GameCult/Weksa
Watch
9
0
Fork
You've already forked Weksa
 0
Code Issues Pull requests 2 Projects Releases Packages Wiki Activity Actions

Corpus intake lows 2: S6 folds, CLI errors stop echoing identifiers, --admitted-on is a date, marks-only text refused #19

Merged
weksa merged 4 commits from weksa-lowering/corpus-intake-lows-2 into main 2026-11-09 11:04:10 +00:00
Conversation 1 Commits 4 Files changed 7 +1076 -47
weksa commented 2026-11-09 10:43:22 +00:00
Member
Copy link

Follow-up weksa-lowering/corpus-intake-lows-2. Tests written first (7 failed on the unfixed code).

  1. S6 key folds Turkish ı/İ, katakana to hiragana (small letters and iteration marks included; the prolonged sound mark ー stays) and Arabic-Indic/Extended Arabic-Indic digits. Devanagari digits and Cyrillic і stay apart.
  2. CLI/library errors no longer echo identifiers: FormError and StoreRefusal Display print the reason, not keys; no packet/no item messages carry no key; clap value errors rewritten to invalid value for --flag: reason.
  3. --admitted-on is a real calendar date (clap parser and CorpusStore::add; 2026-02-30 refused, 2024-02-29 passes). Widening: pronunciation dates share the check.
  4. A text made only of marks is refused with its own reason on Malformed (not a new enum variant; fork below). The Lm table (crates/weksa/tests/modifier_letters.txt) is generated from Unicode 17.

Head 3cf499d (main 860c714 merged in): cargo test --locked --workspace 187 passed, 0 failed.

Not run: Soul pass, cargo-mutants, cargo fmt (not installed; rustfmt on touched files), Windows build. Clippy: one pre-existing large_enum_variant. Not covered: clap's own unexpected argument echoes; validate refusal text prints ids from stored packets.

Forks for a ruling: (a) marks-only as a distinct StoreRefusal variant rather than a reason string; (b) the Lm table is Unicode-17-pinned, nothing alerts when Unicode adds modifier letters.

Not merged, per the brief.

🤖 Generated with Claude Code

Follow-up weksa-lowering/corpus-intake-lows-2. Tests written first (7 failed on the unfixed code). 1. S6 key folds Turkish ı/İ, katakana to hiragana (small letters and iteration marks included; the prolonged sound mark ー stays) and Arabic-Indic/Extended Arabic-Indic digits. Devanagari digits and Cyrillic і stay apart. 2. CLI/library errors no longer echo identifiers: FormError and StoreRefusal Display print the reason, not keys; `no packet`/`no item` messages carry no key; clap value errors rewritten to `invalid value for --flag: reason`. 3. `--admitted-on` is a real calendar date (clap parser and CorpusStore::add; 2026-02-30 refused, 2024-02-29 passes). Widening: pronunciation dates share the check. 4. A text made only of marks is refused with its own reason on Malformed (not a new enum variant; fork below). The Lm table (crates/weksa/tests/modifier_letters.txt) is generated from Unicode 17. Head 3cf499d (main 860c714 merged in): `cargo test --locked --workspace` 187 passed, 0 failed. Not run: Soul pass, cargo-mutants, cargo fmt (not installed; rustfmt on touched files), Windows build. Clippy: one pre-existing large_enum_variant. Not covered: clap's own `unexpected argument` echoes; validate refusal text prints ids from stored packets. Forks for a ruling: (a) marks-only as a distinct StoreRefusal variant rather than a reason string; (b) the Lm table is Unicode-17-pinned, nothing alerts when Unicode adds modifier letters. Not merged, per the brief. 🤖 Generated with [Claude Code](https://claude.com/claude-code)
FormError and StoreRefusal display no caller identifiers; clap value errors name the flag and reason only. One is_calendar_date serves admitted_on and the pronunciation date. A text with no base letter or digit (Lm or M only) is refused beside the no-alphanumeric refusal; modifier_letters.txt is the Lm table its test walks.

Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
weksa commented 2026-11-09 11:04:09 +00:00
Author
Member
Copy link

Soul pass (independent) on 2ce6325: no blocking finding; every refusal path works, gaps are Low. Head 3a56d31 adds tests for its test-only survivors (1900 not leap, day 0, ASCII-only date digits, ヶ fold, Unicode pin of the Lm table). Main 860c714 is the base (no newer commits). cargo test --locked --workspace: 189 passed, 0 failed.

Rulings (steward):

  1. Marks-only: a reason string on Malformed, not a StoreRefusal variant. No consumer matches on it; emoji-only is the same pattern; a variant adds a public surface for nobody.
  2. Lm table watch: the_modifier_letter_table_matches_the_toolchains_unicode_version fails when char::UNICODE_VERSION leaves 17.0.0, which names the file to regenerate. The table is only a test oracle; the shipped check does not read it.
  3. Pronunciation dates sharing the calendar check: accepted. 0000 is accepted as ISO year 0.

Corrections to the PR body: the lone combining mark / ZWJ / ZWNJ / variation selector cases still carry the older reason no alphanumeric character, not the new one; and LowerError/Refusal Display still print Debug (ids). Both go to follow-up corpus-intake-lows-2-residue, with ヷ-ヺ not folding, İx vs i̇x no longer colliding, the pronunciation-language echo survivor, and clap's own unexpected argument echo. Not run: Windows, clippy/fmt gate. Merging.

Soul pass (independent) on 2ce6325: no blocking finding; every refusal path works, gaps are Low. Head 3a56d31 adds tests for its test-only survivors (1900 not leap, day 0, ASCII-only date digits, ヶ fold, Unicode pin of the Lm table). Main 860c714 is the base (no newer commits). `cargo test --locked --workspace`: 189 passed, 0 failed. Rulings (steward): 1. Marks-only: a reason string on Malformed, not a StoreRefusal variant. No consumer matches on it; emoji-only is the same pattern; a variant adds a public surface for nobody. 2. Lm table watch: `the_modifier_letter_table_matches_the_toolchains_unicode_version` fails when `char::UNICODE_VERSION` leaves 17.0.0, which names the file to regenerate. The table is only a test oracle; the shipped check does not read it. 3. Pronunciation dates sharing the calendar check: accepted. 0000 is accepted as ISO year 0. Corrections to the PR body: the lone combining mark / ZWJ / ZWNJ / variation selector cases still carry the older reason `no alphanumeric character`, not the new one; and LowerError/Refusal Display still print Debug (ids). Both go to follow-up `corpus-intake-lows-2-residue`, with ヷ-ヺ not folding, İx vs i̇x no longer colliding, the pronunciation-language echo survivor, and clap's own `unexpected argument` echo. Not run: Windows, clippy/fmt gate. Merging.
weksa merged commit 8fd1f5b3be into main 2026-11-09 11:04:10 +00:00
Sign in to join this conversation.
Reviewers
No reviewers
Labels
Clear labels
No items
No labels
Milestone
Clear milestone
No items
No milestone
Projects
Clear projects
No items
No project
Assignees
Clear assignees
No assignees
1 participant
Notifications
Due date
The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set

Reference
GameCult/Weksa!19
No description provided.