Revision history for BATsh

0.07  2026-07-21 JST (Japan Standard Time)

    - SECURITY / SH REDIRECTIONS: fixed two defects in the pure-Perl SH
      interpreter's I/O redirection target handling (both present through
      0.06; the CMD interpreter was unaffected -- it already dequoted its
      redirection targets).

      (1) Quoting.  A quoted redirection target kept its quote characters
          and was split on whitespace inside the quotes, so

              echo hi > "out.txt"      wrote to a file named  "out.txt"
                                       (quotes included), not  out.txt
              echo hi > "a b.txt"      broke at the space

          Redirection targets are now read with quote/backslash awareness
          (new internal _read_redir_word) and dequoted with the same
          _arr_dequote() the command words already use, so a quoted space
          stays part of the one filename and the quotes are removed.

      (2) Command injection.  Redirection targets reached a 2-argument
          open() unprotected, so a target that ended in "|" was run as a
          pipe and one that began with ">" was re-parsed as a mode:

              f="date|"; cat < $f      ran `date` (a pipe), not open a file
              ... > ">name"            appended to "name" (leading > lost)

          Every plain-filename redirection open() (stdin/stdout/stderr in
          the exec-redirect path, the while-loop input redirect, subshell
          "( ... ) >file / <file", and the general _sh_exec_with_redirs
          path) now uses sysopen() with explicit O_RDONLY / O_WRONLY |
          O_CREAT | (O_APPEND|O_TRUNC) flags.  sysopen() takes the name
          literally, so a leading ">"/"<" or a trailing "|" is never
          treated as a mode or a pipe command.  The file-descriptor
          duplication forms (2>&1, 1>&2, >&STDERR) are unchanged.

          Fcntl (already a dependency) supplies the extra O_* constants
          (O_RDONLY, O_TRUNC, O_APPEND added to the existing import).

      New test t/0024-sh-redirect-quoting.t (7 checks: quoted output /
      input / append targets, a quoted target containing a space, and the
      two injection cases).  All Perl 5.005_03 compatible (index/substr
      scanning, sysopen with Fcntl O_* flags, no prototypes, no 3-arg
      open, no lexical filehandles).

    - SH SUBSHELL REDIRECTIONS: a trailing redirection on a subshell,
      "( ... ) > file" / ">> file" / "< file", extracted its target with a
      bare \S+ match, so a quoted target with a space broke and quotes were
      kept.  It now runs through the same quote-aware redirection parser as
      every other command, so "( ... ) > \"a b.txt\"" and "( ... ) > \$var"
      behave correctly; the target reaches the same sysopen() and so cannot
      be mis-parsed as a pipe/mode either.  (t/0024 RQ08, RQ09.)

    - SH ASSIGNMENT VALUES: a variable assignment dequoted its value by
      stripping only a single OUTERMOST "..." or '...' pair, which left
      concatenated quotes intact and split a backslash-escaped space:

          y=a'b'c     stored  a'b'c    (now abc)
          w=a\ b      ran "b" as a command   (now the value is "a b")

      Both the standalone form (VAR=value) and the prefix form
      (VAR=value command) now dequote with the same _arr_dequote() used for
      command words, and the prefix-assignment value reader honours "\ " so
      an escaped space no longer ends the value word.  New test
      t/0025-sh-assign-quoting.t (6 checks).  Perl 5.005_03 compatible.

    - SH PROCESS SUBSTITUTION path quoting: now that redirection targets
      (and, as before, command words) are dequoted, the temp-file path that
      <(cmd) / >(cmd) substitutes into the line is emitted quoted, so it is
      taken literally.  Without this, a Windows temp path
      (C:\...\Temp\batsh_ps_NN.tmp) lost its backslashes to backslash-escape
      processing, and a space in the temp directory split the word; on
      Unix (/tmp/...) the bug was invisible.  Also fixes process
      substitution used as a command argument (diff <(a) <(b)) when the temp
      directory contains spaces.

    - SH `cd` QUOTING: the cd builtin now dequotes its argument like every
      other command word, so cd "a b", cd 'dir', and cd "$VAR/x" reach
      chdir() as a plain path.  Previously the quote characters were kept
      and cd failed with "No such file or directory".  (t/0026 CD01-CD03.)

    - SH SUBSHELL REDIRECT FAILURE: a subshell whose redirection cannot be
      opened, e.g. "( read L ) < /no/such/file", now reports the error and
      returns a non-zero status WITHOUT running its body.  Previously it
      fell through and ran the body against the inherited stdin, which
      blocks forever when the body reads stdin (cat, read).  Any handle
      already redirected is restored before returning.  (t/0026 RF01; RF02
      guards that a working redirect still reads the file.)  New test
      t/0026-sh-cd-and-redir-fail.t (5 checks).  Perl 5.005_03 compatible.

    - NEW SH BUILTIN getopts: the POSIX single-character option parser,
      the last remaining builtin previously listed as unimplemented.

        getopts optstring name [arg ...]

      Called in a loop, it sets the variable named by "name" to each
      option letter in turn (and OPTARG to an option's argument), tracks
      progress through the argument list in OPTIND (1-based, reset to 1
      by the script before an independent parse), and returns non-zero
      when the options are exhausted so a "while getopts ...; do ...;
      done" loop ends by itself.  Supports: a ':' after a letter in the
      optstring to mark an argument-taking option; the separate
      (-o value) and attached (-ovalue) option-argument forms; clustered
      flags (-abc == -a -b -c); the "--" end-of-options marker (consumed)
      and a leading non-option word (stops parsing, left in place);
      parsing the positional parameters when no explicit args are given;
      and both error-reporting modes -- the default (a diagnostic to
      STDERR and name='?') and the silent mode selected by a leading ':'
      in the optstring (no message; name='?' for an unknown option or
      ':' for a missing argument, with the offending letter in OPTARG).
      The intra-argument character offset used for clustered options is
      internal state that is restarted whenever the script resets OPTIND,
      matching bash's behaviour for a fresh getopts loop; it is also
      reset between top-level runs so one script's half-finished loop
      cannot leak into the next.  All Perl 5.005_03 compatible
      (index/substr scanning, no prototypes, package-variable state).
      New test t/0023-sh-getopts.t (23 checks).

    - Fixed SH "shift N": the builtin accepted an optional count in its
      implementation (shift 1..9 positions) but the dispatcher called it
      with no argument, so "shift 3" and "shift $((OPTIND-1))" always
      shifted by exactly one -- silently corrupting any script that
      shifted more than one positional parameter at a time, including the
      idiomatic "shift $((OPTIND - 1))" that follows a getopts loop.  The
      dispatcher now passes the count through.  (t/0023: GO16/GO17.)

    - Fixed inline single-line function bodies that contain a control
      structure.  A body such as

        loop() { for x in a b c; do echo "$x"; done; }
        count() { i=0; while [ $i -lt 3 ]; do echo $i; i=$((i+1)); done; }
        chk()  { if [ "$1" = yes ]; then echo Y; else echo N; fi; }

      was split naively on ';' at definition time, tearing the loop or
      conditional into fragments ("while C" / "do B" / "done") that no
      longer reassembled into a runnable block -- the construct silently
      produced no output, or ran "done"/"fi" as an external command.
      _parse_function() now detects a control-structure keyword in
      command position in an inline body (new helper
      _inline_body_has_control(), quote/substitution aware) and, when
      present, keeps the whole body as one line so _run_lines() applies
      the same inline-control handling it already uses for a control
      structure typed directly on one physical line (including the
      "prefix; control" split, so "setup; while ...; do ...; done"
      works).  A trailing ';' before the closing brace is dropped so the
      inline "; done"/"; fi"/"; esac" terminator sits where the block
      parsers expect it.  Bodies of only simple commands
      (greet() { echo hi; echo bye; }) still split on ';' exactly as
      before.  All Perl 5.005_03 compatible.  (t/0023: GO14/GO15 exercise
      getopts inside such a function; the fix is general to all inline
      function bodies.)

    - Fixed the SH "$*" special parameter, which was never expanded and
      printed literally: the expansion pass substituted "$@" but not
      "$*", despite the code comment naming both.  "$*" now expands to
      the space-joined positional parameters, like "$@".  This is the
      parameter a script reads after "shift $((OPTIND - 1))" to see the
      operands left by a getopts loop.  (t/0023: GO19.)

