Explicity set $JAVA_HOME in startup wrapper to the OpenJDK version used
for the build to avoid any potential issues with relying on javavmwrapper
parsing bsd.java.mk to determine this at runtime.
Fix an botch in the FreeBSD-specific $LD_LIBRARY_PATH code in the
startup wrapper.
Java is needed at runtime, so flag it as a runtime dependency.
Add an option to build and install the SDK and disable it by default.
The port has been building it all along, but not installing it. We
save only a small amount of time by not building it, but we save about
70 MB of space when the port is installed.
Add an option to build and install the Wiki Publisher extension. If
it is enabled, it is now handled as an integrated extension, so the
user does not have manually use the extension manager to track it down
and install it.
Tweak an existing script to install the icons instead of using a bunch
of complicated code in the Makefile.
Nuke some unused targets in Makefiles.others. The SDK is now explicitly
handled, and the Uno Runtime Environment hasn't had the capability for
separate installation in quite some time.
Nuke an extraneous "-" from the tar command when unpacking the
installation archive during staging.
r1697312 from upstream trunk (with a few tweaks to compensate for
some other upstream changes not included here) to unbreak the build
with OpenJDK8.
Revert r399537 to unmark broken.
PR: 199591 (openjdk8 by default exp-run)
branch to 4.1.2-RC1 in order to give the latter wider exposure before
4.1.2 is released.
Add this missing missing bits of patch-i114430 that were committed
to trunk upstream but that are not in 4.1.2-RC1.
Removing --enable-symbols=small from CONFIGURE_ARGS allows us to avoid
the need to use ${STRIP_CMD}.
Microsoft fonts and the proper definitions within OpenOffice have been
added so that they are recognized as such. [1]
The Linux ports handle the mapping from the printeradmin command to the
spadmin executable in the shell wrapper. Do the same in the FreeBSD
port rather than munging the printeradmin.desktop file. Retain the
spadmin link to the wrapper for backwards compatability.
The setofficelang command went away a long time ago. Remove it from
the shell wrapper and remove the link.
Exec the OpenOffice executables from the wrapper so that the shell for
the wrapper doesn't hang around as an extra process until OpenOffice
terminates.
Nuke an extraneous shell continuation in the Makefile.
Suggested by: pfg [1]
Fix a regression in directory permissions and add a missing directory used
for shared extensions. This combination of issues could cause incomplete
cleanup on pkg removal if OpenOffice had been run by a user in the wheel
group.
Add a note about the location of the Wiki Publisher extension to
pkg-message.
bug with the upstream bug ID.
More do-install target optimization:
* Use the tar -s option to modify the path names when unpacking
the install archive into the staging directory rather than
unpacking, repacking, and unpacking again
* Pass --with-unix-wrapper to configure and patch a makefile
to get the desired Exec and Icon entries in the .desktop
files instead of patching them after they have been staged.
The Icon entries do not need to specify the .png suffix.
* Streamline the usage of ${STRIP_CMD}
* Create hard links to the wrapper for the individual apps
instead of symbolic links, and use a for loop instead of
repeating the command a bunch of times
* Centralize the plist generation, which allows the icon
installation ot be simplified
* Install missing startcenter app icon
* Mute many of the commands, but add strategic ${ECHO_CMD}s
specified but without specifying a document, that application is
started instead of presenting the user with the generic startup
window and requiring him to pick the document type (running
openoffice-*-scalc should open to a new spreadsheet, etc.)
Building with clang 3.7 is reported to have the same failure mode
as clang 3.6, so use the same optimization tweak as a workaround.
Various cleanups to do-install:
* The paths ${PREFIX}/${INSTALLATION_BASEDIR} and
${PREFIX}/${INSTALLATION_BASEDIR}/${AOODIR}, and
${PREFIX}/${INSTALLATION_BASEDIR}/${AOODIR}/share/xdg are
frequently used, so create a few variables to hold those paths
instead of repeating the same, long sequence in a number of places
* Tweak openoffice-wrapper to take advantage of the previous change
* Use INSTALLS_ICONS=yes instead of manually adding the @exec and
@unexec entries to the plist
* Edit bootstraprc only once with ${REINPLACE_CMD} instead of
running it twice to edit the same line
* Remove bootstraprc.bak so that it does not get installed
* With the advent of staging, it is not necessary to remove
${STAGEDIR}${DESKTOPDIR}/${EXECBASE}
* Simplify code for installing icons
Rename remaining OOO* variables to AOO*.
