Some cmake projects use LIB_DESTINATION/pkgconfig to install the pc-files.
For example:
* deskutils/libstreamanalyzer
* sysutils/polkit-qt{,5}
* x11-toolkits/attica
By adding this, we can convert these ports to use pathfix instead of an
uncessessary patch-foo or confusing sed call.
Reviewed by: mat
Approved by: portmgr (mat), mat (mentor)
Differential Revision: https://reviews.freebsd.org/D8141
SeaMonkey now uses FFmpeg directly rather than via GStreamer for H.264
and MP3 playback as part of HTML5. Note, FFmpeg 3.x is unsupported[1].
Changes: http://www.seamonkey-project.org/news#2016-03-14
PR: 210931
Submitted by: net@arrishq.net (based on)
Security: 2c2d1c39-1396-459a-91f5-ca03ee7c64c6
MFH: 2016Q3
[1] SeaMonkey 2.40 is based on Firefox 43 codebase but Firefox 45
is the earliest to have the fix. Backporting maybe tricky:
https://bugzilla.mozilla.org/show_bug.cgi?id=1232268
USES=localbase:ldflags can be used to set LDFLAGS. Normally LDFLAGS
appears too early on the command line causing some ports to link with
their own libraries in LOCALBASE (if installed) instead of WRKSRC.
Also make use of _USES_POST so -L${LOCALBASE}/lib is added as late as
possible after anything a port Makefile might set. Use _USES_POST
instead of .include in libedit.mk and libarchive.mk so things like
'USES=libedit localbase:ldflags' work correctly.
Fix some issues with LIBS in some ports.
Switch ports that don't support LIBS to localbase:ldflags.
PR: 212987
Exp-run by: antoine
Approved by: portmgr (antoine)
hide actual commands executed and only show short summary line (like
"CC foo.c"). CMake and ninja enable this by default, some autotools
using ports do as well. This is unacceptable because we need complete
build logs at any time, so we now switch to verbose build logs
unconditionally. Note that this change deliberately affects ALL
builds and not only package builds on cluster, because we need to
be sure that user experiencing failure can always provide informative
build log regardless of settings and without rerunning the build.
Change summary:
- Always do verbose builds for cmake, ninja and GNU configure (the
latter includes check if --disable-silent-rules is actually supported
by the configure script; there are isolated cases when it's not true)
- Remove CMAKE_VERBOSE, NINJA_VERBOSE and
CONFIGURE_ARGS=--disable-silent-rules from all ports which set them
for this is no longer needed
- Revert hacks for --disable-silent-rules support priorly committed
to biology/ncbi-blast+ and net-p2p/mldonkey - no longer needed as well
Submitted by: amdmi3
Reviewed by: mat
Exp-run by: antoine
Approved by: portmgr (mat, antoine)
Differential Revision: D7534
There are three cases here:
1) USES=ncurses or USES=ncurses:base and devel/ncurses is not present:
use base, no need to add rpath as there is only one libncurses.so
present, in /usr/lib
2) USES=ncurses and devel/ncurses is present or USES=ncurses:port: use
ports, and add rpath to LOCALBASE/lib so that the right one is picked
3) USES=ncurses:base and devel/ncurses is present fail (it would warrant
using -rpath /usr/lib, but it error's out during the sanity stage)
PR: 211592
Reported by: koobs (and many others)
Reviewed by: bapt
Sponsored by: Absolight
Differential Revision: https://reviews.freebsd.org/D7459
Bump PORTREVISION to account for slightly different behavior in other
gecko@ ports and to get a discrete regression window.
Changes: https://www.mozilla.org/firefox/49.0/releasenotes/
PR: 211792
Submitted by: Jochen Neumeister <joneum@bsdproject.de> (based on)
Security: 2c57c47e-8bb3-4694-83c8-9fc3abad3964
MFH: 2016Q3
*_DEPENDS. Helper variables derived from the loop variable aren't
expanded causing *_DEPENDS to be incorrect.
- Bump all ports with more than one value in USE_LINUX.
PR: 211645
Reported by: jkim
Mk/Uses/linux.mk.
- Replace USE_LINUX=yes with USES+=linux and USE_LINUX=(.*) with
USES+=linux:\1 in all ports.
- Replace USE_LINUX_APPS with USE_LINUX in all ports.
