fix: format_P (%P) uses language-specific AM/PM values

[Koan-Bot]

What

%P (lowercase am/pm) now returns the correct language-specific value instead of always returning English "am"/"pm".

Why

format_P in Generic.pm used the lexical @AMPM array directly — same inheritance bug pattern as format_o (#107). Language modules copy format_p into their own namespace (where it references the module's @AMPM), but format_P was never copied, so it always fell through to Generic's English array.

Affected languages: Dutch (VM/NM), Finnish (ap/ip), Hungarian (DE./DU.), Czech (dop./odp.), Chinese, Arabic, Greek, and 10+ others.

How

One-line fix: changed format_P from reimplementing the AM/PM logic with the lexical array to delegating via method dispatch ($_[0]->format_p), which correctly resolves to the language module's copied format_p.

Testing

• Added %P test to English baseline section
• Added Dutch test section with both AM and PM timestamps, verifying %p returns VM/NM and %P returns vm/nm
• Full test suite passes (all files)

🤖 Generated with Claude Code

---

Quality Report

Changes: 2 files changed, 12 insertions(+), 2 deletions(-)

Code scan: clean

Tests: skipped

Branch hygiene: clean

Generated by Kōan post-mission quality pipeline

[atoomic]

LGTM

[greptile-apps]

Greptile Summary

This PR fixes a one-line bug in Date::Format::Generic where format_P (%P, lowercase am/pm) always returned the English value instead of the language-specific value. The fix delegates to format_p via method dispatch ($_[0]->format_p) so that language modules with their own format_p (e.g., Dutch returning VM/NM) are correctly lowercased rather than bypassed. Tests cover both the English baseline and Dutch AM/PM cases.

Confidence Score: 5/5

This PR is safe to merge — minimal one-line fix with correct delegation and well-targeted tests.

The fix is a clean, minimal delegation that correctly resolves the inheritance bug for all language modules. Test count arithmetic checks out (292 + 7 = 299), and the Dutch AM/PM cases verify both the regression scenario and the new behavior. No regressions for languages without a custom format_p since they fall through to Generic's implementation as before.

No files require special attention.

Important Files Changed

Filename	Overview
lib/Date/Format/Generic.pm	One-line fix: format_P now delegates to format_p via method dispatch instead of accessing the lexical @AMPM array directly, fixing language-specific lowercase AM/PM.
t/format.t	Test count correctly updated from 292 to 299 (+7); adds English %P baseline and a Dutch section covering both PM (NM→nm) and AM (VM→vm) timestamps.

Sequence Diagram

sequenceDiagram
    participant C as Caller
    participant G as Generic.pm
    participant D as Dutch.pm

    Note over C,D: %P format dispatch (after fix)
    C->>G: format_P($obj)
    G->>D: $obj->format_p()
    D-->>G: "VM" or "NM"
    G-->>C: lc("VM") → "vm"

    Note over C,G: %P format dispatch (before fix)
    C->>G: format_P($obj)
    G->>G: lc($AMPM[...]) using Generic's @AMPM
    G-->>C: "am" or "pm" (always English)

Loading

_{Reviews (1): Last reviewed commit: "fix: format_P (%P) now uses language-spe..." | Re-trigger Greptile}