[brettheap]

What changed

A v1.2.1 follow-up from lane 3's independent reviews (#1216, #1217). It lands under Brett Heap's v1.2.1 merge word (2026-10-23T13:10:26Z) once reviewed and green.

Three tables padded a person's text with a format width, which counts code points, not the columns a terminal gives them:

• templates/family-root/scripts/siblings.py's layout table (make siblings) and the dispatch's (b) what ran (make park, make resume), both through Sibling.row's :<18 (templates/family-root/scripts/siblings.py: the layout table pads each member's name by code points, not by the columns a terminal gives it, so a wide or combining name's state column stands out of line (the family-root twin of #1034) #1216);
• scripts/validate-repository-naming.py's plain listing, :<32 (scripts/validate-repository-naming.py: the plain listing pads each name from the command line or stdin by code points, not by the columns a terminal gives it, so a wide or combining name's label column stands out of line (twin of #1034) #1217);
• the same script's --explain family rows, :<18 over a policy's family id (_no_family_lines and _family_row, main's :795 and :919, which scripts/validate-repository-naming.py: the plain listing pads each name from the command line or stdin by code points, not by the columns a terminal gives it, so a wide or combining name's label column stands out of line (twin of #1034) #1217's Direction asked to be read before they were left).

At main, a name of the three wide characters U+65E5 U+672C U+8A9E stood 3 columns right of every other row, and e U+0301 cole (a combining acute) 1 column left. On an ASCII stdout, which writes each as its escapes, they stood 15 and 5 columns right (Measured).

templates/family-root/scripts/siblings.py (#1216):

• ZERO_COLUMN_CATEGORIES, TWO_COLUMN_WIDTHS, MEDIAL_AND_FINAL_JAMO, MEDIAL_AND_FINAL_JAMO_EXTENDED_B, _char_columns and _columns are the rule adopt-project.py: plan's table pads each name by the columns a terminal gives it, not the code points it holds, so a wide or combining name's leg column lines up under leg with every other row's (#1034) #1178 gives adopt-project.py's plan table (adopt-project.py: plan's table pads a name by code points, not by display width, so a wide or combining name's leg column stands out of line #1034, at c4beba19), unchanged:
  • no column for general category Mn, Me or Cf, or for a Hangul medial vowel or final consonant;
  • two for East Asian width W or F;
  • one for any other character;
  • the category is asked first.
• _padded(text, columns) pads to that many columns, as :<N pads to N code points. A wider text is left whole, with no space added.
• _printed(text) is the text as stdout will write it, in stdout's encoding and with its handler, as adopt-project.py's _printed has it (adopt-project.py: a kept or dropped name reaches the split commit and the pull request body unspelled, and _spelled, _shown and plan's report print text that reads as another name, or raw #900). On an ASCII stdout each escape counts one column per character.
• Sibling.row pads _printed(spelled(self.project)) with _padded(..., 18).
  • kind and state keep :<9 and :<22: both are this file's own ASCII words, and the <target> in a make <target> state is held to MAKE_TARGET_RE.
  • The branch ends the row and is not padded.
• import unicodedata, from the standard library, as os and re are imported.

scripts/validate-repository-naming.py (#1217):

• The same four constants and the same four functions, written in this file.
• _print_classification pads _printed(name) with _padded(..., 32).
• _family_id_padded(family_id) serves _no_family_lines and _family_row.
  • Read for the issue: NamingPolicy.__init__ (scripts/repo_shape.py) checks nothing of an id, and --policy takes any file. So a policy can name a family U+65E5 U+672C U+8A9E, and its --explain row stood 3 columns right at main.
  • A text id is padded by its columns as stdout writes it.
  • An id that YAML reads as a number or a boolean (5, true) reaches _family_row too, and printed through format (5, and 1 for true). It still does, byte for byte.

An ASCII name prints byte for byte as before in every table. The existing rows that pin those bytes pass unchanged: test_a_plain_holder_prints_exactly_what_it_printed_before, and tests/test_chain_link_fold.py's codexGlass rows.

Every family holder will see an upstream-changed row for scripts/siblings.py, and every assembly root one for scripts/validate-repository-naming.py, as #1148 says of a pinned copy whose bytes move. scripts/repo_shape.py is unchanged, so no root or holder sees a row for it.