- Use INSTALL_SCRIPT instead of INSTALL_PROGRAM to install scripts in some
ports.
- When USE_LINUX_RPM is defined, simplify the way DISTFILES and EXTRACT_ONLY
are defined.
- Remove BRANDELF_DIRS and BRANDELF_FILES handling. In the very rare cases
that it is still necessary ports can run ${BRANDELF} from post-patch.
- Remove AUTOMATIC_PLIST handling. Only one port used it.
- Fix Linux MASTER_SITES.
- Replace OVERRIDE_LINUX_BASE_PORT and OVERRIDE_LINUX_NONBASE_PORTS with
default versions framework.
- bsd.port.mk:
- Move Linux related bits to Uses/linux.mk, except USE_LINUX_PREFIX.
- Put USE_LINUX_PREFIX handling after USES processing.
- Define DOCSDIR, DATADIR, etc. after handling USE_LINUX_PREFIX so it can
give these variables a different default value.
- When a package needs to run Linux ldconfig check before installation if
Linux support is enabled.
- emulators/linux_base-*:
- Use USES=linux and remove duplication.
- Remove files/lp. FreeBSD or CUPS lp(1) should work.
- Remove files/yp.conf. No longer seems to be used.
- Remove pkg-deinstall and move pkg-install into pkg-plist.
- Update pkg-descr and pkg-message.
- Fix handling of ldconfig cache in pkg-plist.
- devel/fb-adb: Use a Linux shell to run a Linux script but patch the script
to use FreeBSD mkdir so mkdir -p $path creates $path and not
/compat/linux/$path.
PR: 211645
Exp-run by: antoine
Approved by: portmgr (antoine)
R warns if it detects .o or .so objects in the source directory, when
running tests. Remove the offending files before running the test.
Take maintainership [1]
PR: 212134
Approved by: wen [1]
Lots of KDE components are providing libraries, and they were,
strangely, written as BUILD_ or RUN_DEPENDS, with a full path to the
library. Change this so that if a component is needed at both build,
and, run-time, then it gets a LIB_DEPENDS.
Reviewed by: rakuco, tcberner
Sponsored by: Absolight
Differential Revision: https://reviews.freebsd.org/D7625
Plasma5 ports
At the moment KDE ports use bsd.kde4.mk to handle their dependencies. When
working on the ports for KDE Frameworks and Plasma5 it seemed to be more
reasonable to create a new kde.mk instead of adding an bsd.kde5.mk.
The kde.mk in this review is a stripped down version of the one we are using in
the KDE Test repositories plasma5 branch [1] to only contain the parts relevant
to the current KDE4 ports in the portstree [2].
Changes to the KDE Ports needed by this:
Replace USE_KDE4 by USE_KDE [3]
Add USES=kde:4 [4]
[1] http://src.mouf.net/area51/view/branches/plasma5/KDE/Mk/Uses/kde.mk
[2] The version in the plasma5 branch also handles frameworks/plasma5 and
handles MASTER_SITES via a KDE_DIST variable similar to bsd.qt.mk for Qt
Ports -- I chose to leave this out for now, as the diff is already large
enough.
[3] I chose USE_KDE instead of USE_KDE4, USE_KDE5, USE_KDEX as the version we
want is already specified as argument to kde:<arg>
[4] For KDE Frameworks and Plasma5 ports this would be kde:5
PR: 210667
Approved by: portmgr, mat (mentor), rakuco (mentor)
Reviewed by: mat, rakuco
Differential Revision: https://reviews.freebsd.org/D6961
The "8" refers to FreeBSD 8.x which is no longer supported by the
ports tree. It's never been used (to my knowledge) so just remove the
option altogether.
Due to the change in r419666 USE_GCC is not conditionally set, however it
was bringing in the dependencies for fortran (silently). Explicitly bring
in the dependencies for fortran.
Also, add a target to strip the compiled shared library.
PR: 211891
Reviewed by: mat
Approved by: wen
The upcoming KDE Frameworks ports need grantlee5
* Add devel/grantlee5 and move installed headers into a subdirectory include/grantlee5
* Move header files from devel/grantlee into a subdirectory
include/grantlee4
to make sure ports do not pick up the wrong headers
* Add Uses/grantlee.mk to handle the PLIST_SUB and LIB_DEEPENDS needed
by ports using grantlee (before we set the PLIST_SUB manual in very
of the depending ports, which now should not be needed anymore).