Nuke patch-clang34 because it has been incorporated upstream.
Replace the default images in the opening splash window and the
Help->About window with FreeBSD-branded versions that also mark
this as a developer snapshot. Thanks to pfg for suggesting this
and reviewing the new images.
UNIQUENAME was never unique, it was only used by USE_LDCONFIG and now,
we won't have conflicts there.
Use PKGBASE instead of LATEST_LINK in PKGLATESTFILE, the *only* consumer
is pkg-devel, and it works just fine without LATEST_LINK as pkg-devel
has the correct PKGNAME anyway.
Now that UNIQUENAME is gone, OPTIONSFILE is too. (it's been called
OPTIONS_FILE now.)
Reviewed by: antoine, bapt
Exp-run by: antoine
Sponsored by: Absolight
Differential Revision: https://reviews.freebsd.org/D3336
building with gcc from ports on FreeBSD. Remove the REINPLACE_CMD
patches in the ports Makefile and $FBSD_LDFLAGS from do-build that
are no longer needed for building with gcc in the general case.
Add a patch to disable optimization when compiling one source file
with base clang 3.4 on FreeBSD 10 amd64 to avoid tripping on a compiler
code generation bug. This allows us to build with base clang on
FreeBSD 10 amd64 instead of using the clang35 port.
Fine tune compiler optimization flags when using clang 3.6 on FreeBSD 11
i386 or gcc 4.9 instead of totally disabling optimization. Unbreak
the build with gcc 4.9 on amd64.
A few Makefile cleanups without functional change:
* Combine two .if ${ARCH} == amd64 sections of Makefile
* Merge the pre-configure target into post-extract
* Simplify the code in the do-build target
Approved by: mat (mentor, implicit)
patch-jpeg and patch-nss have been upstreamed and are no longer needed.
patch-i118574 no longer seems necessary for a clean build.
Regnerate patch-i114430 with make makepatch.
Always build using boost from ports instead of the bundled boost, updating
the BUILD_DEPENDS with the header that configure looks for.
Add missing LIB_DEPENDS that are brought in by other dependencies, but are
directly linked as requested by pkgconfig/*.pc.
Build using sane and xrender headers from ports instead of bundled headers.
Update LICENSE* to account for bundled software.
Reformat pkg-message and mention that the scanner interface can be enabled
by installing sane-backends.
Pet portlint.
Reviewed by: pfg
Approved by: mat (mentor)
Differential Revision: https://reviews.freebsd.org/D3064
does not succeed in any case.
The do-build target does not need to use bash, or even an extra
level of shell.
There are two levels of parallelism in the openoffice build framework.
Split MAKE_JOBS_NUMBER between the two levels instead of potentially
running MAKE_JOBS_NUMBER^2 compilations in parallel.
PR: 199930
Approved by: mat (mentor, implicit)
There is a missing dependency on devel/dbus-glib that causes the
openoffice build to fail when the GCONF and GNOMEVFS options are
disabled. When these options are enabled, one of the dependencies
added by these options brings along dbus-glib so the default build
succeeds.
There is a configure knob to control whether or not dbus-glib is
used, so follow the lead of the libreoffice port and rename the
GCONF option to GNOME and bundle both Gconf and DBUS handling under
the GNOME option.
Borrow the MMEDIA option from the libreoffice port and use that to
control the using of gstreamer.
Add a couple of missing p5-* build dependencies.
Add the --enable-gtk, --disable-kde, --disable-kde4 configure flags
so that the port does not try to build with KDE if it is installed
because compilation fails.