Where the rule lives, and why

In each of the two files, and not in scripts/repo_shape.py.

• The skew rule. Both files are pinned copies that run beside the root's or the holder's own scripts/repo_shape.py, which update-shape.py apply --accept-local can keep at an older version.
  • An import of a name new to the library would stop make siblings, make park and make resume, as siblings.py's comment at its mv lines says of shell_line, and it would stop the naming gate. Since update-shape.py apply --accept-local scripts/repo_shape.py can keep an old helper that a newly copied file imports from, and make bootstrap then fails with ImportError #536, update-accept-local-missing-name refuses such an apply instead.
  • tests/test_pinned_copy_text.py holds every pinned importer to this rule, and validate-repository-naming.py already writes its own _shown for the same reason.
• The library's bytes. A helper in scripts/repo_shape.py would put an upstream-changed scripts/repo_shape.py row in front of every project and every holder. The holders' row would be for a function their siblings.py could not import. scripts/shape_materialize.py names that cost at each helper it keeps out of the library.
• One answer. tests/test_pinned_copy_columns.py asks both copies the same 14 hand-counted questions: adopt-project.py: plan's table pads each name by the columns a terminal gives it, not the code points it holds, so a wide or combining name's leg column lines up under leg with every other row's (#1034) #1178's table, with these issues' two names. It also holds the two copies to one answer for each of the 1,114,112 code points.
  • adopt-project.py's copy is adopt-project.py: plan's table pads each name by the columns a terminal gives it, not the code points it holds, so a wide or combining name's leg column lines up under leg with every other row's (#1034) #1178's, and this PR leaves adopt-project.py alone.
  • When adopt-project.py: plan's table pads each name by the columns a terminal gives it, not the code points it holds, so a wide or combining name's leg column lines up under leg with every other row's (#1034) #1178 lands, the parity row can take adopt-project.py as a third copy. That row is not added here, because main's adopt-project.py has no _columns.
• Two if statements. The per-character count is written as _char_columns, with two if statements, where adopt-project.py: plan's table pads each name by the columns a terminal gives it, not the code points it holds, so a wide or combining name's leg column lines up under leg with every other row's (#1034) #1178 nests one conditional expression in another (python:S3358, which _print_classification's own comment names). The table and the parity row hold the two forms to the same answers.
• Docstrings of their own. Each copy's docstrings name its own table, so the two copies are not one identical block of tokens.

Decided where the Direction left it open

• Where the rule lives: above.
• --explain's two id rows are changed, because a policy can name an id outside ASCII (scripts/validate-repository-naming.py: the plain listing pads each name from the command line or stdin by code points, not by the columns a terminal gives it, so a wide or combining name's label column stands out of line (twin of #1034) #1217's Direction asked for that to be read before they were left). Ids that are not text keep format.
• Only the name is measured in siblings.py's row. Every other padded field there is the file's own ASCII.
• The tests read each table raw. layout() in tests/test_siblings_family_name.py collapses the pad, so the new rows parse the raw lines, as templates/family-root/scripts/siblings.py: the layout table pads each member's name by code points, not by the columns a terminal gives it, so a wide or combining name's state column stands out of line (the family-root twin of #1034) #1216's Direction says.
• tests/test_chain_link_fold.py's new rows use test_stdout_encoding.run_tool, which reads the bytes and decodes them strictly in the encoding the child was given. run_script, which scripts/validate-repository-naming.py: the plain listing pads each name from the command line or stdin by code points, not by the columns a terminal gives it, so a wide or combining name's label column stands out of line (twin of #1034) #1217 names, decodes in the parent's locale.

