Hi there.
The solution we use since several years in the Parma Polyhedra Library to the problem described in this thread
http://lists.gnu.org/archive/html/automake/2005-10/msg00058.html
does not work with the parallel test harness facility. The symptom is that recompilation with the same set of flags is repeated over and over again. The problem goes away if one reverts to the obsolete serial test harness facility, which would very much like to avoid. A Makefile.am and script affected by this issue are
http://www.cs.unipr.it/git/gitweb.cgi?p=ppl/ppl.git;a=blob_plain;f=tests/BD_... http://www.cs.unipr.it/git/gitweb.cgi?p=ppl/ppl.git;a=blob_plain;f=tests/BD_...
Does anyone spot the cause of the problem? Kind regards,
Roberto
P.S. Reading the manual we took note of the fact that we should make changes to Makefile.am, such as using LOG_COMPILER instead of TESTS_ENVIRONMENT, but these are orthogonal issue (i.e., the observation above is with an empty TESTS_ENVIRONMENT).