Re: [Libvir] diagnose invalid domain ID numbers (and invalid integers, in general)

"Daniel P. Berrange" <berrange(a)redhat.com&gt; wrote: > The change I applied was this: ... > TESTS_ENVIRONMENT = \ > - abs_top_builddir='$(abs_top_builddir)' \ > - abs_top_srcdir='$(abs_top_srcdir)' \ > + abs_top_builddir=`pwd`/$(top_builddir) \ > + abs_top_srcdir=`pwd`/$(top_srcdir) \ Hi Dan, That looks fine, but it'd be nice to retain the single quotes: diff --git a/tests/Makefile.am b/tests/Makefile.am index 8539779..d059402 100644 --- a/tests/Makefile.am +++ b/tests/Makefile.am @@ -56,8 +56,8 @@ endif # abs_top_{src/build}dir variables, so don't rely # on them here. Fake them with 'pwd' TESTS_ENVIRONMENT = \ - abs_top_builddir=`pwd`/$(top_builddir) \ - abs_top_srcdir=`pwd`/$(top_srcdir) \ + abs_top_builddir=`pwd`/'$(top_builddir)' \ + abs_top_srcdir=`pwd`/'$(top_srcdir)' \ $(VG) valgrind: They protect against the unusual case in which one does a VPATH (aka non-srcdir) build, where configure is specified via a name containing spaces or shell meta-characters. I'll commit the above on Monday if no one else does it first. ----------

On the portability-to-old-automake front (automake-1.9 is over 3 years old), bear in mind that people building from checked out sources should try hard to use relatively recent versions of autoconf and automake, especially developers and packagers. There are enough portability bugs in automake-1.9 that you don't want to use it in development, and certainly not when building a tarball that might be distributed. That goes double for autoconf.