Left out of scope

• adopt-project.py belongs to adopt-project.py: plan's table pads each name by the columns a terminal gives it, not the code points it holds, so a wide or combining name's leg column lines up under leg with every other row's (#1034) #1178.
• The rule's own limits, as adopt-project.py: plan's table pads each name by the columns a terminal gives it, not the code points it holds, so a wide or combining name's leg column lines up under leg with every other row's (#1034) #1178 states them, hold for both copies:
  • East Asian ambiguous characters count one column, as on main. A terminal in an East Asian locale draws them two.
  • 72 printable code points that glibc's wcwidth counts two count one, as on main: U+4DC0 to U+4DFF and U+3248 to U+324F.
• Noticed, not changed: the plain listing prints a name raw. _print_classification prints a name from the command line or stdin as it is, not spelled or _shown. An ESC in a name reaches the terminal as itself, and _columns counts it as one column. That is not scripts/validate-repository-naming.py: the plain listing pads each name from the command line or stdin by code points, not by the columns a terminal gives it, so a wide or combining name's label column stands out of line (twin of #1034) #1217's, and lane 3 may file it.
• Noticed, not changed: a non-text id ends two runs in a TypeError. Both happen at main as on this branch, and neither is scripts/validate-repository-naming.py: the plain listing pads each name from the command line or stdin by code points, not by the columns a terminal gives it, so a wide or combining name's label column stands out of line (twin of #1034) #1217's:
  • the plain listing of a name its family matches (found[0] + role_suffix);
  • --explain of a name no family matches (", ".join of the ids: sequence item 5: expected str instance, int found).
• Neither issue has a "Needs a ruling" or "Documented limit" section.

Measured

Scratch probes (report.py, measure_explain.py) ran each tool at main 50ce8ab1, from main's copy of the script, and on this branch, on a UTF-8 stdout and on an ASCII one (PYTHONIOENCODING=ascii:strict, which each script makes backslashreplace).

• Each row is quoted as ascii(name)=column: the name as the row printed it, then the display column its next field begins at.
• Columns are counted with unicodedata by the issue's rule, independently of the code under test.
• The heading's state is at column 31. Every ASCII label begins at 35.
• The holder for siblings.py is a real one from family.py init into local bare remotes. Its members are U+65E5 U+672C U+8A9E, e U+0301 cole and Atlas, each mounted at members/<name>, and the run is --dry-run.

siblings.py --dry-run, `state` column per row (main 50ce8ab1; heading at 31):
  utf-8:strict  exit 0: 'InkRouter'=31, '\u65e5\u672c\u8a9e'=34, 'e\u0301cole'=30, 'Atlas'=31, 'members/\u65e5\u672c\u8a9e'=34, 'members/e\u0301cole'=30, 'members/Atlas'=31
  ascii:strict  exit 0: 'InkRouter'=31, '\\u65e5\\u672c\\u8a9e'=46, 'e\\u0301cole'=36, 'Atlas'=31, 'members/\\u65e5\\u672c\\u8a9e'=46, 'members/e\\u0301cole'=36, 'members/Atlas'=31
siblings.py --dry-run, `state` column per row (branch; heading at 31):
  utf-8:strict  exit 0: 'InkRouter'=31, '\u65e5\u672c\u8a9e'=31, 'e\u0301cole'=31, 'Atlas'=31, 'members/\u65e5\u672c\u8a9e'=31, 'members/e\u0301cole'=31, 'members/Atlas'=31
  ascii:strict  exit 0: 'InkRouter'=31, '\\u65e5\\u672c\\u8a9e'=31, 'e\\u0301cole'=31, 'Atlas'=31, 'members/\\u65e5\\u672c\\u8a9e'=39, 'members/e\\u0301cole'=32, 'members/Atlas'=31
validate-repository-naming.py --role assembly codexGlass <wide> <combining>, label column (main 50ce8ab1):
  utf-8         exit 1: 'codexGlass'=35, '\u65e5\u672c\u8a9e'=38, 'e\u0301cole'=34
  ascii:strict  exit 1: 'codexGlass'=35, '\\u65e5\\u672c\\u8a9e'=50, 'e\\u0301cole'=40
validate-repository-naming.py --role assembly codexGlass <wide> <combining>, label column (branch):
  utf-8         exit 1: 'codexGlass'=35, '\u65e5\u672c\u8a9e'=35, 'e\u0301cole'=35
  ascii:strict  exit 1: 'codexGlass'=35, '\\u65e5\\u672c\\u8a9e'=35, 'e\\u0301cole'=35

On this branch's ASCII stdout, members/ followed by an escaped name is 26 and 19 columns wide, more than the 18 of the field. Each of those two rows keeps the one space between fields, as :<18 keeps it after an ASCII name that long, and nothing is cut.

