benchmark_tests FAIL from r42170 - Failed tests: 18-20

[mikehh]

I got a failure in fulltest at r42173 in benchmark_tests

Test Summary Report
-------------------
t/benchmark/benchmarks.t (Wstat: 768 Tests: 34 Failed: 3)
  Failed tests:  18-20
  Non-zero exit status: 3
Files=1, Tests=34, 20 wallclock secs ( 0.04 usr  0.00 sys + 17.53 cusr  1.46 csys = 19.03 CPU)
Result: FAIL
make[1]: *** [benchmark_tests] Error 1

the test had passed at r42164

bisecting the test passed at r42169 but failed at r42170:

mhb@mhb-desktop:~/t.parrot$ date
Fri Oct 30 22:33:53 GMT 2009
mhb@mhb-desktop:~/t.parrot$ prove -v t/benchmark/benchmarks.t
t/benchmark/benchmarks.t ..
1..34
ok 1 - examples/benchmarks/addit.pasm
ok 2 - examples/benchmarks/addit.pir
ok 3 - examples/benchmarks/addit2.pir
ok 4 - examples/benchmarks/array_access.pir
ok 5 - examples/benchmarks/arriter.pir
ok 6 - examples/benchmarks/arriter_o1.pir
ok 7 - examples/benchmarks/bench_newp.pasm
ok 8 - examples/benchmarks/fib.pir
ok 9 - examples/benchmarks/freeze.pasm
ok 10 - examples/benchmarks/gc_alloc_new.pasm
ok 11 - examples/benchmarks/gc_alloc_reuse.pasm
ok 12 - examples/benchmarks/gc_generations.pasm
ok 13 - examples/benchmarks/gc_header_new.pasm
ok 14 - examples/benchmarks/gc_header_reuse.pasm
ok 15 - examples/benchmarks/gc_waves_headers.pasm
ok 16 - examples/benchmarks/gc_waves_sizeable_data.pasm
ok 17 - examples/benchmarks/gc_waves_sizeable_headers.pasm
not ok 18 - examples/benchmarks/oo1.pasm

#   Failed test 'examples/benchmarks/oo1.pasm'
#   at t/benchmark/benchmarks.t line 218.
# Exited with error code: 1
# Received:
# Null PMC access in get_string()
# current instr.: '_Foo@@@(null)' pc 31 (/home/mhb/t.parrot/t/benchmark/benchmarks_18.pasm:56)
#
# Expected:
# 10
#
not ok 19 - examples/benchmarks/oo2.pasm

#   Failed test 'examples/benchmarks/oo2.pasm'
#   at t/benchmark/benchmarks.t line 218.
# Exited with error code: 1
# Received:
# Null PMC access in get_string()
# current instr.: '_Foo@@@(null)' pc 31 (/home/mhb/t.parrot/t/benchmark/benchmarks_19.pasm:43)
#
# Expected:
# 10
#
not ok 20 - examples/benchmarks/oo3.pasm

#   Failed test 'examples/benchmarks/oo3.pasm'
#   at t/benchmark/benchmarks.t line 218.
# Exited with error code: 1
# Received:
# Null PMC access in type()
# current instr.: '_Foo@@@(null)' pc 25 (/home/mhb/t.parrot/t/benchmark/benchmarks_20.pasm:17)
#
# Expected:
# 10
#
ok 21 - examples/benchmarks/oo4.pasm
ok 22 - examples/benchmarks/oo5.pir
ok 23 - examples/benchmarks/oo6.pir
ok 24 - examples/benchmarks/oofib.pir
ok 25 - examples/benchmarks/overload.pir
ok 26 - examples/benchmarks/primes.pasm
ok 27 - examples/benchmarks/primes2.pir
ok 28 - examples/benchmarks/primes2_i.pir
ok 29 - examples/benchmarks/primes_i.pasm
ok 30 - examples/benchmarks/stress.pasm
ok 31 - examples/benchmarks/stress1.pasm
ok 32 - examples/benchmarks/stress2.pasm
ok 33 - examples/benchmarks/stress3.pasm
ok 34 - examples/benchmarks/vpm.pir
# Looks like you failed 3 tests of 34.
Dubious, test returned 3 (wstat 768, 0x300)
Failed 3/34 subtests

Test Summary Report
-------------------
t/benchmark/benchmarks.t (Wstat: 768 Tests: 34 Failed: 3)
  Failed tests:  18-20
  Non-zero exit status: 3
Files=1, Tests=34, 24 wallclock secs ( 0.03 usr  0.00 sys + 16.36 cusr  1.70 csys = 18.09 CPU)
Result: FAIL
mhb@mhb-desktop:~/t.parrot$ cat .parrot_current_rev
42170

r42170 - chromatic: [PMC] Removed implicit double-underscore prefix logic to override vtable entries when adding Subs to NameSpace PMC. If PASM still needs a fix, let's do it the right way and add the :vtable attribute to pcc_sub declarations.

on irc (2009-Oct-30 [UTC]):

[23:30] <chromatic> examples/benchmarks/oo1.pasm fails because I removed the double-underscore implicit vtable overriding behavior.
[23:30] <chromatic> Either we expand PASM's parsing in IMCC to allow attributes on PASM subs or we rewrite that benchmark into PIR.
[23:31] <mikehh> I think the latter option is the better one
[23:32] <chromatic> Fortunately, rewriting is easy.
[23:32] <Coke> I think pasm needs a way to override vtables, no?
[23:33] <chromatic> I'd add it to IMCC for PASM, but why in the world would you write raw PASM these days?

[bacek]

Benchmark tests were rewritten in PIR in r42185, r42186 and r42187. make benchmark_tests now passing again. Closing ticket.

-- Bacek