Eric Blake wrote: > Personally, I use: > > ./gnulib-tool --with-tests --test copy-file > > which says to include the copy-file-tests module automatically, as well > as running the unit test (and clean it up on success), when developing > on a single machine. And when trying to test on a machine that can't > run gnulib, but can share an NFS mount with a more powerful machine, I run: > > ./gnulib-tool --with-tests --create-testdir --dir=/path/to/dir module > > on the powerful machine, then on the machine under test,
On my side, I work more with 'scp' rather than NFS. So, I use ./gnulib-tool --create-testdir --with-tests --dir=/tmp/testdir1 copy-file and then dispatch a build to a number of build machines multibuild testdir1 linux linuxmips32 linuxmips64 linuxppc32 linuxppc64 \ linuxsparc32 linuxsparc64 linuxhppa linuxia64 \ macosx macosx64 freebsd64 openbsd49 netbsd51 \ aix51-cc hpux11.00-cc irix65-cc solaris7-cc solaris8-cc \ solaris9-cc solaris10x86-cc solaris11x8632 solaris11x8664 \ cygwin1.7.9 mingw2009 msvc9 Each of the identifiers corresponds to a machine description that describes how to reach the machine via 'ssh', what environment variables to set (from PATH to CPPFLAGS), and what options to pass to 'configure'. Then I can sit back and watch the log files accumulate. Bruno