    - Fixed the SH "echo" builtin's glob decision, which inspected the
      raw source text for *, ?, or [ with a plain pattern match.  That
      match also fired on the '*' of the "$*" parameter, on the '?' of
      "$?", and on any metacharacter inside quotes -- so
      echo "  x: $*" was needlessly run through word-splitting/globbing,
      collapsing its leading and internal whitespace, and a quoted
      echo "*.txt" risked being globbed against the current directory.
      A new quote- and parameter-aware scanner (_raw_has_glob) triggers
      globbing only on a genuine unquoted glob metacharacter that was
      actually written in the script, skipping quoted regions, $NAME /
      $* / $? parameter references, and ${...} expansions.  Unquoted
      "echo *.txt" still globs as before.  (t/0023: GO20/GO21.)

    - Extended shell features: extglob, brace expansion, here-strings,
      process substitution, select, alias/unalias, exec, and subshell
      command groups (v0.07).
        * shopt -s/-u extglob enables ?(),*(),+(),@(),!() pattern-list
          operators in case patterns and in the ${VAR%pat}-family
          parameter-expansion patterns (_extglob_scan/_extglob_split_alts,
          shared by _case_glob_to_re() and _glob_to_re()). Fixed two
          existing case-pattern parsers (_case_split_patterns and
          _case_parse_clause) to be paren-depth-aware so an extglob
          group's internal '|' and ')' are not mistaken for the
          pattern-separator '|' or the clause-terminator ')'.
        * {a,b,c} and {1..5}/{a..e}[..step] brace expansion runs
          lexically on the raw line (_brace_expand_line and friends),
          protecting quotes, ${...}, $(...), $((...)), `...`, and
          <(...)/>(...) as opaque regions; wired into _exec_line and
          _expand_word_list (for-loop lists).
        * cmd <<< word here-strings (_sh_strip_herestring), reusing the
          here-document temp-file machinery. Fixed a pre-existing
          _hd_detect() bug where "<<<" was only rejected as a
          here-document opener when matched starting at its FIRST '<';
          the scan then re-tried from the second '<' and matched "<<"
          there, silently mistreating "cmd <<< word" as a broken
          here-document. _hd_detect() now skips all three characters.
        * <(cmd) and >(cmd) process substitution
          (_replace_process_subst/_procsub_capture/_procsub_tempfile).
          <(cmd) is captured eagerly into a kept temp file, like $(cmd)
          without the final read-back; >(cmd) defers cmd until after the
          current simple command finishes. _exec_line was split into
          _exec_line_impl plus a thin _exec_line wrapper that drains
          per-nesting-level process-substitution bookkeeping. Fixed
          _split_sh_compound() and _split_sh_pipe(), which both split a
          line on ';'/'&&'/'||'/'|' BEFORE expansion and only tracked
          $(...) nesting: a ';' or '|' inside a <(...)/>(...) body (e.g.
          ">(read LINE; echo $LINE)") was mistaken for the OUTER line's
          separator, corrupting the command. Both now also track <(/>(
          nesting the same way they already tracked $(.
        * select VAR in LIST; do ... done (_parse_select): numbered
          menu to STDERR, $PS3 prompt, REPLY/VAR from STDIN.
        * alias/unalias builtins (_cmd_alias/_cmd_unalias) and
          first-word alias expansion (_alias_expand_line) at the top of
          _exec_line, with a chain-length guard against self-reference.
        * exec (_cmd_exec): "exec > file" (etc.) applies redirections
          permanently to the current shell; "exec cmd" runs cmd and then
          ends the script with cmd's status, approximating process
          replacement (this interpreter never forks).
        * ( cmd1; cmd2 ) subshell command groups (_parse_subshell):
          variable, array, function, and alias changes, and cd, are
          isolated via snapshot/restore of BATsh::Env, %_SH_ARRAY,
          %_SH_ARRAY_TYPE, %_SH_FUNCTIONS, %_SH_ALIAS, and Cwd::cwd()
          around the body; a trailing >, >>, or < on the closing ")" is
          honoured.
      All Perl 5.005_03 compatible (2-argument open()/opendir() with
      bareword filehandles throughout; sysopen(O_CREAT|O_EXCL) for every
      new temp-file class, matching the existing here-document /
      command-substitution / pipeline-stage helpers). New test
      t/0021-sh-extended-features.t (24 tests); all previously existing
      tests continue to pass unmodified.

    - Security hardening: command-substitution capture files and SH
      pipeline stage files are now created with the same symlink-race-
      resistant sysopen(O_CREAT|O_EXCL, 0600) discipline already used by
      _bg_tempfile() (background-job pidfiles) and _hd_tempfile()
      (here-documents). Previously _cmd_subst() ($( ... ) / `...` output
      capture) and _exec_sh_pipe() (per-stage stdout->stdin bridge files)
      opened a predictable path in the shared temp directory with a plain
      open(FH, "> $path"), which (a) relied on umask for the file mode
      (typically world-readable, 0644) and (b) followed a pre-existing
      symlink at that path instead of refusing to open it. Both code
      paths are exercised by nearly every non-trivial script (any $(...),
      backtick, or | pipeline), so the exposure was broad on shared
      multi-user hosts. New helpers BATsh::SH::_subst_tempfile() and
      BATsh::SH::_shp_tempfile() mirror the existing _bg_tempfile() /
      _hd_tempfile() retry-on-EEXIST pattern (sequence-numbered names,
      1000 attempts, 0600 mode); the file is opened once by the helper
      and reused (no separate create-then-reopen step), removing the
      brief window in the old code between creation and use. Stray files
      from an abnormal exit are still tracked for END-time failsafe
      cleanup, as for the existing temp-file classes. No externally
      visible behaviour change; all 890 existing tests pass unmodified.
      Perl 5.005_03 compatible (same sysopen/Fcntl usage as the existing
      helpers; no new dependencies). New test t/0019-tempfile-security.t.

    - Tilde expansion ~/path, ~user/path (v0.07). POSIX word-initial,
      unquoted tilde expansion is now performed on the RAW source line
      inside BATsh::SH::_expand() -- via a new _tilde_prepass() pass
      that runs BEFORE $VAR / $(...) / `...` substitution -- so that a
      literal ~ typed in the script is expanded exactly once, while a
      $VAR that merely happens to hold a string starting with "~" is
      never re-expanded (matching bash: only the source word is
      eligible, not values produced by substitution). Covers cd, word-
      split command arguments (echo, eval, external commands), test/[
      file-test operands, and the right-hand side of NAME=value
      assignments (both the pure "NAME=value" form and the prefix
      "NAME=value command" form, including the extra tilde-eligible
      position right after the assignment '=', as in VAR=~/sub).
      ~user resolves via getpwnam and is a documented no-op on Win32,
      where the word is left literal (same as bash's behaviour for an
      unresolvable login name). NOT implemented: tilde expansion after
      ':' in colon-list assignments (PATH=~/a:~/b); documented as a
      limitation. New test t/0020-tilde-expansion.t (11 tests).

    - Script exit-code propagation and $? <-> %ERRORLEVEL% bridge.
      run() / run_string() / run_lines() now RETURN the script's final
      exit status: "exit 3" gives 3, "EXIT /B 5" gives 5, otherwise the
      status of the last command.  The modulino exits with that code
      (exit(BATsh->main(@ARGV))), so OS-level %ERRORLEVEL% / $? of
      "perl lib/BATsh.pm script.batsh" is the script's own status.
      At every CMD<->SH section boundary the status is mirrored both
      ways, so a failing SH command is visible as %ERRORLEVEL% in the
      next CMD section and vice versa.  New public accessor
      BATsh->last_status.  New sub main() handles --help, --version,
      -e 'source', and "-" (read script from STDIN, remaining arguments
      become %1../$1..).  EXIT with no code keeps the current
      ERRORLEVEL.  "exit N" now also ends the REPL.  NEW FILE
      bin/batsh.pl (EXE_FILES); installed as "batsh".

    - $(( )) arithmetic rewritten as a recursive-descent parser.
      Now supported: comparison and logical operators (result 0/1),
      assignment operators = += -= *= /= %= <<= >>= &= ^= |= (written
      back to the variable store), prefix/postfix ++ and --, the
      ternary ?: and comma operators, ** (right-associative), bitwise
      & ^ | ~ << >> (~ is signed: -v-1), hex 0x.. and octal 0..
      literals, and C-style truncation toward zero for / and %
      (-7/2 = -3, -7%2 = -1).  Errors warn and yield 0.  Fixed a
      here-document false positive on $((1<<4)) ("<<" inside $(( ))
      is no longer taken as a heredoc opener).

