[aeppling]

develop does not compile since #1503 merged (e8541d1). Release PR #3721 and every build job fail on it.

PR #1503 was authored 2026-04-24 and merged 2026-08-31 without a rebase. Two things changed underneath it in that window:

• 5269df7 removed lazy_static from Cargo.toml in favour of LazyLock. phpt_cmd.rs still calls lazy_static::lazy_static!, so the macro fails to expand and STATUS_RE, STAT_RE, NUMBER_OF_TESTS_RE and TIME_RE are never defined.
• RtkRule gained pipeline_final_safe. The new rule in rules.rs spells its fields out rather than using ..RtkRule::DEFAULT, so it misses the new field.

This repairs @iliaal's feature rather than reverting it — the phpt filter itself is fine, only these two touchpoints went stale.

Changes

• phpt_cmd.rs: the four statics become LazyLock, matching the other 25 files and the LazyLock rule in CLAUDE.md.
• rules.rs: use ..RtkRule::DEFAULT like every neighbouring entry, so future fields do not break it again.

Verification

before: e4b9705 (pre-merge develop)  0 errors
        e8541d1 (merge, = develop)   7 errors
after:  this branch                  0 errors

cargo fmt --all --check clean, cargo clippy --all-targets 0 warnings, cargo test --all 2965 passed, 0 failed — the phpt tests run for the first time, since compilation previously stopped before them.

Process note

Checks on #1503 passed against April's develop, not the current one. Requiring branches to be up to date before merge would prevent this class.

[pszymkowiak]

Verified locally: develop at e8541d1 fails with exactly the 7 reported errors (lazy_static macro gone + missing pipeline_final_safe). On this branch: cargo fmt --check clean, clippy 0 warnings, 2965 tests pass. Minimal, correct repair of the two stale touchpoints — and the ..RtkRule::DEFAULT change future-proofs the rule. Agreed on the process note: requiring branches up to date before merge would have caught this. LGTM.