all
temporarily created styles, so assume that works and suppress the
creation of user-visible Undo objects.

Change-Id: I748f0e8304c42e767b331ebd22be0290b9c0d89d

The dialog/sidebar should not actually create styles that don't exist yet,
because it messes up Undo and the (unused) styles are then unnecessarily
exported to documents.

Due to Writer's ... unusual SwDocStyleSheet class this is a bit tricky.

Add a new function GetItemSetForPreview() and use it from the style preview
code.

The implementation does not use FillPhysical so will temporarily create and
then delete any non-existing styles.

Skip page and numbering styles for now since they don't have a useful preview.

(regression from ca953076)

Change-Id: Id6ee30ea467fc24c991547a4c23a9ce14fdd86c7

This causes the temporary creation of frame styles to fail.

Change-Id: I5d148ae008660d9c0f457a4c0e9dc256e0dfc49a

Change-Id: I341fb7c0712dff77ebfed41560a487f85c27c25a

Change-Id: If8d9770717c21875b4472b0d94a1fa5a9d136085

+    rLineBoxes = pFndBox->GetLines()[nEnd]->GetBoxes();

This actually assigns the boxes vector of the last line to the first
line, which the previous code didn't do.

Notably after converting the types from boost::ptr_vector to
std::vector<std::unique_ptr>, the assignment produces a several page
error message from both GCC and clang, and infuriatingly neither
compiler was able to tell on which line in tblrwcl.cxx the template
was instantiated, so i had to bisect that.

(regression from be5e4247)

Change-Id: I646e3ca678895480d38ec48f38d720458860a985

...the computation of nTop, nBottom was done in this curious way ever since its
inception with 108632dc "CWS-TOOLING: integrate
CWS impressaccessibility3"

Change-Id: Ib516e944d96dc10ff12d0a82c0d8c921b57b0466

Change-Id: Id40a6802536120501133968d003267f21aaf5c27

Change-Id: Ic21d88b55b22c650de4fd69479b51d0f640fec6f