    - SH shell options set -e / -u / -x (and set -o errexit /
      nounset / xtrace; +e/+u/+x to turn off; options combinable as
      in "set -eux").  set -e stops the script with the failing
      command's status; if/while/until conditions and non-final
      members of && / || lists are exempt.  set -u makes expansion of
      an unset variable an error (warning + exit status 1;
      ${VAR:-default} is exempt).  set -x traces each simple command
      to STDERR with a "+ " prefix.  Options are reset at the start
      of each top-level run so "set -e" cannot leak into a later
      run() in the same process.  Limitations are documented in POD:
      -x traces the raw pre-expansion line (expanding a copy would
      run $(...) twice), and -u stops after the current command
      completes with the empty expansion.

    - NEW SH BUILTIN eval: one level of quote removal, concatenation,
      and re-execution with a second round of expansion (POSIX
      semantics).  Previously "eval" fell through to external-command
      execution and only worked by accident where /bin/sh handled it.

    - New tests: t/0016-exit-status.t, t/0017-sh-arith.t,
      t/0018-sh-set-options.t.

    - Fixed t/0016-exit-status.t (ES12-ES14) on real Windows: the
      child-process helpers used a single quoted shell string for
      system(), which cmd.exe re-wraps in an extra pair of double
      quotes and mis-parses ("filename, directory name, or volume
      label syntax is incorrect"). Rewritten to use LIST-form
      system() (bypasses the shell entirely) with STDIN/STDOUT
      redirected at the Perl level via dup/reopen instead of a
      shell "<"/">" string. lib/BATsh.pm itself was unaffected; this
      was a test-harness-only bug. Re-verified on Linux (16/16).

      Desk review of the Win32-specific code paths (no Windows host
      available in this environment): BATsh::SH::_bg_launch_decoded()
      uses system(1, $cmdline) (P_NOWAIT) to background a job on
      Win32; the success/failure check (defined $pid && $pid > 0)
      matches the documented Win32 system(1, ...) contract (PID on
      success, 0 on failure). t/0016's _run_child() LIST-form
      system($^X, "-I$lib", $pm, $prog, @args) calls CreateProcess
      directly and does not re-enter cmd.exe, so the original
      quote-re-wrapping failure mode cannot recur. _bg_tempfile()'s
      temp-directory selection already accounts for %TEMP%/%TMP% and
      backslash path separators. No logic defects found in review;
      confirmation on an actual Windows host (locale, path-length,
      %TEMP% permission variables that cannot be reviewed statically)
      remains pending.

    - CP932 (Shift_JIS) multibyte-safe execution: NEW MODULE BATsh::MB.
      A .batsh script written in CP932 -- the encoding of Japanese
      Windows -- now runs correctly even when its characters contain
      trail bytes that collide with ASCII shell metacharacters (the
      classic "dame-moji" / 0x5C problem). Affected characters are
      extremely common: SO (0x835C), HYOU (0x955C), NOH (0x945C) end
      in a backslash; PO (0x837C) in a pipe; CHI (0x8360) in a
      backtick; DA (0x835E) in the cmd.exe caret escape; and many
      trail bytes fall in a-z/A-Z where uc()/lc() corrupt them.

      Design: instead of teaching every byte-oriented scanner in
      BATsh::CMD / BATsh::SH about lead and trail bytes, the script
      text passes through a reversible GUARD TRANSFORM on input. Each
      two-byte character LEAD+TRAIL whose TRAIL is in the dangerous
      ASCII range 0x40-0x7E is rewritten to the three-byte form
      \x01 LEAD (TRAIL+0x80), which contains no ASCII bytes; a literal
      \x01 becomes \x01\x01, making the transform bijective. All
      existing parsing (caret escapes, pipelines, quotes, redirects,
      globs, case patterns, uc/lc variable handling, SETLOCAL
      snapshots, ...) is then automatically DBCS-safe with no scanner
      changes. The inverse transform is applied at the output
      boundaries only:

        print sinks   ECHO/echo/printf/SET display, prompts, _warn
        exec sinks    system() for external commands, background jobs,
                      FOR /F ('command') input pipes
        file sinks    redirect targets, CD/DIR/COPY/DEL/MOVE/MKDIR/
                      RMDIR/REN/TYPE paths, IF EXIST, test -e/-f/-d...,
                      glob patterns, here-document bodies, CALL/source
                      script filenames
        env sink      BATsh::Env::sync_to_env (raw bytes into %ENV)

      and input re-entry points guard incoming raw bytes again:
      SET /P and read line input, FOR /F file/command lines, $(...)
      and `...` captured output, glob results, Cwd for %CD%/PWD.

    - Encoding selection (BATsh::MB): 'auto' is the default -- a
      non-UTF-8 source containing bytes >= 0x80 is detected as CP932
      and the guard switches on (and then stays on for the process,
      because Env may hold guarded values; only an explicit
      set_encoding deactivates it). Pure-ASCII and well-formed UTF-8
      sources need no guarding and are byte-for-byte unaffected, so
      existing behaviour is preserved exactly. Explicit selection:

        BATsh->run($file, encoding => 'cp932');
        BATsh->run_string($src, encoding => 'sjis');
        BATsh->set_encoding('cp932');
        set BATSH_ENCODING=cp932        (environment variable)
        perl lib/BATsh.pm --encoding=cp932 script.batsh

      Supported: cp932/sjis, gbk/cp936, uhc/cp949, big5/cp950 (all
      sharing the same trail-byte hazard), utf8, none, auto. A UTF-8
      BOM on the first line is stripped.

    - Character semantics for substring/length operators under the
      guard: ${#VAR}, ${VAR:N:L}, ${VAR:N} (SH) and %VAR:~n,m% (CMD)
      now count CP932 CHARACTERS, not bytes, via BATsh::MB::mb_length
      / mb_substr. Byte semantics are unchanged while the guard is
      inactive (ASCII/UTF-8 operation identical to 0.06).

    - Modulino command line (perl lib/BATsh.pm ...): new options
      --encoding=ENC and --version; script arguments after the script
      name are now passed through as %1..%9 / %* (previously they
      were dropped).

    - BATsh::Env::sync_to_env no longer copies the batch-parameter
      pseudo keys (%0..%9, %*, %%V FOR variables) into %ENV; they are
      not legal environment variable names and polluted the child
      environment of external commands.

    - t/0015-cp932.t: new regression suite (20 checks, US-ASCII
      source with \xNN escapes) covering echo/ECHO of dame-moji,
      %VAR% and Env-bridge round-trips, pipeline/backtick/caret
      false-positive protection, uc() safety, character-based
      ${#VAR} / ${VAR:N:L} / %VAR:~n,m%, case patterns, redirects,
      CP932 filenames (test -f, IF EXIST), FOR /F over a CP932 file,
      ${VAR^^}, IF string comparison, %ENV export, auto-detection,
      UTF-8 pass-through, and enc/dec bijectivity.

    - eg/13_cp932_demo.pl: runnable demonstration (US-ASCII source)
      of a CP932 mixed CMD/SH script.

    - t/9010-encoding.t: BATsh::MB added to the US-ASCII source
      check list.

    - Fixed a pre-existing echo bug: the SH "echo" builtin decided
      whether to apply filename globbing by testing the already
      variable-expanded $rest for * ? [ characters. This meant a
      variable whose *value* happened to be a glob metacharacter
      (e.g. getopts setting $opt to "?" on an unknown option, or ":"
      on a missing argument) could be silently glob-matched against
      the current directory and replaced by whatever single-character
      filename happened to match, instead of being echoed literally.
      The glob-or-not decision is now made from the raw, pre-expansion
      source text instead, so only a glob metacharacter actually
      written in the script (e.g. "echo *.txt") triggers globbing;
      a variable's expanded value is never re-subject to globbing,
      matching the same "expansion results are not re-expanded"
      principle already used for tilde expansion. (t/0021:
      EF25/EF26 regression tests.)