USE_AUTOTOOLS=autoconf -> USES=autoreconf:autoconf
PR: 199865
Reported by: pi
Differential Revision: https://reviews.freebsd.org/D2441
Reviewed by: pi, pfg
Approved by: mat (mentor)
Word around broken build on FreeBSD 11.0-CURRENT i386, which appears
to be a code generation bug in clang 3.6 when using -Os optimization.
Build using the ports version of graphics/silgraphite instead of
building an internal copy.
When building with clang, use the ports version of devel/boost-libs
instead of an internal copy.
Use the dictionaries already installed by ports.
Update the patch-bridges and patch-*.mk patches to versions that
will be sent upstream. The latter are tuned for clang to use as
is, with only some minor additional tweaks needed for gcc.
Nuke patch-webdav because it has been incorporated upstream.
Wordsmith pkg-message.
Various Makefile cleanups.
Differential Revision: https://reviews.freebsd.org/D2108
Reviewed by: pfg
Approved by: mat (mentor)
Xorg is now looking in ${LOCALBASE}/share/fonts by default
Xorg now accepts symlinks in etc/X11/fontpath.d (as decribed in Xserver(1))
Large cleanup on lots of font ports
All fonts are now properly dynamically generating fonts.dir and fonts.scale
instead of sometime overwriting existing ones)
All fonts are generating fontconfig's cache
Improve consistency in fonts ports
where clang is the base compiler. The issue was that these ports
would only successfully build with gcc and libstdc++, so they
specified USE_GCC=yes, but they linked to other C++ ports that were
compiled with clang, which brought in libc++. The conflict between
libstdc++ and libc++ caused the application to crash whenever an
operation that popped up a dialog box was attempted. Thanks to
dim@ for helping me track this down. The fix is to patch various
bits of the openoffice souce to allow it to be built with clang
on systems where the C++ dependencies are also compiled with clang. [1]
Add a CUPS option so that CUPS can be disabled [2].
Register print/cups-client as a LIB_DEPENDS when CUPS is enabled.
pkg-message claims that user settings are stored in
~/.openoffice,org4, whereas all other platforms seem to use
~/.openoffice.org/4 (or equivalent), and both openoffice-4 and
openoffice-devel actually use ~/.openoffice.org-devel/4. The
addition of -devel to the location happened with r325370.
The / appears to have been introduced in r297259. Change the
location match other platforms. Introduce a new variable
${AOOUDIR} so that the actual location and pkg-message stay in
sync.
Rename ${OOOTAG} to ${AOOTAG} and restore its value so that it
can once again be substituted into pkg-message. It has not
been set since r296269.
Various Makefile cleanups:
* Gather and sort USE_*
* Simplify use of ${REINPLACE_CMD}
* --x-includes and --x-libraries are automatically passed to configure,
which ignores them
* Get rid of unnecessary include of bsd.port.options.mk
PR: 188088 [1]
PR: 198458 [2]
Differential Revision: https://reviews.freebsd.org/D2055
Reviewed by: pfg
Approved by: mat (mentor)
The problem is that libreoffice installs its own copy of unopkg in
${PREFIX}/bin and that the openoffice build attempts to use this copy
of unopkg instead of the openoffice version which is in subdirectory
of ${WRKSRC}.
The reason is that the openoffice build expects to find its own copy
of unopkg by depending on having "." in its $PATH, but its $PATH has
${PREFIX}/bin before ".". Openoffice attempts to do the right thing
by first constructing $PATH by prepending "." and a small number of
other directories where it stashes executables used during the build
to the value of $PATH that it inherits from the environment. Things
go wrong when it tries to add the paths for ${CC}, perl, and java
to $PATH. If $PATH has /usr/bin before ${PREFIX}/bin, the openoffice
build finds the perl symlink in /usr/bin before it finds perl in
${PREFIX}/bin, so it prepends ${PREFIX}/bin to $PATH to try to ensure
that the correct version of perl will be found first. This moves
${PREFIX}/bin earlier in $PATH than ".".
The operation to put the path to ${CC} in $PATH has a different
problem. It uses the variable COMPATH for this, which is the dirname
of the patch to $CC, with the trailing /bin stripped off. That
results in /usr/local being added to $PATH, which is nonsensical,
though mostly harmless.