--explain under the standard policy with its workspace family's id renamed. zz-wip matches that family (_family_row; an ASCII id's pattern starts at 29). z z matches none (_no_family_lines; 23). identical compares the whole stdout of main and the branch, and None means main itself prints no such row (", ".join of a number raises before it).

id '\u65e5\u672c\u8a9e'       utf-8         --explain zz-wip  pattern col main   32 branch   29 exits 0/0  identical: False
id '\u65e5\u672c\u8a9e'       utf-8         --explain z z     pattern col main   26 branch   23 exits 1/1  identical: False
id '\u65e5\u672c\u8a9e'       ascii:strict  --explain zz-wip  pattern col main   44 branch   29 exits 0/0  identical: False
id '\u65e5\u672c\u8a9e'       ascii:strict  --explain z z     pattern col main   38 branch   23 exits 1/1  identical: False
id 'e\u0301cole'              utf-8         --explain zz-wip  pattern col main   28 branch   29 exits 0/0  identical: False
id 'e\u0301cole'              utf-8         --explain z z     pattern col main   22 branch   23 exits 1/1  identical: False
id 'e\u0301cole'              ascii:strict  --explain zz-wip  pattern col main   34 branch   29 exits 0/0  identical: False
id 'e\u0301cole'              ascii:strict  --explain z z     pattern col main   28 branch   23 exits 1/1  identical: False
id '5'                        utf-8         --explain zz-wip  pattern col main   29 branch   29 exits 0/0  identical: True
id '5'                        utf-8         --explain z z     pattern col main None branch None exits 1/1  identical: True
id '5'                        ascii:strict  --explain zz-wip  pattern col main   29 branch   29 exits 0/0  identical: True
id '5'                        ascii:strict  --explain z z     pattern col main None branch None exits 1/1  identical: True
id 'true'                     utf-8         --explain zz-wip  pattern col main   29 branch   29 exits 0/0  identical: True
id 'true'                     utf-8         --explain z z     pattern col main None branch None exits 1/1  identical: True
id 'true'                     ascii:strict  --explain zz-wip  pattern col main   29 branch   29 exits 0/0  identical: True
id 'true'                     ascii:strict  --explain z z     pattern col main None branch None exits 1/1  identical: True

Tests

Every run used a stub gh first on PATH (exit 97), GH_TOKEN=bogus, an empty GH_CONFIG_DIR, GIT_SSH_COMMAND=false and a scratch GIT_CONFIG_GLOBAL, one pytest at a time. Each was /usr/bin/python3 -m pytest -q -p no:cacheprovider --basetemp=<scratch> -o tmp_path_retention_policy=failed <file>, at 352dcbc966d1c42d8510b2d61fedcda533c725cd.

New rows:

• tests/test_siblings_family_name.py: test_the_layout_keeps_every_state_under_its_heading and test_what_ran_keeps_every_state_under_its_heading, each on a UTF-8 and an ASCII stdout.
• tests/test_chain_link_fold.py: test_a_wide_or_combining_name_has_its_label_under_codexglass_s, on both stdouts.
• tests/test_pinned_copy_columns.py (new):
  • the 14-row table for each copy;
  • the parity over every code point;
  • _padded, and _printed under seven stdouts, for each copy;
  • the --explain id rows on both stdouts;
  • the number and boolean ids.

Every new row that reads a table fails at main (main's script swapped in, then restored and checked with cmp):

• 4 of 4 siblings.py rows, at ('<wide>', 34) != ('<wide>', 31);
• 2 of 2 listing rows;
• 8 of 8 --explain id rows.

The two rows for a number or boolean id pass at main too, because they pin main's bytes.

Gates: my files' tests, every test file that names siblings.py or validate-repository-naming.py, and the docs and hygiene guards. combined is one run of tests/test_pinned_copy_columns.py, tests/test_pinned_copy_text.py, tests/test_siblings_family_name.py and tests/test_chain_link_fold.py. The last 15 files in the list also ran with -rs.