    - Practical-level fixes to the pure-Perl SH and CMD interpreters
      (bugs found by exercising real scripts; all covered by the new
      t/0022-sh-compound-fixes.t regression suite, 16 checks):
        * A simple-command prefix followed on the SAME physical line by
          a control structure introduced with ';' now dispatches the
          control structure through its block parser instead of handing
          the pieces to /bin/sh -- e.g. x=""; if [ -z "$x" ]; then ...;
          fi, i=0; while ...; done, v=cat; case ...; esac
          (_run_lines/_find_control_split).
        * Inline "if COND; then A; else B; fi" (and elif) now honours
          the else/elif branch; the old single-line parser dropped it
          (_parse_if via _inline_has_terminator/_inline_expand).
        * Nested "if ... fi" is depth-aware so an inner fi no longer
          closes the outer if, both multi-line and inline
          (_parse_if body collector + _if_depth_delta).
        * Escaped double quotes inside a double-quoted word are handled
          per POSIX: echo "she said \"hi\"" -> she said "hi"
          (_arr_dequote rewritten with backslash escaping).
        * A for-loop / array list built from $(...) no longer leaks a
          stray ")" token: for f in $(echo a b c); do ...; done, and
          arr=($(echo x y z)) (_arr_split_words made substitution-aware).
        * A trailing "# comment" is stripped from SH command lines when
          the '#' begins a word and is unquoted, while $#, ${#var},
          ${var#pat}, an in-word '#' (a#b, http://h#frag) and quoted
          '#' are left intact, and here-document bodies keep their '#'
          (_strip_sh_comment, applied per physical line in _run_lines).
        * A control structure used as a pipeline element or && / ||
          operand now runs correctly: cmd | while read x; do ...; done,
          cmd | for i in ...; do ...; done, true && for ...; done.
          _split_sh_compound became control-structure-grouping aware so
          a ';' / && / || inside a while/for/if/case/until/select block
          is no longer treated as a top-level separator, and a single
          compound command reaching _exec_line_impl as a pipe/operand is
          routed through the block runner (_seg_is_control). (The
          multi-line form with "do" on its own line after a pipe remains
          a known limitation; the inline "; do" form is supported.)
        * Single-line CMD "IF cond (body) ELSE (body)" (and IF EXIST /
          IF DEFINED / negated forms) now parses both branches on one
          physical line instead of echoing the raw ") ELSE (" text
          (_parse_if_bodies via _match_paren).

0.06  2026-06-27 JST (Japan Standard Time)

    - SH indexed and associative arrays (BATsh::SH). The single most useful
      missing piece of bash-script compatibility is now implemented:

        arr=(a b c)            indexed array literal
        arr+=(d e)             append at the next index
        arr[i]=v, arr[i]+=v    element assignment / string append
        declare -a arr         declare an (empty) indexed array
        declare -A map         declare an associative array
        typeset -a / -A        accepted as an alias of declare
        map[key]=value         associative element assignment
        map=([k1]=v1 [k2]=v2)  associative (or sparse indexed) literal
        ${arr[i]}, ${map[key]} element access; $arr is short for ${arr[0]}
        ${arr[-1]}             negative index counts back from the last element
        ${arr[@]}, ${arr[*]}   all element values
        ${#arr[@]}             number of set elements
        ${#arr[i]}             length of one element's value
        ${!arr[@]}             list of indices (indexed) or keys (associative)
        unset arr, unset arr[i] remove whole array / single element

      Indexed subscripts are evaluated arithmetically (so ${arr[$i]} and
      ${arr[i+1]} work) and indexed arrays may be sparse; associative
      subscripts are literal strings. Array names are case-insensitive, like
      scalar variables, and a name is either a scalar or an array, never both.
      Element order for ${arr[@]} is ascending numeric index for indexed
      arrays and sorted key order for associative arrays -- bash leaves the
      associative order unspecified, so a deterministic order is used for
      portable output. In a for list, "${arr[@]}" and "${!arr[@]}" word-split
      to one item per element or key, so elements containing spaces survive.
      Array operations are detected on the raw line before expansion so the
      "(a b c)" literal and the "[sub]" subscripts are never mangled by
      variable or command substitution. All Perl 5.005_03 compatible (no
      prototypes; storage via plain package-level hashes).

    - SH for-loop word list: the list of a "for VAR in LIST" header is now
      variable- and command-substitution-expanded (previously the words were
      taken literally), with quote-aware word splitting and filename globbing.
      This fixes "for x in $LIST" and enables "for x in "${arr[@]}"".

    - SH echo: quote removal is now structural (per word) rather than only
      stripping one pair of surrounding quotes, so e.g.
      echo "${arr[@]}" tail no longer prints the literal double quotes.

    - t/0010-sh-arrays.t: new regression suite (26 checks) covering indexed
      and associative arrays, append, sparse arrays, negative indexing,
      arithmetic and variable subscripts, ${#arr[@]} / ${#arr[i]} / ${!arr[@]},
      unset of whole arrays and single elements, declare/typeset -a/-A,
      inline associative initialisers, and for-loop iteration over both
      "${arr[@]}" and "${!arr[@]}".

    - lib/BATsh.pm, lib/BATsh/SH.pm: POD updated to document array support and
      to move arrays out of the "not implemented" list.

    - CMD CALL :label argument passing and subroutine call frame
      (BATsh.pm, BATsh::CMD). CALL :label arg1 arg2 ... now installs the
      subroutine's own positional parameters -- %0 is the :label token,
      %1..%9 are the call arguments, and %* is their join -- and the caller's
      %0..%9 / %* frame is saved on entry and restored on return, so a
      subroutine no longer clobbers the caller's parameters and a shorter call
      no longer inherits the caller's stale arguments. Arguments are
      %-expanded before the call and split with double-quote awareness, so
      CALL :sub "a b" %FILE% passes "a b" as a single argument. The two CALL
      code paths (the BATsh-level directive interceptor and BATsh::CMD's
      _cmd_call) now both delegate to BATsh->call_sub, which performs the
      save/install/restore, so they behave identically. The same arguments
      are mirrored to BATSH_ARG1.. so an SH-mode subroutine body sees them as
      $1..$9 / $@. Previously CALL set BATSH_ARG* but not %1..%9 (so %1 and
      %~dp1 inside the subroutine saw the caller's arguments), set neither %0
      nor %* for the callee, and never restored the caller's frame.

    - CMD %~N path modifiers on CALL arguments: because %1..%9 are now
      populated by CALL, the tilde modifiers %~f1 %~d1 %~p1 %~n1 %~x1 (and
      combinations such as %~dp1, %~nx1) operate on a subroutine's passed
      path arguments, not just on %0.

    - CMD SHIFT / SHIFT /N is now an actual dispatched builtin (BATsh::CMD).
      It was documented but unimplemented, so SHIFT inside a subroutine fell
      through to an external command ("Can't exec SHIFT"). SHIFT now moves
      %2 into %1, %3 into %2, ..., clears %9, rebuilds %*, and keeps the
      BATSH_ARG* mirror in step; SHIFT /N begins the shift at %N.

    - t/0011-cmd-call-args.t: new regression suite (14 checks) for CALL
      argument passing (%1..%9, %*, %0 = label), caller-frame save/restore,
      quoted and %-expanded arguments, %~nx1 / %~n1 / %~x1 / %~dp1 on passed
      arguments, SHIFT and SHIFT /N, nested CALL frames, SH-mode subroutine
      bodies seeing $1..$9, and non-inheritance of unused parameters.

    - lib/BATsh.pm, lib/BATsh/CMD.pm: POD documents the CALL argument frame,
      %~N modifiers on passed arguments, and SHIFT / SHIFT /N.

    - CMD subroutine-internal GOTO labels (BATsh.pm). A subroutine body may
      now contain its own :labels as GOTO targets -- loops, forward skips,
      or an early GOTO :EOF return -- so idioms such as a SHIFT loop that
      sums a variable-length argument list work inside a CALL'd subroutine.
      _extract_subroutines() previously used only a "label ... RET" heuristic
      that mis-split such a subroutine: an internal label appearing before
      the RET demoted the real entry label, so CALL :SUB reported "undefined
      subroutine" and only the last internal label was registered. The
      extractor now (a) treats any label named by a "CALL :LABEL" anywhere in
      the script as a subroutine entry point (unioned with the existing RET
      heuristic for backward compatibility), and (b) opens a subroutine only
      at the top level: once inside a body, every :label is an internal label
      that travels with the body and only RET/RETURN closes it. Internal
      labels are resolved by the CMD interpreter when the body runs.

    - CALL is no longer intercepted in _exec_cmd_section; it is handled by the
      CMD interpreter's _cmd_call (which already delegates to call_sub /
      source_file). The old interception flushed and split the CMD batch at
      each CALL, so a GOTO whose loop body contained a CALL landed in a
      different exec_block than its label and failed with "label not found".
      Keeping CALL inside the batch lets each CMD section -- and each
      subroutine body -- run as one block with a complete label index, so
      GOTO across a CALL now resolves (in main code and in subroutines
      alike). The now-unused BATsh::_split_call_args helper was removed.

