StringRangeEnumerator (i.e. user's input) contains page numbers in a different
page range (it excludes empty pages). So:
- first map page numbers to a common range, then compare
- user's input can't contain empty pages, remove this check

Change-Id: I4fce5215272fc90f39c9e05d3f3604734a8aebe3

Change-Id: I5e5bcebeb804256d0ad8cf6afafc392b2853b88f

Change-Id: I230e8972340b409c6da70a4c988ac1d531a9f3b5

Change-Id: Ic64d2cae56787884ff107b78c3b6567ab5d38785

Change-Id: Id7349d1ff6547afe0f46404340b20132b97ce854

FINDPRODUCT property was not available to this deferred custom action.
Not to mention that registry keys are also deleted at his stage of uninstallation.
The proper solution is to set the installation directory with a type 51 custom action,
and pass it to RemoveExtensions custom action via CustomActionData property.

Change-Id: I0ac18b3a0b19ff1a87bcf580fad9c7fdadb26f76

- inverted check in SwDoc::RemoveInvisibleContent
- various too short end positions in DeleteAndDestroy calls in
  SwDoc::~SwDoc and SwDoc::ClearDoc
- std::copy to begin() of empty vector in SwReader::Read replaced
  with std::back_inserter; the vector is used as a set here so the
  end position should work just as well

Change-Id: Ib2ddd788fade4ee1a8beb702d5321c503985fba4

This test uses gb_CppunitTest_use_library_objects, and MSVC won't link
it now. Seems this was added in 1015cd4f,
but why?

Change-Id: I206132cf2b0c4075334ee7a692a12c54da9ac953

Change-Id: I91d84a7a8d8f18883798541e6a171ea91f135edb

we get a large performance win by evaluating the formulas at the end of import where we can set up the ScAutoNameCache cache.

Change-Id: Id3763afb805eb9bfa9a6a55e1de3923f02e621ae

Change-Id: Ia521752c4c71cc7df2bb7fc4659fcf6ec528e5bc

Change-Id: I021b4d70a02de7a66e0af82a859ae89e1a7d1231

Along the way, simplify a lot of the ForEach callback methods.

Change-Id: I54bc87c1060f1d73470ca8b7fd645ffa23059606

Change-Id: Idffc8a90fcd4ada856339fc0be3e4a11478db6cc

Change-Id: If85e2bc65a99e4854d92454b3bfd40e436df2aa1

Change-Id: I9f26d380526c63a8b86b42a5b0b74a7cfff71249

Change-Id: I3b6f41b97d1fc375985465ea9e5c0cca9e05ba10

Change-Id: I91c33be902e20f72623c3d9642057db5eb6e30ba

Change-Id: I2e80e17ed79e156bcfe379ddc5b11fd2cb2804d2

Change-Id: Id2f35eba75628ed1bc1080e9671378e461e84233

Change-Id: I65b6f6e1f615c8f6e778b10945c60a28875d88ba

Change-Id: I905d17fbbda11199065237b396c52bacc481e414

I couldn't use ptr_vector because some of the code deletes
entries without deallocating them.

Change-Id: Iafbb19dc8ca9f8337169e116d47dcf0ccf78ed0e

Change-Id: I49cefb7e4b8df0f9b95b7a3ef2beaf2d8c0c98e7

Change-Id: I29d7b6f55f227f11abbf7e41fba3ffd7af85f268

Change-Id: I15f7449c8bf0b1d93c7185687d833136d6fbc26f

Change-Id: I92492e051d859537b140ccebe279e09ba4eed566

Change-Id: I851a3ef2bda7f427f6562ed946c7889c97ad83a0

Change-Id: I3a09158ee1d864d3185b015d344391a60a926b65

Change-Id: I12762582ccf871828e27900f643fcbe4b59dbbf3

Change-Id: Id557b0116ec2574ef3e11a0664fb408ecc850e6e

Change-Id: I196f7a6211a3c11d976a95d592af4205476fe81f

...(i.e., after logging out of a session with soffice still running, but only
with the backing window still open).  The key issue was that bSessionDataExists
would always be true if /org.openoffice.Office.Recovery/RecoveryInfo/SessionData
is true, even if /org.openoffice.Office.Recovery/RecoveryList is empty.  Always
calling doRestore (unless bExistsRecoveryData, but even if !bExistsSessionData)
is there to ensure that /org.openoffice.Office.Recovery/RecoveryInfo/SessionData
is eventually reset (whether or not that is really necessary).

Change-Id: I9c3fff23f8019069f36e1dd5ded634ea1917bbd4

The combination of marking it as ACT_IGNORE and osl_Signal_Terminate was
strange, as osl_Signal_Terminate caused SalMainPipeExchangeSignal_impl
(desktop/source/app/officeipcthread.cxx) to call DisableOfficeIPCThread, but
ACT_IGNORE (instead of ACT_EXIT, like other signals marked osl_Signal_Terminate)
caused the process to continue running.  The net effect was that calling an
additional soffice instance after a SIGHUP would no longer connect to the first
instance, but instead would lead to two soffice instances running simultaneously
on the same UserInstallation.

Marking SIGHUP as ACT_HIDE most closely resembles the old behavior (of having
~no effect).  Other options would be to mark it as ACT_IGNORE and
osl_Signal_System or as ACT_EXIT and osl_Signal_Terminate, which would both
result in instant termination.  I do not know if that wouldn't actually be
better.

Change-Id: Id1ee7c1108765301e6010e0e05d6c1ff5894a414

Change-Id: If3a6dcd2216cbac63b43fb393af37206e8730a20

- I had to convert SwFrmFmts, SwSpzFrmFmts, SwCharFmts, SwSectionsFmts,
  SwTextFormatColl, SwGrfFmtColl all together because of the code that wants
  to work with any of them.
- Introduce a pure virtual base class SwFmtsBase, since there are a handful
  of methods that want to work with any of the std::vector<Sw*Fmt> types.
- the class SwSpzFrmFmts was dropped, and the code changed to use SwFrmFmts
  instead, since the 2 types are identical.

Change-Id: I8c4056b5aedd574a33666e2d649a8b849958441c
Signed-off-by: Fridrich Štrba <fridrich.strba@bluewin.ch>

Change-Id: Ic8f3885965a46cc7cea891472f7d902254004543

This commit partially reverts 9e8957de
(as it interpreted mso-wrap-style as the wrap type of the textbox
itself, not the wrap type of the contained text) and also extends the
effect of commit 0016757e to the
SimpleShape class as well.

Change-Id: If2bded6ad94f2866e8ea398ff0f2792e8c8e86b0

06e2005e changed String to OUString and
initialized some variables with -1 instead of STRING_NOTFOUND, which was
unsigned 0xffff so var1<var2 was true if var1 had a value other than
default and var2 was default. After the change the comparison had logic
reversed and the result was a default negative currency format with
parentheses for all locales.

Change-Id: I3fe49fed7c1047618dc653fce992eb01582c398d