Port main/i18npool to gbuild.

Allow ICU's version.mk to be used by GNU make (= instead of *=).

Add support for building generated C files.

Reference ICU libraries using "gb

Port main/i18npool to gbuild.

Allow ICU's version.mk to be used by GNU make (= instead of *=).

Add support for building generated C files.

Reference ICU libraries using "gb_Library_use_externals" across the tree.
Also ICU is an external library, remove it from RepositoryFixes.mk
which is only for internal ones, and keep it in RepositoryExternal.mk
exclusively. Since we get to specify the full linker parameter over there,
remove the half-baked hacks for it from RepositoryFixes.mk.

Patch by: me



git-svn-id: https://svn.apache.org/repos/asf/openoffice/trunk@1852965 13f79535-47bb-0310-9956-ffa450edef68

show more ...

Fix the Ant dependency extraction for the case when
there are no dependencies.

Fix the format of Ant paths on Windows.

Patch by: me



git-svn-id: https://svn.ap

Fix the Ant dependency extraction for the case when
there are no dependencies.

Fix the format of Ant paths on Windows.

Patch by: me



git-svn-id: https://svn.apache.org/repos/asf/openoffice/trunk@1847506 13f79535-47bb-0310-9956-ffa450edef68

show more ...

Complete enough of the assembly language building for now,
to successfully build main/basic on Windows.

Patch by: me



git-svn-id: https://svn.apache.org/repos/asf/openo

Complete enough of the assembly language building for now,
to successfully build main/basic on Windows.

Patch by: me



git-svn-id: https://svn.apache.org/repos/asf/openoffice/trunk@1842965 13f79535-47bb-0310-9956-ffa450edef68

show more ...

Add a gbuild API for building assembly language source files.

This will be needed by basic, bridges and possibly other modules.

Patch by: me



git-svn-id: https://s

Add a gbuild API for building assembly language source files.

This will be needed by basic, bridges and possibly other modules.

Patch by: me



git-svn-id: https://svn.apache.org/repos/asf/openoffice/trunk@1841998 13f79535-47bb-0310-9956-ffa450edef68

show more ...

Fix potential memory alignment issues on X86_64.

Clang version 4.0 and newer uses SSE instructions that require
16-byte alignment to zero memory allocated using the C++ new operator.

Fix potential memory alignment issues on X86_64.

Clang version 4.0 and newer uses SSE instructions that require
16-byte alignment to zero memory allocated using the C++ new operator.
The internal memory allocator does not understand anything larger
than 8-byte alignment. Modify it to be capable of doing 16-byte
alignment when necessary.

There is also a debug layer beneath the C++ new and delete operators
that is enabled by the --enable-debug configure option. This layer
adds 8 to the requested size of any allocations before calling the
underlying allocator, adds a known signature to the start of the
memory block, and then adds an 8 byte offset to the pointer before
returning it to the caller. The delete operator basically does the
reverse, checking for the proper signature. Modify this code so
that it adds and subtracts a 16-byte offset on X86_64 so that a
properly aligned block from the underlying allocator does not cause
new to return a misaligned pointer.

Modify set_soenv.in so that it always requests 16-byte alignment
on X86_64 so that the ABI is the same independent of the toolchain.



git-svn-id: https://svn.apache.org/repos/asf/openoffice/trunk@1830406 13f79535-47bb-0310-9956-ffa450edef68

show more ...

Unbreak build on FreeBSD with clang 6.0 and newer. Clang 6 changed
the its default C++ standard from C++98 to C++14 and our old code
is not prepared for that. Avoid the problem by adding th

Unbreak build on FreeBSD with clang 6.0 and newer. Clang 6 changed
the its default C++ standard from C++98 to C++14 and our old code
is not prepared for that. Avoid the problem by adding the -std=gnu++98
compiler flag.

Add the -fstack-protector compiler flag on FreeBSD INTEL and X86_64.



git-svn-id: https://svn.apache.org/repos/asf/openoffice/trunk@1829757 13f79535-47bb-0310-9956-ffa450edef68

show more ...

Work around a bug in clang versions 3.6.x and 3.7.x on 32-bit Intel.
Using -Os optimization causes clang to generate bad DWARF CFI which
is needed for stack unwinding during exception handlin

Work around a bug in clang versions 3.6.x and 3.7.x on 32-bit Intel.
Using -Os optimization causes clang to generate bad DWARF CFI which
is needed for stack unwinding during exception handling. See:
<https://llvm.org/bugs/show_bug.cgi?id=24792>
Instead of using -Os, optimize using "-O2 -fno-unroll-loops" as a
reasonably close substitute.



git-svn-id: https://svn.apache.org/repos/asf/openoffice/trunk@1829681 13f79535-47bb-0310-9956-ffa450edef68

show more ...