    - t/0012-cmd-sub-labels.t: new regression suite (7 checks) for internal
      GOTO loops with SHIFT, subroutine reuse, internal forward GOTO skips,
      coexistence of top-level GOTO labels with internal-label subroutines,
      multiple subroutines each with their own internal labels, early
      GOTO :EOF return from a subroutine, and nested CALL where both the
      caller and callee subroutines use internal labels.

    - eg/09_cmd_subroutines.batsh: the SHIFT demonstration now uses the
      idiomatic internal GOTO loop (:SUMARGS_LOOP / :SUMARGS_DONE) to sum a
      variable-length argument list.

    - SH case..esac pattern branching (BATsh::SH). _parse_case() and
      _match_pattern() were rewritten to parse clauses structurally rather
      than line by line, adding:
        * the bash fall-through terminators ;& (run the next clause's body
          unconditionally) and ;;& (keep testing the remaining patterns),
          alongside the normal ;;;
        * character-class patterns [abc], ranges [a-z], and negation
          [!abc] / [^abc], in addition to the existing * and ? globs;
        * quoted and backslash-escaped patterns that match literally (e.g.
          "*") matches a literal asterisk);
        * a fully-inline construct on one physical line
          (case $x in a) echo a ;; *) echo b ;; esac);
        * an optional leading "(" before the pattern list ((pattern)).
      Multiple |-separated patterns and the default *) clause already worked
      and remain supported; clause and pattern splitting is now quote- and
      [class]-aware so a | or ) inside quotes or a [...] class is not a
      delimiter.  The closing esac is recognised only at a clause boundary,
      so the word "esac" appearing inside a body no longer ends the construct
      prematurely.  break / continue / return / exit inside a clause body
      stop the case and propagate as before.  All Perl 5.005_03 compatible
      (hand-rolled scanners; no regex features beyond \A \z and character
      classes).

    - t/0013-sh-case.t: new regression suite (17 checks) for |-patterns, the
      *) default, glob and [class] patterns including negation and ranges,
      quoted/literal patterns, ;& and ;;& fall-through, fully-inline and
      multi-line forms, leading-paren clauses, empty bodies, first-match
      semantics, and esac-inside-a-body.

    - SH trap / signal handling (BATsh::SH, BATsh.pm): a minimal trap built-in
      bridged to Perl's %SIG.

        trap 'commands' SIGSPEC...   register a handler
        trap - SIGSPEC...            reset to the default action
        trap '' SIGSPEC...           ignore the signal
        trap        / trap -p        list the current traps

      A SIGSPEC is a signal name with or without a leading SIG (INT, SIGINT),
      a number (2), or the EXIT pseudo-signal (also 0).  Real signals are
      bridged to %SIG: trap 'cmd' INT installs a %SIG{INT} handler that runs
      cmd, trap '' INT sets IGNORE, and trap - INT restores DEFAULT.  The EXIT
      trap runs internally when the script finishes or when exit is called
      (exactly once), wired in via BATsh.pm's run / run_string / run_lines and
      BATsh::SH::fire_exit_trap(), plus _cmd_exit().  The handler command is
      captured on the raw line (a new interceptor in _exec_line, before
      expansion) and expanded only when it fires, so trap 'rm -f $tmp' EXIT
      removes the file named by $tmp as it stood at exit.  EXIT/ERR/DEBUG/
      RETURN are treated as pseudo-signals and never touch %SIG (only EXIT
      runs a handler today); %SIG assignment is eval-guarded so signal names
      unsupported by the host (common on Windows) degrade quietly.  trap was
      previously listed as unimplemented.  Assigning a handler for a signal
      the platform lacks (e.g. HUP/USR1/USR2 on Windows) no longer prints a
      "No such signal" warning: _sh_set_os_sig() filters just that warning
      while the best-effort assignment still succeeds.

    - t/0014-sh-trap.t: new regression suite (14 checks) for the EXIT trap on
      normal end, on exit, with deferred expansion, cancellation, and
      single-fire; %SIG handler install / IGNORE / DEFAULT / clearing;
      handler execution and real signal delivery (kill); trap listing;
      multiple signals per command; SIG-prefix and numeric normalization.

    - eg/12_cmd_vs_sh.batsh: new teaching example that writes the same nine
      tasks twice -- once in cmd.exe (CMD) style and once in bash/sh (SH)
      style -- as a side-by-side parallel-translation reference for students:
      printing, variables, arithmetic, an if/else string test, a counted
      loop, a list loop, a subroutine/function, a multi-way branch (CMD
      IF-chain vs SH case), and the shared variable bridge between the two
      modes.  A quick CMD-vs-SH syntax table heads the file; output lines are
      labelled CMD:/SH: so the two notations line up.

    - doc/ cheat sheets (all 21 languages): added three new sections covering
      the 0.06 SH features -- "15. SH Arrays", "16. SH case Statements", and
      "17. SH trap and Signals" -- with the same language-independent code
      examples in every file and consecutive section numbering preserved
      (t/9080 still passes).  Inline comments and prose are fully localized
      for EN and JA, and additionally for ZH, KO, FR, ID, VI and TR; the
      remaining 12 languages (BM BN HI KM MN MY NE SI TL TW UR UZ) carry the
      translated section headings with the comment text left in English as a
      draft pending native-speaker review.

0.05  2026-06-14 JST (Japan Standard Time)

    - SH single-line loops: "for VAR in LIST; do BODY; done" and
      "while/until COND; do BODY; done" written entirely on one physical
      line are now executed correctly. Previously _parse_for() captured the
      whole inline tail (BODY; done) into the iteration list and collected
      no body, so the loop produced no output and -- because the body
      collector then consumed the rest of the script looking for a "done"
      that never arrived as its own line -- it re-ran the remainder of the
      file once per whitespace token in the list; the single-line while
      form died with a shell "do unexpected" / "Can't exec done" error.
      _parse_for() and _parse_while() now detect an inline ";do" (matched
      with \b after "do" so a "; done" terminator is never taken for the
      "do" keyword), split the header from the inline body, and recognise
      an inline "; done" terminator (greedy, so the LAST "; done" closes the
      loop and a body that legitimately contains the word "done" or a nested
      single-line loop is handled). The inline while form also honours a
      redirect on the closing done, e.g. "while read L; do ...; done < FILE".
      Multi-line loops are byte-for-byte unaffected. All forms are Perl
      5.005_03 compatible (no regex features beyond \A \z \b and /s). This
      makes the single-line "for i in 1 2 3; do ... done" shown in the
      doc/*.txt cheat-sheet MIXED-MODE SAMPLE run as documented.

    - t/0002-sh-interpreter.t: added regression tests for the inline
      single-line for/while/until forms, nested single-line loops, a body
      that contains the literal word "done", inline filename globbing, and
      an inline while-read with a redirect on the closing done.

    - Section splitter: BATsh::_sh_depth_delta() now computes the NET block
      depth change for an ENTIRE SH line instead of inspecting only the first
      token. A single-line "for ...; do ...; done" (or while/until/case)
      counted the leading "for" (+1) but never the inline "done" (-1), so the
      SH section was left open; the FOLLOWING CMD line was then absorbed into
      the SH section and executed by the SH interpreter, where "%VAR%" is not
      expanded -- e.g. the cheat-sheet sample's "ECHO Back in CMD: %result%"
      printed "%result%" literally instead of bridging the SH-side value. The
      new scanner counts opener (if for while until case function select, "{")
      and closer (fi done esac, "}") keywords only in command position and
      skips quoted text, $( ) substitutions and backticks, so an inline loop
      nets to zero and the section boundary is detected correctly. Pure
      Perl 5.005_03 (substr-based scan, no regex features). Multi-line input
      is unaffected (the fast path returns the previous single-token result).

    - Section splitter: a CMD-style comment (":: ..." or "REM ...") that
      follows an SH section is no longer handed to the SH interpreter
      verbatim. _process_lines() routed every comment line into the current
      section unchanged; because the SH interpreter treats only "#" as a
      comment, a ":: ..."/"REM ..." line was dispatched to the external shell,
      and any "( )", "|" or other metacharacter in the comment then made
      /bin/sh fail with 'Syntax error: "(" unexpected'. Comment (and empty)
      lines are now routed into a section as a BLANK line, which both
      interpreters skip identically at every nesting depth and which also
      avoids the real cmd.exe quirk of "::" inside a "( )" block.