There are three fixes here:
* Always keep the parts of $PATH for the directories under ${WRKSRC},
including "." at the beginning of the path.
* Ignore symlinks to executables when deciding to prepend a directory
to the path.
* Append "/bin" when using COMPATH so that the result points to the
directory where ${CC} actually resides. There is actually another
variable CC_PATH, but it has an extra trailing "/", so it doesn't
match ${PREFIX}/bin.
Differential Revision: https://reviews.freebsd.org/D1958
Reviewed by: pfg
Approved by: mat (mentor)
Make sure we use openjpeg1 in poppler, while openjpeg2 support
was added there still bugs in openjpeg2 that need to be fixed.
Add patches to inkscape for API changes in poppler 0.29.0 [2].
PR: 196599 [1]
Submitted by: olivierd@ [1]
Exp-run by: antoine@
Obtained from: upstream [2]
Resynchronize to openoffice-4 including:
* portlint fixes
* fix all but one stage-qa warning
* add source makefile patch to unbreak --with-system-jpeg
* add the --with-system-hyphen, --with-system-nss, --with-system-jpeg,
and --with-system-mythes configure options
* remove the --with-system-saxon option to allow coexistance with
math/scilab
* add all distfiles that were previously downloaded during build
to the ext_sources distfile, and add --disable-fetch-external
* make the graphics/png dependency explicit
* change from USE_GCC=yes to USE_GCC=4.8 because of a compile error with
gcc 4.9
* remove SVNREVISION from the ext_sources tarball name and
tweak how it is extracted to allow it to be shared by
multiple AOO releases
* remove trailing cruft from files/patch-sal (from pfg@)
* nuke files/patch-sal-inline (from pfg@)
* modernize MASTER_SITES syntax and update MASTER_SITE for unowinreg.dll
* unbreak sdk, package-canonical, and languagepack targets (from pfg@)
* don't truncate generated plist (from mat@)
* don't add @dirrm to generated plist, but add @dir for empty directories
* don't add license files to generated plist because the framework
does this (from mat@)
* don't add a blank line to generated plist
* merge post-install target into do-install (from mat@)
* misc Makefile cleanup
PR: 192557
Differential Revision: https://reviews.freebsd.org/D1571
Reviewed by: pfg
Approved by: mat (mentor)
a zeising, kwm production, with help from dumbbell, bdrewery:
NEW XORG ON FREEBSD 9-STABLE AND 10-STABLE
This update switches over to use the new xorg stack by default on FreeBSD 9
and 10 stable, on osversions where vt(9) is available.
It is still possible to use the old stack by specifying WITHOUT_NEW_XORG in
/etc/make.conf .
FreeBSD 8-STABLE and released versions of FreeBSD still use
the old version.
A package repository with binary packages for new xorg will
be available soon.
This patch also contains updates of libxcb and related ports, pixman, as well
as some drivers and utilities.
Bump portrevisions for xf86-* ports, as well as virtualbox-ose-additions due
to xserver version change.
Apart from these updates, the way shared libraries are handled has been
changed for all xorg ports, as well as libxml2 and freetype, which means
ltverhack is gone and as a consequence shared libraries have been bumped.
The plan is that this change will make library bumps less likely in the
future.
All affected ports have had their portrevisions bumped as a consequence of
this.
Fix some issues where WITH_NEW_XORG weren't detected properly on CURRENT.
Update instructions, hardware support, and more notes can be found on
https://wiki.freebsd.org/Graphics
Thanks to: all testers, bdrewery and the FreeBSD x11@ team
exp-run by: bdrewery [1]
PR: ports/187602 [1]
Approved by: portmgr (bdrewery), core (jhb)
GCC 4.6.4 to GCC 4.7.3. This entails updating the lang/gcc port as
well as changing the default in Mk/bsd.default-versions.mk.
Part II, Bump PORTREVISIONs.
PR: 182136
Supported by: Christoph Moench-Tegeder <cmt@burggraben.net> (fixing many ports)
Tested by: bdrewery (two -exp runs)
