test: ask what case conversion answers where it follows ASCII - #7223

build_config/ci/gcc-clang.rb:65-70 says what the ascii-case build is there for:

  # The one build here that indexes by character and converts case by ASCII.
  # Both halves of that pair are what it covers: core's ASCII conversion, and
  # the refusal mruby-regexp answers a pattern with when /i is asked for a
  # folding the build has no table for.

The second half holds, and holds nowhere else. mrbgems/mruby-regexp/test/ascii_case.rb:12
opens with skip unless __ENCODING__ == "UTF-8", so the byte-string build reaches
neither of its blocks and this build is their only home.

The first half asserts nothing. Every assertion that would show the narrowing is gated
off rather than mirrored:

So the conversion is held to "nothing raised" rather than to an answer. Counted against
the build beside it, ascii-case reports a Total of 2332 where bintest reports 2336,
and none of the four blocks that went missing came back as a mirror. Nothing in any
build configuration pins this:

"Ä".downcase       #=> "Ä" where case follows ASCII, "ä" where it follows Unicode
"ä".upcase         #=> "ä"
"ä".casecmp?("Ä")  #=> false

Fix

Gate the mirrors the other way. UNICODECASE is false both under MRB_USE_ASCII_CASE
and on a build reading its strings as bytes, so one addition covers both configurations,
and each block sits beside the Unicode block it answers.

Core holds String#downcase, #upcase and #capitalize, and asks for the bytes of a
run that spells no character handed back, the ArgumentError such a run draws elsewhere
belonging to the walk over characters. mruby-string-ext holds String#swapcase and
#casecmp?. mruby-encoding asks what the bytes of "Ä" read as UTF-8 answer, which is
what the same bytes read as bytes answer where there is nothing above ASCII to map.

Verification

rake -m test, the mrbtest counts, on e5517c30 and on this branch:

KO and Crash are 0 in every column. The three builds that convert case by Unicode
gain the two blocks in mruby-string-ext as Skip; the three that convert by ASCII, which
default.rb is one of, gain three blocks that run. The bintests are unchanged, 106 OK
in default.rb and the same in the bintest build.

Thirteen of the twenty one new assertions answer differently on a build converting by
Unicode, which is the narrowing they are there to show. The other eight pin that the
ASCII half still converts, and answer alike either way.

Environment

# ci/gcc-clang, full-debug
gcc -std=gnu99 -g -O3 -Wall -Wundef -Werror-implicit-function-declaration -Wwrite-strings -g3 -O0 -DMRB_GC_STRESS -DMRB_USE_DEBUG_HOOK -DMRB_DEBUG -DMRB_USE_BIGINT -DMRB_USE_COMPLEX -DHAVE_MRUBY_ENCODING_GEM -DMRB_UTF8_STRING -DHAVE_MRUBY_IO_GEM -DMRB_USE_RATIONAL -DHAVE_MRUBY_REGEXP_GEM -DMRB_USE_SET -DMRB_USE_TASK_SCHEDULER src/string.c

# ci/gcc-clang, bintest
gcc -std=gnu99 -g -O3 -Wall -Wundef -Werror-implicit-function-declaration -Wwrite-strings -DMRB_GC_FIXED_ARENA -DMRB_USE_BIGINT -DMRB_USE_COMPLEX -DHAVE_MRUBY_ENCODING_GEM -DMRB_UTF8_STRING -DHAVE_MRUBY_IO_GEM -DMRB_USE_RATIONAL -DHAVE_MRUBY_REGEXP_GEM -DMRB_USE_SET -DMRB_USE_TASK_SCHEDULER -DMRB_USE_DEBUG_HOOK src/string.c

# ci/gcc-clang, cxx_abi
gcc -g -O3 -Wall -Wundef -Wwrite-strings -x c++ -std=gnu++03 -DMRB_GC_FIXED_ARENA -DMRB_USE_CXX_EXCEPTION -DMRB_USE_CXX_ABI -DMRB_USE_BIGINT -DMRB_USE_COMPLEX -DHAVE_MRUBY_ENCODING_GEM -DMRB_UTF8_STRING -DHAVE_MRUBY_IO_GEM -DMRB_USE_RATIONAL -DHAVE_MRUBY_REGEXP_GEM -DMRB_USE_SET -DMRB_USE_TASK_SCHEDULER src/string.c

# ci/gcc-clang, byte-string
gcc -std=gnu99 -g -O3 -Wall -Wundef -Werror-implicit-function-declaration -Wwrite-strings -DMRB_USE_BIGINT -DMRB_USE_COMPLEX -DHAVE_MRUBY_IO_GEM -DMRB_USE_RATIONAL -DHAVE_MRUBY_REGEXP_GEM -DMRB_USE_SET -DMRB_USE_TASK_SCHEDULER src/string.c

# ci/gcc-clang, ascii-case
gcc -std=gnu99 -g -O3 -Wall -Wundef -Werror-implicit-function-declaration -Wwrite-strings -DMRB_USE_ASCII_CASE -DMRB_USE_BIGINT -DMRB_USE_COMPLEX -DHAVE_MRUBY_ENCODING_GEM -DMRB_UTF8_STRING -DHAVE_MRUBY_IO_GEM -DMRB_USE_RATIONAL -DHAVE_MRUBY_REGEXP_GEM -DMRB_USE_SET -DMRB_USE_TASK_SCHEDULER src/string.c

# build_config/default.rb
gcc -std=gnu99 -g -O3 -Wall -Wundef -Werror-implicit-function-declaration -Wwrite-strings -DHAVE_MRUBY_REGEXP_GEM -DMRB_USE_SET -DHAVE_MRUBY_IO_GEM -DMRB_USE_RATIONAL -DMRB_USE_COMPLEX -DMRB_USE_BIGINT -DMRB_USE_DEBUG_HOOK src/string.c

🤖 Generated with Claude Code

https://claude.ai/code/session_01PiFpCR4tWrc5iWH8DRfNeQ

Summary by CodeRabbit

• Tests
  • Added coverage for string case conversion when Unicode support is unavailable.
  • Verified ASCII characters continue to convert correctly while non-ASCII and malformed byte sequences are handled consistently.
  • Added tests for swapcase, casecmp?, and bang methods, including unchanged-result behavior.