* The ports depending on devel/grantlee have been modified from
LIB_DEPENDS=libgrantlee_gui.so:devel/grantlee
to
USES=grantlee:4
as this does not affect the package no version bumped was made.
Reviewed by: mat, rakuco
Approved by: rakuco (mentor)
Differential Revision: https://reviews.freebsd.org/D7434
file, so the order remains the same.
Every PHP (or Zend) extension now installs its own .ini file in
/usr/local/etc/php. A PHP extension will be automatically activated
when installed. The order into which extensions are loaded is
automatically guessed. In some very rare cases, the guess will be
wrong, and PHP_MOD_PRIO will need to be set. Refer to the USES=php
section of the Porter's Handbook for more information.
Convert ports touching etc/php/extensions.ini manually, or telling the
OP to do it.
And finally, bump PORTREVISION for all php extensions.
PR: 210697
Submitted by: mat
Sponsored by: Absolight
Differential Revision: https://reviews.freebsd.org/D7022
The variable defined in it are now always available after including
bsd.port.pre.mk.
PR: 210666
Submitted by: mat
Exp-run by: antoine
Sponsored by: Absolight
Differential Revision: https://reviews.freebsd.org/D6933
- If a port links to libtk then it will always also link to libtcl. Include
libtcl in the LIB_DEPENDS when linking to libtk.
- Fix indentation of '.if' statements
PR: 211261
Approved by: gahr
MFH: 2016Q3
print/texinfo which requires gmake. To break this circular dependency
gmake-lite was introduced. This can be removed now as well.
PR: 210623
Reviewed by: bapt
Exp-run by: antoine
Approved by: portmgr (antoine)
postgresql-9.0.x was declared EoL in September 2015.
Summary:
Remove 9.0 from the list of postgresql versions available in ports
Disconnect postgresql90 ports from the build
Remove postgresql90-pgtcl port
Remove postgresql90-client port
Move the master postgreslXY-plperl makefile to postgresql95-plperl/Makefile.
Adjust include lines in other postgresqlXY-plperl ports
Delete postgresql90-plperl
Move the master postgreslXY-plpython/{Makefile,pkg-descr} to
postgresl95-plpython/{Makefile,pkg-descr}
Adjust all other postgresqlXY-plpython/Makefile to include the new master
Remove postgresql90-server
Reviewers: jgh, girgen, #portmgr, O5 Ports Framework, bapt, crees
Reviewed By: #portmgr, O5 Ports Framework, bapt, crees
Subscribers: mat
Differential Revision: https://reviews.freebsd.org/D6898
Add a qa hint about needing, or not, USES=ssl.
Fix ports doing silly things, like including bsd.openssl.mk directly.
PR: 210322
Submitted by: mat
Exp-run by: antoine
Sponsored by: Absolight
Differential Revision: https://reviews.freebsd.org/D6866
PYTHON_ABIVER/PYTHON_INCLUDEDIR, when they use default options for
lang/python3x, by providing a reasonable default value to PYTHON_ABIVER
With hat: portmgr
the default when WANT_PGSQL is not defined). Compare to
WANT_PGSQL=client which only adds a RUN_DEPENDS on
postgresqlXY-client. 'client' can be part of a list of different
components, but 'lib' is ignored if any other terms are specified,
which seems an arbitrary and incorrect restriction.
WANT_PGSQL=lib allows expressing that the port links against
libpq.so.X rather than just needing one of the client programs like
psql(1) at runtime. However, other than that there is little
practical difference between 'lib' and 'client' on the resulting pkg.
Approved by: crees (pgsql@)
Sponsored by: https://reviews.freebsd.org/D6893
[1] Add "compiles" argument to cran to allow R-cran ports need to compile
code to bring in the correct compiler dependencies. Also, ensure
consistency between the compiler selected by the R port and the R-cran ports.
[2] Migrate R-cran ports that manually specify the compiler to
USES=cran,compiles.
PR: 210343
Approved by: wen (1), portmgr (2, blanket)
Turns out that the conversion to USES was broken, and did not take the
DEFAULT_VERSIONS=bdb=foo into account.
PR: 210149
Submitted by: mat
Exp-run by: antoine
Sponsored by: The FreeBSD Foundation, Absolight