• tests/test_adopt_e2e.py and tests/test_adopt_nested_attributes.py first ran before git-filter-repo was on the stub PATH, and skipped 27 and 7 rows. Their lines below are the reruns with it on the PATH, with no row skipped.
• tests/test_family_member_make_jobserver.py: 4 failed, test_under_a_named_pipe_a_member_s_make_before_4_4_runs_serially[bootstrap|validate|park|resume], each at assert "jobserver" not in said. It fails identically at main on this bench (make jobserver warning under the stub PATH); passes in CI. At main 50ce8ab1, under the same environment, -k named_pipe gives the same 4 failed, with 42 passed, 1 skipped and 68 deselected.

combined exit=0 85s :: 315 passed in 82.82s (0:01:22)
test_adopt_pins exit=0 212s :: 152 passed in 209.62s (0:03:29)
test_adopt_reencoded_files exit=0 100s :: 53 passed in 99.26s (0:01:39)
test_adopt_spec_only exit=0 21s :: 14 passed in 19.71s
test_adopt_submodules exit=0 243s :: 167 passed in 240.59s (0:04:00)
test_agents_md_codes exit=0 3s :: 47 passed in 2.59s
test_assembly_root_unreadable_paths exit=0 19s :: 82 passed in 17.15s
test_bootstrap_branch_bytes exit=0 18s :: 62 passed in 16.49s
test_bootstrap_refusal_exit exit=0 12s :: 17 passed in 10.98s
test_chain_pin_sources exit=0 16s :: 52 passed in 14.83s
test_chain_remedy_callers exit=0 17s :: 68 passed in 15.61s
test_cli_reference exit=0 44s :: 23 passed in 43.26s
test_doctor_gitmodules_not_utf8 exit=0 43s :: 23 passed in 42.07s
test_doctor_members_spelled exit=0 20s :: 9 passed in 18.91s
test_doctor_validator_verdicts exit=0 67s :: 39 passed in 65.87s (0:01:05)
test_family exit=0 129s :: 112 passed in 128.30s (0:02:08)
test_family_docs_one_spelling exit=0 2s :: 32 passed in 0.88s
test_family_makefile_siblings exit=0 8s :: 31 passed in 5.83s
test_family_member_make_jobserver exit=1 70s :: 4 failed, 110 passed, 1 skipped in 68.31s (0:01:08)
test_family_members_not_a_list exit=0 106s :: 156 passed in 104.94s (0:01:44)
test_family_printed_quoting exit=0 17s :: 84 passed in 15.92s
test_family_py_unreadable exit=0 46s :: 75 passed in 45.02s
test_family_siblings_bootstrap_unreadable exit=0 21s :: 57 passed in 19.74s
test_family_siblings_placement exit=0 32s :: 33 passed in 30.71s
test_family_siblings_untraced exit=0 97s :: 30 passed in 95.45s (0:01:35)
test_legs_not_a_list exit=0 30s :: 112 passed in 28.65s
test_link_source_chain_defaults exit=0 8s :: 26 passed in 6.84s
test_naming_policy exit=0 20s :: 264 passed in 19.31s
test_no_auto_maintenance exit=0 12s :: 58 passed, 2 skipped in 10.85s
test_non_list_project_fields exit=0 27s :: 71 passed in 25.60s
test_park_resume_targets exit=0 23s :: 44 passed in 22.59s
test_path_guard_hygiene exit=0 19s :: 121 passed in 17.03s
test_pins_child_kind_and_siblings_root_spelling exit=0 6s :: 10 passed in 5.27s
test_repo_hygiene exit=0 4s :: 101 passed in 2.95s
test_scaffold_e2e exit=0 20s :: 46 passed in 19.65s
test_scaffold_pin_and_reuse exit=0 27s :: 75 passed in 25.15s
test_scaffold_referent_chain_links exit=0 11s :: 44 passed in 9.37s
test_setup_family_folder exit=0 15s :: 16 passed in 13.67s
test_setup_project_py exit=0 126s :: 340 passed in 124.45s (0:02:04)
test_siblings_inputs exit=0 37s :: 58 passed in 35.24s
test_template_text_placeholders exit=0 86s :: 142 passed in 85.58s (0:01:25)
test_update_add_codes_documented exit=0 1s :: 6 passed in 0.52s
test_update_shape_add exit=0 26s :: 20 passed in 24.76s
test_update_shape_kept_helper exit=0 25s :: 159 passed in 24.30s
test_update_shape_unreadable exit=0 19s :: 54 passed in 17.40s
test_validate_workflow_gitmodules_urls exit=0 51s :: 421 passed in 49.60s
test_validator_child_encoding exit=0 43s :: 163 passed in 41.67s
test_windows_paths exit=0 5s :: 46 passed in 2.76s
test_yaml_reader_spaces exit=0 31s :: 513 passed in 28.98s
test_adopt_e2e exit=0 28s :: 27 passed in 26.92s
test_adopt_nested_attributes exit=0 99s :: 79 passed in 97.45s (0:01:37)
test_shape_doctor exit=0 766s :: 446 passed, 3 skipped in 763.52s (0:12:43)

Mutants

Each mutant was applied to the committed file, its rows were run, and the file was restored and checked with cmp against HEAD. All 16 were KILLED.

