allows to build it with bmake(1) (as well as other implementations)
- Unbreak the build against Clang (main() must return int)
- Sanitize formatting of the port description while I am here
Reported by: marino
bmake(1), which does not play nice with libtool way of building things
- While here, convert LIB_DEPENDS to new form, sort the knobs a bit, and
hook test target to our standard "regression-test" one
Reported by: marino
Tested by: marino (poudriere testport)
Approved by: miwi, bapt (portmgr, implicit)
"make fetch" failure on non-POSIX-compliant SHELL:
1. Replace ${SHELL} by ${SH} so the fetchwrapper.sh gets executed with a
POSIX-compliant shell (such as /bin/sh)
2. Just because it's meant that way, add an svn:executable=* property
on files/fetchwrapper.sh.
PR: ports/181252
Reported by: Mike Harding (private email)
This update include a patch for sgen which avoids an assertion failure
(condition `info->doing_handshake' not met) that will be part of 3.2.2.
Because our mono version is unstable due to this, we decided not to wait for
3.2.2 to be out and update to 3.2.1, cherry-picking the appropriate fix.
Many thanks to Charles Randall for reporting and helping.
- Trim Makefile header
- Remove dead link from MASTER_SITES
- Remove indefinite article from COMMENT
- Add LICENSE (BSD)
- Convert to the new options framework
- Update WWW field in pkg-descr
Build details: http://goo.gl/l5Fmwu
- Trim header
- Remove WWW indent
- Remove indefinite article from COMMENT
There are other complaints by portlint that need reviewing by maintainer.
Approved by: bapt/culot (mentors, implicit)
Many thanks to the Mike Bernard for reporting and helping me debug
this problem.
While here remove obsolete MAKE_JOBS_SFE flag
Reported by: Mike Barnard Kwatampora <mike.barnardq@gmail.com>
bmake(1) now. The build fails because bmake(1) treats MFLAGS variable
slightly differently: while by default its value would be '-j X', our
old make(1) allows to set it from the makefile or environment, bmake(1)
will append to it instead. Pass the empty value via MAKE_ARGS to make
every make(1) happy, and thus fix parallel builds (once again).
1) `devel/libnotify' no longer pulls gtk20; we must depend explicitly;
2) `x11-toolkits/pango' now depends on `print/harfbuzz', which installs
header file ($localbase/include/harfbuzz/hb.h) that in turn confuses
bundled libass' configure script and conflicts with HandBrake's own
internal header.
NB: supposedly it should have been caught by exp-run for r324037, but it
was not for some reason.
Reported by: Sergey V. Dyatko
