Created attachment 303191 [details] build log On my hardened ~amd64 machine with gcc-4.6.2 I'm unable to run the testsuite of maxima. It allways stalls for an infinite time. It does not create any load. All the processes are constantly sleeping. The last line in the log is as follows: Running tests in rtestnset: The process tree looks as follows: /usr/bin/python2.7 /usr/bin/emerge -1 maxima └─ [sci-mathematics/maxima-5.26.0] sandbox "/usr/lib64/portage/bin/ebuild.sh" test └─ /bin/bash /usr/lib64/portage/bin/ebuild.sh test └─ /bin/bash /usr/lib64/portage/bin/ebuild.sh test └─ make -j5 -j1 check └─ /bin/sh -c fail= failcom='exit 1'; \ for f in x $MAKEFLAGS; do \ case $f in \ *=* | --[!k]*);; \ *k*) failcom='fail=yes';; \ esac; \ done; \ dot_seen=no; \ target=`echo check-recursive | sed s/-recursive//`; \ └─ /bin/sh -c fail= failcom='exit 1'; \ for f in x $MAKEFLAGS; do \ case $f in \ *=* | --[!k]*);; \ *k*) failcom='fail=yes';; \ esac; \ done; \ dot_seen=no; \ target=`echo check-recursive | sed s/-recursive//`; \ └─ make check └─ /bin/sh -c echo "Running test suite with ecl..."; \ /bin/sh ../maxima-local --lisp=ecl --batch-string="run_testsuite();" < /dev/null └─ /var/tmp/portage/sci-mathematics/maxima-5.26.0/work/maxima-5.26.0/src/binary-ecl/maxima -- --lisp=ecl --batch-string=run_testsuite(); └─ /var/tmp/portage/sci-mathematics/maxima-5.26.0/work/maxima-5.26.0/src/binary-ecl/maxima -- --lisp=ecl --batch-string=run_testsuite();
Created attachment 303193 [details] emerge --info
I see you use ecl. Have you tried some other lisps? ecl is very slow (only clisp is slower); sbcl or cmucl or gcl are much faster. I cannot reproduce your problem. On my ~x86 box maxima runs its testsuite with all 6 lisps without problems. One suspecious thing is hardened. In general, "hardened" and "lisp" always tend to contradict each other. Lisps tend to allocate some memory somewhere, to compile stuff on the fly into this memory, and to execute it immediately. This is a normal way of life for lisp. Usually, each small upgrade of the hardened stuff tends to break all lisps; later, lisps tend to recover somehow from this breakage; but by then, the hardened stuff change again, and all lisps are again completely broken. If possible, it would be much more reasonable to run lisp applications on non-hardened computers (probably, guarded from the hostile internet by hardened ones). Why mix these very different lifestyles on a single computer? Anyway, I cannot check this bug report, because I don't run hardened. I prefer to be able to have some work done on my laptop.
I retested this with 5.29.1 and it works just fine.
OK thanks. Of course, I run tests for each version of maxima for all 6 lisps (I have gcl unmasked) on a ~x86 box; recently I got a new laptop, ~amd64, and also run tests on it for all 5 lisps (no cmucl on amd64). There are some failures on 32-bit ccl, all other combinations run tests fine. I have not seen your problem.