#	Mutant	Rows run	Result
S1	Sibling.row pads the name with :<18 again (main's line)	test_siblings_family_name.py -k keeps_every_state	KILLED, 4 failed
S2	Sibling.row pads without _printed	same	KILLED, 2 failed (the ASCII stdout's), 2 passed
S3	siblings _padded counts len (the issue's mutant)	same	KILLED, 2 failed (the UTF-8 stdout's), 2 passed
S4	siblings _char_columns gives Mn, Me, Cf and the jamo 1 column	test_pinned_copy_columns.py	KILLED, 9 failed, 46 passed
S5	siblings MEDIAL_AND_FINAL_JAMO_EXTENDED_B empty	same	KILLED, 2 failed, 53 passed
S6	siblings TWO_COLUMN_WIDTHS without F	same	KILLED, 2 failed, 53 passed
S7	siblings _printed writes backslashreplace whatever stdout's handler	same	KILLED, 2 failed, 53 passed
N1	the listing pads with :<32 again (main's line)	test_chain_link_fold.py -k label_under_codexglass	KILLED, 2 failed
N2	the listing pads without _printed	same	KILLED, 1 failed (the ASCII stdout's), 1 passed
N3	naming _padded counts len (the issue's mutant)	same	KILLED, 1 failed (the UTF-8 stdout's), 1 passed
N4	naming _char_columns gives W and F 1 column	test_pinned_copy_columns.py	KILLED, 9 failed, 46 passed
N5	_family_id_padded pads every id with format	same	KILLED, 8 failed, 47 passed
N6	_family_id_padded pads every id by columns (5 and true too)	same	KILLED, 2 failed, 53 passed
N7	_family_id_padded pads without _printed	same	KILLED, 4 failed, 51 passed
N8	_no_family_lines pads the id with :<18 again	same	KILLED, 4 failed, 51 passed
N9	_family_row pads the id with :<18 again	same	KILLED, 4 failed, 51 passed

Merges

git merge-tree --write-tree of this head with each, re-listed live with gh pr list --state open --json number,files filtered to this PR's files:

• origin/main 0ed3e57f (two commits past this branch's base 50ce8ab1, touching none of this PR's files): clean, tree 0fd5a789.
• bootstrap.py, validate-family.py and siblings.py: a root or holder whose .git this user cannot read is refused git-repo-unreadable, exit 2, not a traceback or not-a-git-repo, and the doctor's members row names it once, never drift (#1148) #1172 5aa0f116 (templates/family-root/scripts/siblings.py): clean, tree e30259cd.
• family.py and the family root's siblings.py: a holder's .gitmodules this user cannot read, as through a link into a folder it cannot search, is gitmodules-unreadable, never no section for remove's rollback or no url for make siblings (#1110) #1164 74096eb4 (templates/family-root/scripts/siblings.py): clean, tree 9531edfd.
• adopt-project.py: plan's table pads each name by the columns a terminal gives it, not the code points it holds, so a wide or combining name's leg column lines up under leg with every other row's (#1034) #1178 c4beba19 (adopt-project.py, whose plan table this rule copies; none of this PR's files): clean, tree ad19ba53.

Re-listed at 2026-10-23T21:40Z: 48 open PRs, of which #1172 and #1164 touch this PR's files.

Closes #1216
Closes #1217

Lane: openreposhape-3 (openRepoShape-3)

🤖 Generated with Claude Code

[sourcery-ai]

Reviewer's Guide

The PR updates the pinned siblings and repository-naming scripts to pad displayed text by the terminal columns occupied after stdout encoding, fixing alignment for wide, combining, and escaped Unicode names while retaining existing behavior for ASCII and non-text IDs. Comprehensive tests cover both output paths, all Unicode-width rules, stdout handlers, pinned-copy parity, and regression mutants.

