Word the tail-part comma report without naming a field (#629)

[derek73]

Summary

Segment's comma-structure report on a part past the second comma said the part was "consumed as suffix best-effort". That was true through 2.3.0, when the whole part read as suffix. Since #603 (rules.md#C2), a title word there reads as a title, so the report named a field the word isn't in. That happened this cycle and never shipped.

The report now names no field:

Input	Before	Now
John Smith, Jr., Dr. Bart	…; consumed as suffix best-effort	…; read best-effort
John Smith, Jr., Bart	…; consumed as suffix best-effort	…; read best-effort

It says what is certain, that the part is beyond the recognized comma structures, and the fields say where each word went. The input is malformed, so the fix is one string, not machinery. Report kinds are unchanged; only the detail text moves.

History

The first commit (0d5fea4) named every field the part's words land in ("consumed as title and suffix best-effort"), worded at assemble from the final roles. Review found that it withdrew the report from a name emptied of content (", , -"), against rules.md#A2, and it needed a second tail on PendingAmbiguity, an assemble branch and an equality check in the field guard. The second commit (14c750d) reverts all of that for the neutral wording. decisions.md#A1's entry records the declined design and why.

Measured (decisions.md#A1, 2026-10-10)

Against master bc24386, comparing every report's kind, detail and referents plus the seven fields:

• Population: the corpora and the case table (1,925 texts), plus a tail grid of 97,650 texts, each under the three orders.
• What moves: only the detail of comma-structure reports (90 in the named texts, 247,353 in the grid). No report appears or goes; ", , -" keeps its report as on master.
• Five differential gates: all exit 0. They compare report kinds, which don't move.

Tests

• test_segment.py: the report's verbatim detail on John Smith, Jr., Dr. Bart.
• The field guard (test_a_report_names_the_field_its_word_lands_in) is unchanged: the new wording names no field, so there is nothing for it to check.

Docs

• decisions.md#A1: new 2026-10-10 entry (decided, declined with the review's finding, measured); replaces the Open: #629 line.
• mechanisms.md and AGENTS.md: the Smith, Jr., Dr. Bart reports its last part as "consumed as suffix" when 'Dr.' is read as a title #629 notes now say the report names no field.
• Release log: the Should a comma part opening with a credential be a suffix comma? John Smith, PhD Jones reads family John Smith #603 bullet shows the new wording through parse(...), and closes Smith, Jr., Dr. Bart reports its last part as "consumed as suffix" when 'Dr.' is read as a title #629.

Closes #629

🤖 Generated with Claude Code

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 99.01%. Comparing base (bc24386) to head (14c750d).

Additional details and impacted files

@@           Coverage Diff           @@
##           master     #630   +/-   ##
=======================================
  Coverage   99.01%   99.01%           
=======================================
  Files          46       46           
  Lines        4486     4486           
=======================================
  Hits         4442     4442           
  Misses         44       44           

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.