Port main/jvmfwk to gbuild / Ant.

Fix the gbuild platform CPUDEFS to conform to what modules expect.

Patch by: me



git-svn-id: https://svn.apache.org/repos/asf/ope

Port main/jvmfwk to gbuild / Ant.

Fix the gbuild platform CPUDEFS to conform to what modules expect.

Patch by: me



git-svn-id: https://svn.apache.org/repos/asf/openoffice/trunk@1829205 13f79535-47bb-0310-9956-ffa450edef68

show more ...

Implement the ability to build "UDK versioned" libraries in gbuild
for *nix platforms. Do this by setting the ELF SONAME to the
libxyz.so.3 style library output name, changing the output file

Implement the ability to build "UDK versioned" libraries in gbuild
for *nix platforms. Do this by setting the ELF SONAME to the
libxyz.so.3 style library output name, changing the output file
that the linker writes to to be in this format (from libxyz.so),
making a symlink from libxyz.so to libxyz.so.3, adding libxyz.so.3
as a delivery AUXTARGET for the deliverable libxyz.so, and
changing the deliver commands to copy symlinks properly. This is
all what dmake does too.

Use this to port main/registry to gbuild.

Patch by: me



git-svn-id: https://svn.apache.org/repos/asf/openoffice/trunk@1827456 13f79535-47bb-0310-9956-ffa450edef68

show more ...

#127664# $CCNUMVER from dmake to configure

Move the calculation of $CCNUMVER and some other variables from
main/solenv/inc/tg_compv.mk, where it is only usable by dmake, to
configure

#127664# $CCNUMVER from dmake to configure

Move the calculation of $CCNUMVER and some other variables from
main/solenv/inc/tg_compv.mk, where it is only usable by dmake, to
configure, where it can be used by both dmake and gbuild. This is
a requirement to upstream some compiler bug workaround patches from
the FreeBSD port.

A bit of logic from set_soenv is also moved into configure. A bunch
more should probably be moved so that the configuration logic is
not spread across so many different places, but that can wait.
Something else to consider is that it would be nice to use a different
value of $COM for Apple's clang, maybe "ACLANG" or "APPLECLANG"
since it has a different version numbering scheme that the open-source
version of clang and having a unique identifier would simplify
version checking when applying compiler bug workarounds.

Note: I think the old value of $CCNUMVER on the Mac is wrong. It
should look something like 000800010000 or 000700030000, depending
on the installed version.

Change -DCPPU_ENV on the Mac from $(COMID) to $(COMNAME) for
consistency with the dmake side. It shouldn't make a difference
in practice since both have the same value on the Mac.



git-svn-id: https://svn.apache.org/repos/asf/openoffice/trunk@1826296 13f79535-47bb-0310-9956-ffa450edef68

show more ...

Nothing uses GXX_INCLUDE_PATH, so remove it. The uretest/README
file mentions is being needed by the STLport included in the URE,
but that does not appear to be true any longer, and it is no

Nothing uses GXX_INCLUDE_PATH, so remove it. The uretest/README
file mentions is being needed by the STLport included in the URE,
but that does not appear to be true any longer, and it is not
used by boost.



git-svn-id: https://svn.apache.org/repos/asf/openoffice/trunk@1820351 13f79535-47bb-0310-9956-ffa450edef68

show more ...

Improvements to per-object optimization overrides in gbuild:

* Less Library.mk editing required to add an optimization exception
than what has been normally done to date.

*

Improvements to per-object optimization overrides in gbuild:

* Less Library.mk editing required to add an optimization exception
than what has been normally done to date.

* Add a variable to specify an override to -O1 instead of only
allowing -O0.

* Don't lose debug flags when overriding optimization.


Convert the recent Mac optimization override in framework/Library_fwk.mk
to the new style.


Improvements to Windows debug builds in gbuild:

* There are multiple dependency paths that trigger compiling a source
file. Add a target specific $(PDBFILE) to the root of each so
that $(PDBFILE) is set properly when compiling.

Note: $(PDBFILE) is not getting set when compiling the sources
for the per-module (Google Test) tests.

* Don't inadvertantly disable the use of pre-compiled headers when
enabling debug (not specific to Windows).

Note: If both debug and pre-compiled headers are enabled, building
of modules that contain more than one library will fail due to
limitations of Visual Studio. The problem is that when we
build the precompiled header, the output file(s) are based on the
module name. Even though each Library_*.mk specifies the rule
to build the precompiled header, it is only built once for the
module. Visual Studio requires that the $(PDBFILE) used when
compiling the sources match the $(PDBFILE) used when compiling
the precompiled header, but the value of $(PDBFILE) is specific
to each library. The best fix would be to compile the headers
once for each library and stash the output in a per-library
specific place, but this requires some fairly extensive changes.
Another potential fix would be to use -Z7 optimization so that the
debug information is stored in the object files.



git-svn-id: https://svn.apache.org/repos/asf/openoffice/trunk@1817946 13f79535-47bb-0310-9956-ffa450edef68

show more ...

