On Mon, 09/25 11:11, Paolo Bonzini wrote: > On 25/09/2017 05:09, Fam Zheng wrote: > > On Fri, 09/22 17:49, Paolo Bonzini wrote: > >> diff --git a/tests/docker/common.rc b/tests/docker/common.rc > >> index 6865689..1522aab 100755 > >> --- a/tests/docker/common.rc > >> +++ b/tests/docker/common.rc > >> @@ -36,3 +36,11 @@ build_qemu() > >> $QEMU_SRC/configure $config_opts > >> make $MAKEFLAGS > >> } > >> + > >> +install_qemu() > >> +{ > >> + make install $MAKEFLAGS DESTDIR=$PWD/=destdir > > > > Why provide DESTDIR? build_qemu already has "--prefix=$INSTALL_DIR", can a > > simple "make install $MAKEFLAGS" work? Is there a permission problem? > > Just to test the $DESTDIR case. I have plans to extend the test a bit > further, such as: > > - checking that $INSTALL_DIR remains empty > > - comparing the destdir installation against the $INSTALL_DIR (to ensure > that the destdir doesn't end up in the installed files; distro packaging > conventions usually don't like that). > > - doing the DESTDIR install as a non-root user > > >> + ret=$? > >> + rm -rf $PWD/=destdir > > > > Why is this cleanup needed given the container will go away anyway? > > Again it's just for cleanliness. Another possible test to add is that > "make distclean" actually leaves the build tree in a pristine state, and > this would fail if we leave the installation tree around. > > All these tests, by the way, are inspired by the "make distcheck" > produced by Automake.
Yeah, this is fine. We can add pure "make install" in addition. I'll queue this patch. Thanks. Fam