    - t/0004-bridge.t: added regression tests proving (a) a single-line
      for/while does not leave the SH section open so a following CMD
      "%VAR%" still bridges, and (b) a ":: ..."/"REM ..." comment containing
      shell metacharacters after an SH section is not dispatched to the
      shell (STDERR is captured and checked for no shell syntax error).

    - CMD variable substring/substitution modifiers: %VAR:~n,m% and
      %VAR:str1=str2% in-place substitution are now supported in
      BATsh::Env::expand_cmd() via a new _expand_var_modifier() helper.
      Substring forms: %VAR:~n% (from offset), %VAR:~n,m% (n+m chars),
      %VAR:~-n% (last n chars), %VAR:~n,-m% (all but last m chars).
      Substitution forms: %VAR:str1=str2% (replace first, case-insensitive),
      %VAR:*str1=str2% (replace from start through first str1).
      All forms are processed before the plain %VAR% pass and are
      Perl 5.005_03 compatible (substr, index, no regex features).

    - CMD dynamic pseudo-variables: %DATE%, %TIME%, %CD%, %RANDOM%,
      %ERRORLEVEL%, and %CMDCMDLINE% are now resolved at expansion time
      by a new _expand_named_var() dispatcher inside expand_cmd().
      %DATE% returns YYYY-MM-DD, %TIME% returns HH:MM:SS.cc, %CD% is the
      current working directory, %RANDOM% is a pseudo-random integer 0-32767,
      %ERRORLEVEL% reads the current value from BATsh::CMD via the new
      public accessor BATsh::CMD::_get_errorlevel(), and %CMDCMDLINE%
      returns an empty string (not meaningful in pure-Perl mode).

    - SH filename globbing: unquoted words that contain *, ?, or [...]
      metacharacters are now expanded to matching pathnames using Perl's
      built-in glob(). Expansion occurs in BATsh::SH::_parse_args() (for
      echo and external-command arguments) and in BATsh::SH::_parse_for()
      (for the word list of for VAR in GLOB; do ... done). Single-quoted
      and double-quoted patterns are NOT expanded (POSIX behaviour).
      If no file matches the pattern, the literal pattern is returned
      unchanged (nullglob-off behaviour, which is the shell default).
      Two new helpers are added to BATsh::SH: _glob_expand() (expand one
      word) and _glob_expand_args() (convenience wrapper over a list).

    - t/0009-new-vars.t: new test file, 25 tests (NV01-NV25) covering
      all three feature areas above. Added to MANIFEST.

    - POD updated: BATsh.pm BUGS AND LIMITATIONS now records that
      %VAR:~n,m% / %VAR:str1=str2% and all dynamic pseudo-variables are
      supported; the SH filename-globbing limitation item is removed.
      BATsh::Env Variable Expansion section documents all new forms.
      BATsh::SH Supported Features table lists glob expansion.

    - Version bumped to 0.05 in lib/BATsh.pm, lib/BATsh/CMD.pm,
      lib/BATsh/SH.pm, lib/BATsh/Env.pm, Makefile.PL, META.yml,
      META.json, and README.

0.04  2026-06-07 JST (Japan Standard Time)

    - Inline-Perl portability: every shelled-out Perl one-liner in the
      distribution is rewritten into a form that satisfies BOTH external
      shells that BATsh dispatches to -- cmd.exe (system STRING on Win32)
      and /bin/sh -c (system STRING on Unix/BSD). The previous forms hit
      two distinct, OS-specific foot-guns:
        (A) Unix: a dollar token inside the code (e.g. "$_") was expanded
            by /bin/sh from the environment variable "_" (the last-arg /
            path that the shell exports), which is unpredictable on CPAN
            smokers and produced random failures such as
            "Bareword found where operator expected ... 1EERDtQcrK".
        (B) Win32: cmd.exe does not honour single quotes, so a one-liner
            wrapped in '...' was split on whitespace and Perl died with
            "Can't find string terminator".
      The portable form uses DOUBLE quotes and no shell-expandable dollar
      token, e.g.  perl -e "..."  ->  perl -ne "print uc"  (the default
      variable is consumed implicitly by uc, so nothing leaks to the
      shell). Rewritten in: lib/BATsh.pm POD, lib/BATsh/SH.pm POD, README,
      all 21 doc/batsh_cheatsheet.*.txt files, eg/05_cmd_comprehensive.batsh,
      eg/06_sh_comprehensive.batsh, and t/0006-new-features.t. The stderr
      sample in eg/06_sh_comprehensive.batsh is likewise switched from a
      single-quoted body to a double-quoted "print STDERR qq(...)" form.

    - t/0007-extcmd-env.t: new regression test that locks in the
      portability fix above. EE01/EE02 run the pipeline and here-document
      patterns under several hostile values of the environment variable
      "_" and confirm correct uppercase output (this passes on every OS
      and actively defeats vector (A) on Unix). EE03 is the clean-
      environment baseline. EE04 is a static guard against vector (B):
      no inline "perl -e/-ne/-pe" anywhere in t/ eg/ doc/ lib/ README may
      be wrapped in single quotes. EE05 is a static guard against vector
      (A): no double-quoted inline Perl may contain a shell-expandable
      dollar token ($name, $_, ${...}, $1..$9); the harmless numeric $$
      is exempt. Both static guards run on any OS, so a Windows run still
      catches a Unix-introduced regression and vice versa. Added to
      MANIFEST.

    - External-Perl PATH portability (vector C): the test suite and the
      eg/ examples shell out to a bareword "perl", but a CPAN smoker
      frequently does NOT have the perl under test on PATH as "perl"
      (perlbrew/plenv, or perl invoked by absolute path). The bareword
      then resolves to nothing ("perl: not found", empty output), which
      failed t/0007 EE01/EE02 and t/0006 NF23/NF60 and -- worse -- let
      t/0006 NF07/NF21/NF22 report a corrupted, empty-named "ok" when the
      failed pipe disturbed the captured-STDOUT save/restore; in eg/06 it
      also hung (an empty "perl" command substitution fed a
      "while read ... < $EMPTY" redirect whose read fell back to terminal
      STDIN and blocked). Fixed WITHOUT touching the command strings or
      the examples (a bareword "perl" is the correct thing for an end
      user to type, and embedding an absolute $^X path would expose a
      Win32 backslash path to SH-mode quote/escape processing): each
      affected test now prepends the directory of the running interpreter
      ($^X) to PATH so the bareword "perl" resolves to the very perl now
      running the suite. The prepend is installed before the first
      BATsh::Env::init() (init() snapshots %ENV into STORE and
      sync_to_env() copies STORE back to %ENV before each external
      command). Touched: t/0006-new-features.t, t/0007-extcmd-env.t,
      t/9070-examples.t (the last for the eg/ child process). Verified on
      Linux with perl deliberately removed from the child PATH (and under
      a hostile "_"): all 606 tests pass, no failures, no hangs; the
      command strings and all eg/*.batsh examples are byte-for-byte
      unchanged.

    - t/0006-new-features.t: the END block now sets "$? = 1 if $fail"
      instead of calling "exit 1", matching the INA_CPAN_Check.pm END-block
      convention adopted in 0.03 (an END block must not call exit, so that
      the harness sees the real plan/ok reconciliation).

    - Documentation: the "self-contained" qualifier is removed from
      lib/BATsh.pm (header comment, module description, the run() banner,
      and the NAME / DESCRIPTION POD) and from README. BATsh dispatches
      external commands to a real shell, so describing the interpreter
      itself as "self-contained" was misleading; "bilingual shell
      interpreter written in pure Perl" is retained and accurate.

    - SH nested command substitution fixed (lib/BATsh/SH.pm). $( ... )
      has been advertised as supporting full nesting since 0.02, but a
      nested $( ... $( ... ) ) -- especially with a pipeline at each
      level -- collapsed to an empty string, and on Unix a nested
      pipeline could hang; the failure mode also differed between Windows
      and Unix. Three independent defects were responsible:
        (1) _cmd_subst() named its stdout-capture temp file with the
            process id alone (batsh_cap_$$.tmp). An inner $(...) reused
            the same path and unlink()'d it, so the outer level captured
            nothing. The capture file is now tagged with the active
            substitution-nesting depth.
        (2) _split_sh_pipe() counted the "(" of a "$(" twice, leaving the
            $( nesting depth stuck at 1 after a nested $(...); a bare "|"
            that followed it was then not recognised as a pipe. "$(" now
            consumes both characters and bumps the depth exactly once.
        (3) _exec_sh_pipe() named its per-stage temp files with the
            process id alone (batsh_shp_$$) and left its dup STDOUT/STDIN
            globs un-local()ised. A nested pipeline therefore clobbered
            the outer pipeline's stage file and saved handles; the outer's
            final segment found no input file and blocked on the real
            STDIN (a hang on Unix). The stage files are now tagged with
            the active pipeline-nesting depth and the handle globs are
            local()ised. All fixes are Perl 5.005_03 compatible (use vars
            package globals, bareword filehandles, 2-argument open). The
            command strings and the externally-visible API are unchanged.

