On Friday 05 August 2011, Stefano Lattarini wrote:
> Here is a brief illustration of the issue:
> 
>   $ cat > Makefile.am <<'END'
>   AUTOMAKE_OPTIONS = foreign parallel-tests -Wall -Werror
>   TEST_LOG_COMPILER = false
>   TESTS = foo.test bar.test
>   END
>   $ touch foo.test bar.test
>   $ autoreconf -vi && ./configure
>   ...
> 
>   $ freebsd-make -j2 check; echo %% STATUS = $? %%
>   freebsd-make  check-TESTS
>   FAIL: foo.test
>   FAIL: bar.test
>   ====================
>   2 of 2 tests failed
>   See ./test-suite.log
>   ====================
>   %% STATUS = 0 %%
> 
> It works as expected with GNU make and NetBSD make though:
> 
>  $ gmake -j2 check; echo %% STATUS = $? %%
>   gmake  check-TESTS
>   gmake[1]: Entering directory `/tmp/...'
>   gmake[2]: Entering directory `/tmp/...'
>   FAIL: foo.test
>   FAIL: bar.test
>   ====================
>   2 of 2 tests failed
>   See ./test-suite.log
>   ====================
>   gmake[2]: *** [test-suite.log] Error 1
>   gmake[2]: Leaving directory `/tmp/...'
>   gmake[1]: *** [check-TESTS] Error 2
>   gmake[1]: Leaving directory `/tmp/...'
>   gmake: *** [check-am] Error 2
>   %% STATUS = 2 %%
> 
> 
>   $ netbsd-make -j2 check; echo %% STATUS = $? %%
>   --- check-am ---
>   netbsd-make  check-TESTS
>   --- check-TESTS ---
>   --- foo.log ---
>   --- bar.log ---
>   FAIL: bar.test
>   --- foo.log ---
>   FAIL: foo.test
>   --- test-suite.log ---
>   ====================
>   2 of 2 tests failed
>   See ./test-suite.log
>   ====================
>   *** [test-suite.log] Error code 1
>   1 error
>   netbsd-make: stopped in /tmp/...
>   *** [check-TESTS] Error code 2
>   1 error
>   netbsd-make: stopped in /tmp/...
>   *** [check-am] Error code 2
>   1 error
>   netbsd-make: stopped in /tmp/...
>   %% STATUS = 2 %%
> 
> The problem is present both in automake 1.11.1 and in the developement
> version from the 'test-protocols' branch.
> 
OK, the attached hacky patch seems to fix the bug.  I have no idea why
exactly it is so yet, so we might want to wait to apply the patch until
we have fully understood the reasons of the original failure, and why
the proposed patch seems to fix it.

Regards,
  Stefano