.... if I knew, what exactly to ask, it where easier. I'm implementing a parrot interpreter in immc. I seem to have memory corruptions or unreproducible errors. E.g. moving the stacktop variable only slightly gives different test results, as well as results on a second machine differ. Blocking DOD gives slightls worse results and Mops go down to 1/5.
One difference to test_main.c is, that my interpreter is a global. Is this allowed? Are there other issues related to this area? What could I try to do? BTW: $ time perl6 --test -r Failed Test Status Wstat Total Fail Failed List of failed ------------------------------------------------------------------------------- t/compiler/1.te 12 2 16,67% 11-12 t/compiler/2.te 5 3 60,00% 3-5 t/compiler/b.te 2 2 100,00% 1-2 t/rx/call.test 2 1 50,00% 1 t/rx/special.te 2 2 100,00% 1-2 Failed 5/15 test scripts, 66.67% okay. 10/65 subtests failed, 84.62% okay. real 0m28.892s user 0m26.570s sys 0m0.580s Though results vary (~60 - 85% okay). Execution time is approximately halfed compared to "perl6 --test" (800 MHz Athlon, i386-linux). Thanks for tips, leo