    - t/0008-nested-subst.t: new regression test for the fix above. NS01/
      NS02 prove the capture-file depth fix with pure builtins (no "perl"
      on PATH required); NS03 is the single-level pipeline-in-$() baseline;
      NS04/NS05 cover nested $() with a pipeline at each level (defects 2
      and 3); NS06 checks that two sibling $() pipelines on one line do
      not collide; NS07 covers an assignment from a nested pipeline
      substitution and its reuse. Like t/0006/0007 it prepends the running
      interpreter's directory to PATH so the bareword "perl" resolves on a
      smoker. Added to MANIFEST.

    - eg/05_cmd_comprehensive.batsh: the SET/IF demonstration variable was
      renamed LANG -> GREETING. As LANG, the example exported LANG=BATsh
      into %ENV, and the external "perl" it later spawns then emitted a
      glibc "Setting locale failed ... LANG = BATsh" warning to STDERR on
      Unix (harmless, and invisible during "make test" because
      t/9070-examples.t captures and discards child STDERR, but visible
      when the example is run by hand; Windows perl does not warn). The
      rename keeps the example's behaviour identical and silences the
      Unix-only noise.

    - eg/00_hello.pl: normalised from a CRLF line ending to LF, matching
      the rest of eg/ (the other examples are already LF). Perl tolerates
      the trailing CR on Unix, so this is a cosmetic consistency fix.

    - Version bumped to 0.04 in lib/BATsh.pm, lib/BATsh/CMD.pm,
      lib/BATsh/SH.pm, lib/BATsh/Env.pm, Makefile.PL, META.yml and
      META.json. BATsh::CMD and BATsh::Env carry no changes other than
      the version; BATsh::SH changes are the version, the two POD
      one-liner rewrites noted above, and the nested command-substitution
      fix noted above.

0.03  2026-06-06 JST (Japan Standard Time)

    - t/lib/INA_CPAN_Check.pm: emit exactly one TAP plan line per test
      file. Each check_* helper previously called plan_tests() itself,
      while the .t files also called plan_tests(count_*); this produced
      multiple "1..N" lines in a single file. Under a real TAP harness
      (prove / Test::Harness, as used by CPAN Testers) this raised
      "More than one plan found in TAP output" and made the affected
      files FAIL, even though every individual "ok" line passed when the
      scripts were run by hand. Affected files: t/9010-encoding.t,
      t/9030-distribution.t, t/9040-style.t. The plan_tests() call is
      now removed from every check_* helper, leaving the .t file as the
      sole owner of the plan line.
    - t/lib/INA_CPAN_Check.pm: count_A() now returns the actual number
      of MANIFEST entries instead of a fixed 1, so that the plan
      computed by t/9030-distribution.t matches the number of A1 checks
      that check_A() emits.
    - t/lib/INA_CPAN_Check.pm: remove the mid-stream plan_skip() calls
      from check_A() and check_C(); the MANIFEST-absent guard is handled
      by the .t file and by count_C() before any plan line is printed.
    - t/lib/INA_CPAN_Check.pm: check_K() now honours the k3_exempt
      option passed by t/9040-style.t. The argument was previously
      discarded by "my ($root) = @_;", so the intended exemption of
      accessor-style hash names (%env, %opts, %args) never took effect;
      the K3 detector also did not capture the hash name. check_K() now
      accepts "k3_exempt => REGEX", captures the returned hash name via
      /return \%(\w*)/, and skips a "return \%name" only when the name
      matches the supplied pattern. Behaviour is unchanged when
      k3_exempt is not passed (every "return \%..." is still flagged),
      so distributions that call check_K($root) without the option are
      unaffected.
    - t/lib/INA_CPAN_Check.pm: add a regression guard for the two TAP
      defect classes above. plan_tests() now refuses to emit a second
      "1..N" line, and an end-of-run reconciliation reports
      "planned X but ran Y" (setting a non-zero exit) when the emitted
      plan does not match the number of ok()/not-ok() lines. Both
      problems now FAIL immediately on a plain "perl t/foo.t", not only
      under a real harness.
    - t/lib/INA_CPAN_Check.pm: add selfcheck_suite(), which runs t/*.t
      (and xt/*.t) in a child Perl and verifies one plan line per file,
      plan == number of ok/not-ok lines, and no failures.
    - pmake.bat: at "pmake dist" time, after the existing source checks,
      run INA_CPAN_Check::selfcheck_suite() as check3 and abort the
      build if any test file fails the plan-sanity check (disable with
      --no-check3). Bump $PMAKE_BAT_VERSION to 0.34.
    - t/lib/INA_CPAN_Check.pm: pass \@files / \@pm_files (a reference)
      instead of [ @files ] (an anonymous copy) to _find_pm_t() in
      _scan_code(), check_D(), check_E(), and check_K(). The copy form
      meant the collected file list never reached the caller, so E1
      (no shebang in lib/*.pm) and K3 (return { %hash } form) silently
      scanned zero files and always passed.

    - Documentation: BATsh.pm BUGS AND LIMITATIONS corrected. It no longer
      claims SH-mode background execution is unsupported (it is supported
      for external commands; see above and BATsh::SH), and it now clarifies
      that non-builtin commands (FINDSTR, SORT, etc.) are invoked as
      external programs rather than "unsupported". README and BATsh.pm POD
      additionally enumerate previously undocumented limitations: CMD
      "%VAR:~n,m%" / "%VAR:str1=str2%" and dynamic "%RANDOM%/%DATE%/%TIME%/
      %CD%" variables; SH arrays, filename globbing, "~" tilde expansion,
      brace expansion, and the trap/getopts/select/alias/declare/eval/exec
      builtins and set -e/-u/-x options; and the shared (no sub-shell)
      "( ... )" grouping common to both modes.

    - SH expansion: a backslash-escaped "\$", "\`" or "\\" inside double
      quotes is now preserved literally and no longer triggers variable
      or command substitution (e.g. "\$_" yields a literal "$_").

    - SH read: the "read" built-in now returns a non-zero status at end of
      input so that "while read VAR; do ...; done < FILE" terminates
      instead of looping. Leading option flags such as "-r" are skipped
      and are no longer treated as target variable names.

    - SH assignment prefix: "VAR=value command args" (POSIX) now applies
      the assignment and then runs the command (e.g. "IFS= read -r LINE",
      "LC_ALL=C sort"); multiple prefixes are supported. A standalone
      assignment whose value merely contains spaces or a "$(...)"
      substitution (e.g. UPPER=$(echo "a b")) keeps the full value and is
      no longer mistaken for a prefix.

    - SH while/until: an input redirection on the "done" line
      ("while read L; do ...; done < FILE") now reopens STDIN from FILE for
      the duration of the loop so the loop's "read" consumes the file.

    - eg/06_sh_comprehensive.batsh: I/O-redirection section simplified to a
      plain "while read" loop now that the loop terminates correctly.

    - Tests: t/9070-examples.t now executes each eg/*.batsh in a child
      process and guards against runaway output and "syntax error"
      breakage (E4). t/9060-readme.t verifies the README advertises every
      eg/ example by name (R5). README gains an EXAMPLES section.

    - SH background execution: an unquoted trailing "&" starts an external
      command asynchronously and returns immediately. On Win32 the job is
      spawned via system(1, ...) (P_NOWAIT, PID returned); on Unix it is
      started through /bin/sh without a Perl fork, capturing the job PID
      via the shell's $! into a sysopen O_CREAT|O_EXCL temp file (Pure
      Perl, 5.005_03). The new $! parameter expands to the most recent
      background PID (empty before any job); $? is 0 on a successful
      launch (the job's own exit status is not awaited). Built-ins,
      functions, assignments and control words ignore the trailing "&"
      and run in the foreground; "&&", ">&"/"2>&1", quoted and escaped
      "\&" are not treated as background. No job control; CMD-mode "&"
      remains a sequential separator (see BUGS AND LIMITATIONS).