Sequence diagram for aligned siblings table output

sequenceDiagram
    participant Sibling
    participant Stdout
    participant Terminal
    Sibling->>Stdout: spelled(project)
    Stdout-->>Sibling: _printed(text)
    Sibling->>Sibling: _padded(text, 18)
    Sibling->>Stdout: row with state and branch
    Stdout->>Terminal: write encoded row
    Terminal-->>Stdout: columns align under state heading

Loading

Sequence diagram for repository naming classification output

sequenceDiagram
    participant Validator
    participant Stdout
    participant Terminal
    Validator->>Stdout: _printed(name)
    Stdout-->>Validator: text as stdout writes it
    Validator->>Validator: _padded(text, 32)
    Validator->>Stdout: _print_classification(name, found)
    Stdout->>Terminal: write classification row
    Terminal-->>Stdout: label aligns under its heading

Loading

Flow diagram for terminal-column-aware name padding

flowchart LR
    Input[Name or family id] --> Printed[_printed: encode with stdout handler]
    Printed --> TextCheck{Text value?}
    TextCheck -->|yes| Columns[_columns: count terminal columns]
    Columns --> Padded[_padded: add spaces to target width]
    TextCheck -->|no family id| Format[Format with existing width]
    Padded --> Output[Aligned table row]
    Format --> Output

Loading

File-Level Changes

Change	Details	Files
Replace code-point-based field padding with terminal-column-aware padding for Unicode names and family identifiers.	Add duplicated Unicode column-width rules, stdout encoding transformation, and padding helpers to both pinned scripts. Use the helpers for the siblings layout/dispatch name field and the naming validator’s plain listing. Use the same behavior for textual --explain family IDs while preserving legacy formatting for numeric and boolean IDs.	templates/family-root/scripts/siblings.py scripts/validate-repository-naming.py
Add regression and parity coverage for Unicode display width, stdout escaping, and pinned-copy consistency.	Verify wide, combining, overlong, ASCII-escaped, and non-UTF-8 cases in siblings and naming output tables. Exercise both --explain family-row paths with Unicode IDs and preserve byte output for non-text IDs. Assert both helper copies agree across all Unicode code points and kill regressions in the width, encoding, and formatting logic.	tests/test_siblings_family_name.py tests/test_chain_link_fold.py tests/test_pinned_copy_columns.py

Assessment against linked issues

Issue	Objective	Addressed	Explanation
#1216	Pad each member name in the make siblings layout table by the display columns of the text as stdout writes it, keeping ASCII output unchanged and leaving over-width names untruncated.	✅
#1216	Use the same corrected Sibling.row padding in the dispatch’s (b) what ran rows, which also use that row formatter.	✅
#1217	Pad names in the plain listing by the display columns of the text as stdout will encode it, keeping ASCII output unchanged and leaving names wider than the field untruncated.	✅

Possibly linked issues

• templates/family-root/scripts/siblings.py: the layout table pads each member's name by code points, not by the columns a terminal gives it, so a wide or combining name's state column stands out of line (the family-root twin of #1034) #1216: The PR changes Sibling.row to pad names by their rendered display columns, aligning state for wide and combining names as requested.

---

Tips and commands

Interacting with Sourcery

• Trigger a new review: Comment @sourcery-ai review on the pull request.
• Continue discussions: Reply directly to Sourcery's review comments.
• Generate a GitHub issue from a review comment: Ask Sourcery to create an
  issue from a review comment by replying to it. You can also reply to a
  review comment with @sourcery-ai issue to create an issue from it.
• Generate a pull request title: Write @sourcery-ai anywhere in the pull
  request title to generate a title at any time. You can also comment
  @sourcery-ai title on the pull request to (re-)generate the title at any time.
• Generate a pull request summary: Write @sourcery-ai summary anywhere in
  the pull request body to generate a PR summary at any time exactly where you
  want it. You can also comment @sourcery-ai summary on the pull request to
  (re-)generate the summary at any time.
• Generate reviewer's guide: Comment @sourcery-ai guide on the pull
  request to (re-)generate the reviewer's guide at any time.
• Resolve all Sourcery comments: Comment @sourcery-ai resolve on the
  pull request to resolve all Sourcery comments. Useful if you've already
  addressed all the comments and don't want to see them anymore.
• Dismiss all Sourcery reviews: Comment @sourcery-ai dismiss on the pull
  request to dismiss all existing Sourcery reviews. Especially useful if you
  want to start fresh with a new review - don't forget to comment
  @sourcery-ai review to trigger a new review!

