Checkout
 

hdu in openoffice

#i105098# fix drag&drop insertion of files/images on Mac

Fixed-by: Manik Malhotra <mm30f@yahoo.com>

Reviewed-by: Herbert Duerr <hdu@apache.org>

    • ?
    /main/vcl/aqua/source/dtrans/DropTarget.hxx
#i125897# fix color palette entry for "Sky Blue 1"

Patch-by: Erik Engstrom <engstromea@g.cofc.edu>

Suggested-by: V Stuart Foote <vstuart.foote@utsa.edu>

Found-by: Jay Philips <philipz85@hotmail.com>

Reviewed-by: Herbert Duerr <hdu@apache.org>

#i125921# fix autodetection of the OSX 10.10 SDK

which failed because the autodetection only worked with single-digit

minor-versions.

Patch-by: Thorsten Wagner <thorsten.wagner.4@gmail.com>

Found-by: Thorsten Wagner <thorsten.wagner.4@gmail.com>

Review-by: Herbert Duerr <hdu@apache.org>

#i125815# fix build breaker with Mac 10.10 SDK

#i125776# handle OSX SDK 10.9 changes regarding NSPrintingOrientation -> NSPaperOrientation

SDK 10.9 changed the NSPrintInfo::orientation method to take a parameter of type

NSPaperOrientation (i.e. enum) instead of NSPrintingOrientation (i.e. NSUInteger).

The actual values are the same so the binaries work fine on older and newer platforms,

but the type mismatch on the newer SDK would break the build.

#i63015# always default to WinAnsiEncoding for non-symbol PDF-Type1 export

Suggested-By: edv@gruene.de

Reviewed-By: hdu@apache.org

#i125234# the scrollbar arrows disappeared in OSX>=10.7

so the native-widget assumptions about the scrollbar layout has to be adjusted.

#i125226# expect DDE servers to be already up

"soffice" is always up anyway and other DDE apps should be started manually

#i125226# disallow absolute and relative paths for DDE servers
#i125226# typo fix
#i125226# some macro preferences are directly applicable to DDE servers

#i125226# don't try to access known-bad DDE servers

#i125176# add Debian package maintainer and extended description

the package maintainer is a to-be-created email alias to the target that

will be decided by the Apache OpenOffice PMC.

#i125176# fix the Debian control-file-bad-permission problem
#i122471# make the Mac lookup feature handle the correct word

    • ?
    /main/vcl/aqua/source/a11y/aqua11ytextwrapper.mm
#i125143# reduce helpex stack usage to prevent intermittent failures

moving the 1MB copy buffer from the stack to the heap solves the problem

that was first observed on the Windows buildbot.

#i125145# build helpcontent only for enabled languages

and treat AOO's default language "en-US" as always enabled to allow that

languages with only partial translations for helpcontent can be used by

falling back to the default language for not yet translated parts.

    • ?
    /main/helpcontent2/source/auxiliary/makefile.mk
#i125143# use bulk-copying instead of line-wise in helpex Export::CopyFile()

This simple change has been measured to speed up the helpex by almost 30%.

#i124970# fix DocumentFocusListener::notifyEvent's handling of IndexOutOfBoundsException

the DocumentFocusListener::notifyEvent() throw list only allows a RuntimeException to

propagate. The methods called in notifyEvent() allow IndexOutOfBoundsException though,

so it must be handled in all cases to prevent C++ from invoking its unexpected() abort

mechanisms. Ceterum censeo, non-empty throw lists should be removed altogether...

    • ?
    /main/vcl/aqua/source/a11y/documentfocuslistener.cxx
#i125115# fix unoinfo output on Windows

by using plain fwrite instead of trusting fprintf to handle the wchar_t string

#i125121# speed up the "make writeable" step when building external libraries

avoiding the many individual invocations of chmod for each directory makes

this step about ten times faster on Windows without SSD (460sec -> 41sec

for boost alone). With SSD storage the speedup should be even higher.

#i125003# force gtest sysroot to match the AOO build on OSX

#i125112# fix CStringHash functor to match full strings only
#i125071# check SwFmt::GetItemState()'s ppItem parameter for NULL

before assigning something into it. The ppItem default value is NULL

so it is allowed. The method should be prepared to handle that.

#i125003# fix google-test for osl_condition in the sal module

the test didn't initialize the nanosecond part of the condition timeout and so

the absolute timeout calculated from the second and nanosecond values could

become invalid depending on the pseudo-random value of the nanosecond member.

This resulted in spurious gtest failures (first seen on Ubuntu).

#i125003# sal's autotest submodule depends on sal's cpprt submodule

#i125003# use gtest's internal tuple class for its own needs

test scripts using gtest include that gtest.h header with its dependencies.

Using the system tuple header results in slight problems on RHEL5 that can

be easily avoided by enabling gtest's internal tuple functionality. Other

platforms also benefit from the reduced header dependencies.

    • ?
    /ext_libraries/gtest/use-own-tuple.patch
#i125003# use gtest tarball provided configure script

on the RHEL5 baseline using autoconf for building gtest results in problems

because the platform only provides autoconf 2.59 while the provided aclocal.m4

requires autoconf 2.68 or newer. Fortunately the gtest tarball also provides

a perfectly fine configure script and the autoconf step is not needed.

#i125003# mark gtest as being dependent on soltools

the dmake build environment needs to be functional before such an external

module can be unpacked, built, delivered, checked, etc. Adding a dependency to

the solenv module could suffice already but current and future build recipes

might also expect stuff from the soltool module such as mkdepend or checkdll.

#i125020# allow antialiasing on all update borders for Aqua graphics