On 03/06/2012 01:53 PM, Peter Rosin wrote: > Subject: [PATCH] tests: do not clobber the modified CC > Maybe s|the modified CC|the $CC set by AM_PROG_CC_C_O (for losing compilers)| here?
> If CC is originally a losing compiler, AM_PROG_CC_C_O will > modify CC. "$MAKE -e" will then clobber this modified CC > and reintroduce the raw losing compiler causing the test > to fail, as subdir-objects is in effect. > > * tests/yacc-dist-nobuild-subdir.test: Drop -e from the $MAKE > invocations. This is safe since DISTCHECK_CONFIGURE_FLAGS is not > specified in the Makefile and since the YACC override is carried > over from the configure invocation into the Makefile making it > redundant to specifiy it again in the environment for $MAKE. > --- > tests/yacc-dist-nobuild-subdir.test | 4 ++-- > 1 files changed, 2 insertions(+), 2 deletions(-) > > Hi! > > Ok for master? > > Cheers, > Peter > > diff --git a/tests/yacc-dist-nobuild-subdir.test > b/tests/yacc-dist-nobuild-subdir.test > index 67568d7..3fdae8d 100755 > --- a/tests/yacc-dist-nobuild-subdir.test > +++ b/tests/yacc-dist-nobuild-subdir.test > @@ -85,9 +85,9 @@ chmod u+w . > # Try to enable dependency tracking even with slow dependency > # extractors, to improve coverage. > ../configure --enable-dependency-tracking YACC=false > -YACC=false $MAKE -e > +$MAKE > ls -l sub/*.[ch] && Exit 1 > > -env YACC=false DISTCHECK_CONFIGURE_FLAGS='YACC=false' $MAKE -e distcheck > +env DISTCHECK_CONFIGURE_FLAGS='YACC=false' $MAKE distcheck > > : The original version of the test case was being overly paranoid to ensure the bug it tested for was truly exposed; but keeping up with such paranoia wouldn't justify a spurious failure, nor a change introducing even more hoops or extra indirections. So ACK. Thanks, Stefano