The RT System itself (via RT) wrote:
# New Ticket Created by The RT System itself # Please include the string: [perl #19192]
# in the subject line of all future correspondence about this issue. # <URL: http://rt.perl.org/rt2/Ticket/Display.html?id=19192 >
I can build parrot fine on my FreeBSD box. I can re-run the tests fine with
/usr/local/bin/perl5.8.0 t/harness quick
All tests successful, 18 subtests skipped.
Files=42, Tests=581, 205 wallclock secs (65.07 cusr + 38.41 csys = 103.48 CPU)
However, if I try to test the JIT:
/usr/local/bin/perl5.8.0 t/harness -j quick
Failed Test Stat Wstat Total Fail Failed List of Failed
-------------------------------------------------------------------------------
t/op/trans.t 2 512 18 2 11.11% 11 18
t/pmc/perlstring.t 1 256 8 1 12.50% 3
t/pmc/pmc.t 1 256 80 1 1.25% 36
18 subtests skipped.
Failed 3/42 test scripts, 92.86% okay. 4/581 subtests failed, 99.31% okay.
Did I already say 'dubious'?
All failing tests are floatingpoint, all use the fp_eq macro (though a lot more tests use this macro and don't fail). But that's all. No common schme beyond above, why tests fail.
Could you set FLOAT_REGISTERS_TO_MAP to 0 near the end of jit/i386/jit_emit.h? (Maybe there are some FP issues, which the compiler can workaround but JIT isn't aware of)
Tracking this further down would need some extensive debug sessions on this machine, which would currently imply to prepare pasm's w/o macros and empty lines.
Nicholas Clark
leo