Add the ability to set a linker script in gbuild that is used to
version symbols.

This has been a bottleneck in continued gbuild porting and should
help further development considera

Add the ability to set a linker script in gbuild that is used to
version symbols.

This has been a bottleneck in continued gbuild porting and should
help further development considerably.

Patch by: me



git-svn-id: https://svn.apache.org/repos/asf/openoffice/trunk@1816518 13f79535-47bb-0310-9956-ffa450edef68

show more ...

Add debug symbols to gbuild modules when any of --enable-debug,
--enable-symbols, or --enable-crashdump are passed to ./configure (just
like it already is for dmake modules), as opposed to th

Add debug symbols to gbuild modules when any of --enable-debug,
--enable-symbols, or --enable-crashdump are passed to ./configure (just
like it already is for dmake modules), as opposed to the previous behaviour
of only doing it on --enable-debug.

Also implemented --enable-symbols=small for gbuild modules on platforms
where AOO is built with GCC and Clang compilers (MSVC on Windows doesn't
seem to support that).

Patch by: me


git-svn-id: https://svn.apache.org/repos/asf/openoffice/trunk@1780246 13f79535-47bb-0310-9956-ffa450edef68

show more ...

Fix a FreeBSD regression in the gbuild gb_CPUDEFS variable caused by
r1773166, where it was always set to POWERPC64 due to a swapped if-else.

This needs further work, on all platforms!

Fix a FreeBSD regression in the gbuild gb_CPUDEFS variable caused by
r1773166, where it was always set to POWERPC64 due to a swapped if-else.

This needs further work, on all platforms!

Patch by: me



git-svn-id: https://svn.apache.org/repos/asf/openoffice/trunk@1780134 13f79535-47bb-0310-9956-ffa450edef68

show more ...

Don't use the "archive" package format, which is to be extrated by
smoketest as a side effect, for the subsequent tests. The smoketest
is currently broken and won't do that, and the path is w

Don't use the "archive" package format, which is to be extrated by
smoketest as a side effect, for the subsequent tests. The smoketest
is currently broken and won't do that, and the path is wrong anyway,
as it was for OpenOffice 3 and we're now on 4.

Instead, use the office instance from the "installed" package format
for subsequent tests, as it doesn't have to zipped up and unzipped,
resulting in faster building and faster testing, and doesn't require
a side effect in a module that will probably be moved from main/
to test/.

Patch by: me


git-svn-id: https://svn.apache.org/repos/asf/openoffice/trunk@1779838 13f79535-47bb-0310-9956-ffa450edef68

show more ...

Fix a FreeBSD (and probably Solaris) regression in commit 1776288
caused by listing static libraries before objects, causing the
main/slideshow module to fail to build, as the one-pass linkin

Fix a FreeBSD (and probably Solaris) regression in commit 1776288
caused by listing static libraries before objects, causing the
main/slideshow module to fail to build, as the one-pass linking described
in that commit won't find anything from the static library as it appears
too early. Rather link all libraries after objects, like Linux does.

Patch by: me



git-svn-id: https://svn.apache.org/repos/asf/openoffice/trunk@1776379 13f79535-47bb-0310-9956-ffa450edef68

show more ...

I've heard rumours it's a POSIX feature, but so far the problem only
appears on Linux, not FreeBSD.

Linking is one-pass, using symbols exported by libraries to resolve
missing symbol

I've heard rumours it's a POSIX feature, but so far the problem only
appears on Linux, not FreeBSD.

Linking is one-pass, using symbols exported by libraries to resolve
missing symbols in previous, but *NOT* subsequent, objects (including
libraries). This means the order libraries are given to the linker
does matter, because symbols missing in latter objects cannot be
satisfied by former objects.

Static libraries should be therefore be linked to before dynamic
libraries, as static libraries cannot specify dependencies on dynamic
libraries, yet might need symbols from them.

This should fix the main/slideshow build problem on Linux.

Only Linux has been tested. The other platforms are guessed.

Patch by: me



git-svn-id: https://svn.apache.org/repos/asf/openoffice/trunk@1776288 13f79535-47bb-0310-9956-ffa450edef68

show more ...

Update endianness detection on FreeBSD.

This produces a working AOO on FreeBSD PowerPC64!

This is related to i126615 which was also fixed by Curtis a while ago.
Kudos for both c

Update endianness detection on FreeBSD.

This produces a working AOO on FreeBSD PowerPC64!

