On Thu, Dec 31, 2009 at 03:52:07PM +0100, Auclair Vincent wrote: > On Thu, Dec 31, 2009 at 2:28 PM, Landry Breuil <[email protected]> wrote: > > On Thu, Dec 31, 2009 at 01:46:44PM +0100, Auclair Vincent wrote: > >> attached is the google-glog port > >> > >> Information for inst:glog-0.3.0 > >> > >> Comment: > >> c++ application-level logging > >> > >> Description: > >> The glog library implements application-level logging. This library > >> provides logging APIs based on C++-style streams and various helper > >> macros. > >> > >> Maintainer: Vincent Auclair <[email protected]> > >> > >> WWW: http://code.google.com/p/google-glog/ > >> > >> It is missing a library to correctly dump the signal handlers. > >> one of these probably : execinfo, libunwind or ucontext > > > > CONFIGURE_ARGS should use --with-gflags=${LOCALBASE}, not ${TRUEPREFIX} : > > - LOCALBASE is for already installed stuff > > - TRUEPREFIX is where the current port will actually install its stuff > > > > fails to build @amd64 here: > > src/logging.cc:1545: warning: comparison between signed and unsigned > > integer > > expressions > > src/logging.cc: In function `int google::posix_strerror_r(int, char*, > > long > > unsigned int)': > > src/logging.cc:1758: error: reinterpret_cast from `char*' to `int' loses > > precision > > > > Builds fine on macppc though. > > > > LIB_DEPENDS should precise the gflags::devel/gflags libspec it depends on. > > $make port-lib-depends-check > > LIB_DEPENDS: gflags.0 from gflags-1.2p0 > > (/usr/local/lib/libglog.so.0.0) > > > > Updated tarball contains an upstream patch for the > error on amd64. It was trying to fit an adress into an int. > Changed TRUEPREFIX to LOCALBASE > and fixed the LIBDEPENDS.
Commited, with minor tweaks: - comment in patch to link to the bug tracker issue - 'library' appended to COMMENT - fix depends : gtest is not a LIB_DEPENDS actually, glog doesn't link with it. It's a BUILD_DEPENDS, and can be inherited from REGRESS_DEPENDS. thanks!