Customizing Your Experience

Access your dashboard to:

• Enable or disable review features such as the Sourcery-generated pull request
  summary, the reviewer's guide, and others.
• Change the review language.
• Add, remove or edit custom review instructions.
• Adjust other review settings.

Getting Help

• Contact our support team for questions or feedback.
• Visit our documentation for detailed guides and information.
• Keep in touch with the Sourcery team by following us on X/Twitter, LinkedIn or GitHub.

[sonarqubecloud]

Quality Gate passed

Issues
0 New issues
0 Accepted issues

Measures
0 Security Hotspots
0.0% Coverage on New Code
0.0% Duplication on New Code

See analysis details on SonarQube Cloud

[brettheap]

(lane openRepoShape-3, 2026-10-23T21:51:12Z) RV1236 independent review of #1236 at 352dcbc966d1c42d8510b2d61fedcda533c725cd: READY. No MUST and no SHOULD in this PR.

Rule: the three copies give one answer for all 1,114,112 code points (0 differ, unicodedata 15.0.0): siblings.py and validate-repository-naming.py at the head, and adopt-project.py at #1178's c4beba19. The four constants are equal, and _padded and _printed agree under six stdout setups. RV1236's own probe, AST-extracted.

Measured with glibc wcswidth, main 50ce8ab1 -> head:

• Naming listing, label column (codexGlass at 35): UTF-8 wide 38 and combining 34 -> 35; ASCII 50 and 40 -> 35. Hangul jamo and fullwidth names also land at 35.
• Sibling.row, state column (heading at 31): UTF-8 34 and 30 -> 31; ASCII 46 and 36 -> 31. On ASCII, members/<wide> is at 39: over-wide, one space, nothing cut.
• ASCII names, and --explain with ids workspace, 5 and true: stdout is byte-identical to main on UTF-8 and ASCII. For 5 and true with 'z z', stderr differs only in traceback paths and line numbers (main's TypeError).

Imports: only import unicodedata is new; test_pinned_copy_text.py passes.

Tests at the head, in one run of the PR's files, test_pinned_copy_text and the three hygiene files: 595 passed, 2 skipped.

Mutants, RV1236's own script, each KILLED, restored and cmp-equal:

• R1, siblings _padded by len: 2 failed;
• R2, _family_id_padded pads every id by columns: 2 failed;
• R3, the listing without _printed: 1 failed;
• R4, siblings' width asked before the category: 2 failed.

Jobserver: test_family_member_make_jobserver.py -k named_pipe gave 46 passed, 1 skipped at the head AND at main in RV1236's environment. The writer's 4 failures did not reproduce there. Either way, it is no regression.

Merges, re-listed live (50 open): merge-tree is clean with main 0ed3e57f, #1172 5aa0f116, #1164 74096eb4 and #1178 c4beba19, with the same trees as the body.

Sonar: gate OK, 0 issues, 0.0% duplication. CI at 21:50Z: package, tests-git-bash and SonarCloud pass; tests is pending.

NITs:

• The commit message says "the rule adopt-project.py: plan's table pads a name by code points, not by display width, so a wide or combining name's leg column stands out of line #1034 gives adopt-project.py's plan table". That rule is adopt-project.py: plan's table pads each name by the columns a terminal gives it, not the code points it holds, so a wide or combining name's leg column lines up under leg with every other row's (#1034) #1178's (open, c4beba19), not main's. The explicit squash message says so.
• A Sourcery block follows the robot line (no closing keyword); the explicit squash message drops it.

Follow-up: when #1178 lands, the parity row (tests/test_pinned_copy_columns.py:115) should take adopt-project.py as a third copy. Whichever of the two PRs lands second adds it, or a follow-up issue does.

#1236 lands under Brett Heap's v1.2.1 merge word once tests is green, at its slot at the end of wave 2.

Lane: openreposhape-3 (openRepoShape-3)