This is related to i126615 which was also fixed by Curtis a while ago.
Kudos for both contributions, makes me want to get a PowerPC to run
FreeBSD + AOO on it!

Author: Curtis Hamilton -- cmhamilto at gmail


git-svn-id: https://svn.apache.org/repos/asf/openoffice/trunk@1773166 13f79535-47bb-0310-9956-ffa450edef68

show more ...

Merge branches/gbuild-reintegration to trunk.



git-svn-id: https://svn.apache.org/repos/asf/openoffice/trunk@1755455 13f79535-47bb-0310-9956-ffa450edef68

#i125003# Eliminate cppunit from our tree completely.



git-svn-id: https://svn.apache.org/repos/asf/openoffice/trunk@1707955 13f79535-47bb-0310-9956-ffa450edef68

#i125003# Provide gbuild integration for Google Test. Windows, Linux and FreeBSD have
been tested, the other platforms are educated guesses.



git-svn-id: https://svn.apache.org/

#i125003# Provide gbuild integration for Google Test. Windows, Linux and FreeBSD have
been tested, the other platforms are educated guesses.



git-svn-id: https://svn.apache.org/repos/asf/openoffice/trunk@1703444 13f79535-47bb-0310-9956-ffa450edef68

show more ...

Don't pass -finline-limit=0 and -fno-default-inline command line options to Clang
during debug builds, since it doesn't want to run with them.



git-svn-id: https://svn.apache.or

Don't pass -finline-limit=0 and -fno-default-inline command line options to Clang
during debug builds, since it doesn't want to run with them.



git-svn-id: https://svn.apache.org/repos/asf/openoffice/trunk@1700874 13f79535-47bb-0310-9956-ffa450edef68

show more ...

FreeBSD build fixes.

This allows out the box builds with gcc and to
simplify the build with clang and also the FreeBSD port.

From Don Lewis (FreeBSD port maintainer):

B

FreeBSD build fixes.

This allows out the box builds with gcc and to
simplify the build with clang and also the FreeBSD port.

From Don Lewis (FreeBSD port maintainer):

Because we need to use different CFLAGS for gcc and clang, I had to add
some compiler detection logic. On most platforms, the value of $(COM)
is either set statically by set_soenv, or set_soenv parses the compiler
name to figure out which compiler is being used and then set $(COM)
appropriately. The latter doesn't work for FreeBSD because cc could
either be gcc or clang. For FreeBSD, I added the compiler detection
logic to configure, which then passes that to set_soenv, in a somewhat
hackish manner.

When building with ports gcc on FreeBSD, we need to pass the rpath for
the gcc runtime to the linker. The FreeBSD port attempts to to this by
adding this information to LDFLAGS, which the openoffice configure
script then steps on, and in any case, this does not help the out of the
box build. My solution is to add some logic to configure to generate
the necessary linker flags, which it then passes to set_soenv for
inclusion in FreeBSD*Env.Set.sh.

On FreeBSD, the out of the box build needs to pass $LIBINTL_PREFIX in
the environment to the build phase. I added some code to configure to
figure out the value of this variable and to pass it to set_soenv for
inclusion in FreeBSD*Env.Set.sh so that this does not need to be done as
a extra step in the build.

Changing $(COM) from GCC to CLANG for clang builds caused a number
regressions elsewhere in the build framework. These were mostly caused
by the framework checking for $(COM) == GCC and $(OS) == FREEBSD, with
$(COM) == CLANG case unhandled. The fix was generally to just ignore
the value of $(COM) and only test the value of $(OS). One special case
was the bridgetest regression test, which started dumping core on INTEL
32-bit when built with clang. It turns out that this entire test is was
skipped for $(COM) == gcc, $(OS) == FREEBSD, and $(CPU) == I. Rather
than also skipping this test when building with clang, I tracked down
the failure to a particular subtest involving polymorphic structures
that also fails on OS/2 and tweaked the code to also skip that subtest
on FREEBSD INTEL (32-bit). Now bridgetest is run and passes on FreeBSD
with both gcc and clang, on both i386 and amd64.

Submitted by: Don Lewis


git-svn-id: https://svn.apache.org/repos/asf/openoffice/trunk@1694132 13f79535-47bb-0310-9956-ffa450edef68

show more ...

FreeBSD build structure changes

Part of
Code Review: https://reviews.freebsd.org/D2108

Author: Don Lewis (truckman at FreeBSD)


git-svn-id: https://svn.a

FreeBSD build structure changes

Part of
Code Review: https://reviews.freebsd.org/D2108

Author: Don Lewis (truckman at FreeBSD)


git-svn-id: https://svn.apache.org/repos/asf/openoffice/trunk@1669465 13f79535-47bb-0310-9956-ffa450edef68

show more ...