[tannevaled]

Three units, +283 measured, and the first wave whose numbers are trustworthy throughout: #660 repaired the harness at the end of wave 31, so this is the first measurement in the campaign with no known vacuous assertions beneath it.

PR	area	gain
#666	core/kernel, numerics, symbols	+162, core/rational and core/builtin_constants at the ceiling
#668	IO/Dir encodings resolved once at open, #syswrite reaching the descriptor, IO::Buffer's base/size distinction	+56, and 13 of 16 files touched now sit exactly at the ceiling
#665	singleton classes of nil/true/false, post parameters in Method#arity and #parameters, !~/!/!= dispatched

45 examples were invisible, not failing

core/integer/pow_spec.rb and exponent_spec.rb each scored 0 — never finishing. The corpus contains 100000000 ** 1000000000, which rbgo computed unbounded until the harness killed the process. bignum.c's BIGLEN_LIMIT refuses it. Same shape as the shim story: a file reporting 0 needs its ceiling measured before that 0 is read as failures.

One cause, three defects

#663 reported that !~ was compiled to !(a =~ b) and never dispatched. Asking whether its neighbours shared the shape found ! and != doing the same — all three definitions were dead code on main. Any object redefining them (a null object, a matcher, a DSL) lost the definition silently.

Measurement

scripts/conformance/rubyspec/run.sh, pinned corpus 87b1631, 16-way, TIMEOUT=25, three consecutive runs:

passing examples: 22489   (files loaded: 2194, failed to load: 12)
passing examples: 22489
passing examples: 22488
frozen floor:     22205

Floor set thirteen below the lowest.

Three briefing targets dissolved under measurement — the third wave running

• core/io/write_spec.rb's "9 failures": a real gap of 1. MRI fails the other eight through this shim.
• core/array/fill_spec.rb's "9": a gap of zero. Ceiling 38, rbgo 38.
• core/numeric/step_spec.rb's 12.700000000000001: what MRI prints.

Running MRI through our own shim for a per-file ceiling is what catches these, and it is now done before planning rather than after.

The largest lever left is in none of these lanes

core/file/printf_spec.rb scores 0 against a ceiling of 160: rbgo cannot parse a do…end block inside an arrow-lambda that is a paren-less command argument.

foo :a, -> { [1].each do |x| x end }   # rbgo: parse error. MRI: fine.
foo(:a, -> { … })                      # parses
foo :a, lambda { … }                   # parses

One parser fix, 160 examples — issue #669.

🤖 Generated with Claude Code