    - eg/05_cmd_comprehensive.batsh: the "IF ERRORLEVEL" diagnostic line
      "ECHO   ERRORLEVEL>=0: ELTEST=%ELTEST%" contained a bare ">", which
      CMD mode correctly treats as output redirection (matching cmd.exe).
      As written, the message was silently redirected to a file named
      "=0:" instead of being printed, and that stray file was created in
      the current directory each time the example ran (including under
      "make test" via t/9070-examples.t). The ">" is now caret-escaped
      ("ECHO   ERRORLEVEL ^>= 0: ELTEST=%ELTEST%"), so the line prints
      as intended and no file is written.

0.02  2026-04-28 JST (Japan Standard Time)

    [Highlights]
    - Full bash/sh interpreter implementation: if/for/while/until/case,
      function definitions (name() { ... }), local variable scoping,
      && / || / ; compound commands, pipelines (|), I/O redirection
      (> >> < 2> 2>> 2>&1), variable expansion (${var%pat}, ${var#pat},
      ${#var}, ${var^^}, ${var,,}, ${var:N:L}, ${var/p/r}, ${var//p/r}),
      positional parameters $1..$9 / $@ / $* / $#, shift, read, source.
    - cmd.exe pipeline (|) support via temporary file (Pure Perl, 5.005_03).
    - I/O redirection: stdout overwrite (>), append (>>), stdin (<),
      stderr (2>), stderr-to-stdout (2>&1), stdout-to-stderr (1>&2).
      Supported in both CMD mode and SH mode.
    - SH here-documents on STDIN: cmd <<DELIM ... DELIM, <<-DELIM (strip
      leading tabs), and <<'DELIM' (literal, no expansion). Body is
      materialised to a temp file created with sysopen O_CREAT|O_EXCL
      (Pure Perl, 5.005_03) and fed through the existing "< file" path,
      so both built-ins (read) and external commands see it on STDIN.
      Top-level mode dispatch is here-document aware, so uppercase body
      lines are not misrouted to CMD mode. Single here-document per line;
      here-strings (<<<) and same-line pipeline/compound combos are not
      supported (see BUGS AND LIMITATIONS).
    - cmd.exe batch-parameter tilde modifiers: %~0, %~f1, %~dp0, %~nx1,
      %~n0, %~x0, %~p1 etc. (f d p n x modifiers, combinable).
    - SET /P VAR=Prompt  interactive prompt input from STDIN.
    - $0 normalised to absolute path via File::Spec on run().

    [BATsh::Env]
    - Variable names are now stored and looked up in uppercase, matching
      cmd.exe's case-insensitive environment variable behaviour.
      SET myvar=x  followed by  ECHO %MYVAR%  now correctly outputs "x".
    - Added $DELAYED_EXPANSION package variable (default 0).
    - setlocal() now accepts an options string and parses
      ENABLEDELAYEDEXPANSION / DISABLEDELAYEDEXPANSION.
      The delayed-expansion flag is saved/restored with the variable store.
    - expand_cmd() now expands !VAR! references when $DELAYED_EXPANSION is on.

    [BATsh::CMD]
    - Implemented ^ escape character:
        ^X          -> literal X (protects & | < > etc.)
        ^^          -> literal ^
        trailing ^  -> line continuation (joins next line)
    - Implemented I/O redirection parsed before command dispatch:
        >file       stdout overwrite
        >>file      stdout append
        2>file      stderr overwrite
        2>>file     stderr append
        <file       stdin redirect
      Redirects with ^> are correctly treated as escaped > (not a redirect).
      fd-digit stripping limited to isolated '1' or '2' before '>' to avoid
      consuming trailing digits of command arguments (e.g. "ECHO line1 >f").
    - SETLOCAL now passes its option string to BATsh::Env::setlocal() so
      ENABLEDELAYEDEXPANSION and DISABLEDELAYEDEXPANSION take effect.
    - !VAR! delayed expansion: pre_expanded block bodies now call expand_cmd()
      at runtime when delayed expansion is active, so SET inside an IF/FOR
      block followed by ECHO !VAR! correctly reflects the updated value.
    - IF block pre-expansion: %VAR% in parenthesised IF/ELSE bodies is now
      expanded at parse time (matching cmd.exe semantics), so a SET inside the
      block does not affect %VAR% references in the same block.
    - FOR block pre-expansion: %VAR% in parenthesised FOR bodies is expanded
      once before the first iteration (at FOR-line parse time) and cached;
      the loop variable is substituted per-iteration via an internal placeholder.
    - IF /I (case-insensitive comparison) is now parsed before plain == so
      that "/I" is not consumed as part of the left-hand operand.
    - IF EXIST now handles quoted paths that contain spaces.
    - ECHO no longer resets ERRORLEVEL to 0 after printing.
    - FOR /F fully implemented:
        tokens=N,M-P  select specific token columns
        tokens=N*     select token N and put the remainder in the next variable
        delims=CHARS  field delimiters (default space/tab)
        skip=N        skip the first N lines of the source
        eol=C         skip lines beginning with character C (default ;)
        usebackq      swap quoting: "file" reads a file, 'cmd' runs a command
      Sources: bare filename, quoted filename, 'command' (backtick output),
      and ("literal string").
    - & (sequential), && (conditional-success), || (conditional-failure)
      compound commands are now supported.
    - SET VAR=value: variable name regex relaxed to accept any non-'=' prefix,
      matching cmd.exe's permissive variable naming.

    [BATsh::SH]
    - Full bash/sh interpreter implemented as Pure Perl (no external shell).
    - Control structures: if/then/elif/else/fi, for/do/done, while/do/done,
      until/do/done, case/esac with glob-pattern matching.
    - Function definitions: name() { ... } and function name { ... } syntax,
      including inline single-line bodies.  Functions receive positional
      arguments $1..$9; caller's arguments are saved and restored on return.
    - local variable scoping: local VAR=value saves the caller's value and
      restores it when the function returns.
    - Compound commands: cmd1 && cmd2, cmd1 || cmd2, cmd1 ; cmd2.
    - Pipeline: cmd1 | cmd2 [| cmd3 ...] implemented via temporary files
      (Perl 5.005_03 compatible bareword filehandles, no fork/exec).
    - I/O redirection: > >> < 2> 2>> 2>&1 1>&2, parsed after variable
      expansion so that filenames may contain variables.
    - Variable expansion: $VAR, ${VAR}, $1..$9, $@, $*, $#, $?, $$, $0.
      Parameter expansion forms: ${VAR:-default}, ${VAR:=default},
      ${VAR:+alt}, ${VAR%pat}, ${VAR%%pat}, ${VAR#pat}, ${VAR##pat},
      ${VAR/pat/rep}, ${VAR//pat/rep}, ${VAR^^}, ${VAR^}, ${VAR,,}, ${VAR,},
      ${VAR:N:L}, ${VAR:N}, ${#VAR}.
      Glob patterns in %/%%/#/## support *, ?, and [abc] character classes.
    - Arithmetic expansion: $(( expr )) with +, -, *, /, %, and positional
      parameters $1..$9 inside the expression.
    - Command substitution: $( cmd ) with full nesting/quoting support, and
      backtick `cmd` form.
    - shift [N]: shifts positional parameters left by N positions (default 1),
      updating both %N and %* in BATsh::Env.
    - read VAR: reads one line from STDIN, chomps it, stores in VAR.
    - source / . file: executes an external file in the current SH context.
    - Builtin commands: echo, printf, cd, pwd, exit, true, false, :, export,
      unset, set (noop), test / [ ... ] with -f -d -e -r -w -x -s -z -n
      and string (= == != < >) and integer (-eq -ne -lt -le -gt -ge) ops.
    - _cmd_subst uses fixed bareword filehandles (_SUBST_SAVOUT etc.) to
      avoid collision under Perl strict and recursive invocations.
    - _sh_strip_redirects and _sh_exec_with_redirs added for I/O redirection.
    - _replace_cmd_subst added: walks $( ) depth-tracking the nesting so that
      $( cmd | perl -e "print uc" ) parses the closing ) correctly.
    - _split_sh_compound and _exec_sh_compound added for && / || / ; handling.
    - _split_sh_pipe and _exec_sh_pipe added for pipeline handling.
    - $0 is set to the absolute path of the running script by BATsh::run().

    [BATsh.pm]
    - _exec_cmd_section no longer intercepts SETLOCAL/ENDLOCAL itself;
      both are passed through to BATsh::CMD::_dispatch so the option string
      (ENABLEDELAYEDEXPANSION etc.) is correctly forwarded.

0.01  2026-04-26 JST (Japan Standard Time)

    - Initial CPAN release.
