Fix main CI red on stale digest pending-list entries

[Mikola Lysenko (mikolalysenko)]

Problem

Main's test suite fails on every platform. utils::digest::tests::production_digests_go_through_the_helpers (socket-patch-core lib) fails on origin/main (db83f01):

left:  ["utils/group_commit.rs", "vendor/jvm/mod.rs", "vendor/maven_repo.rs", "vendor/pypi.rs", "vendor/redownload.rs", "vendor/yarn_berry_lock.rs"]
right: ["crawlers/gradle_cache.rs", "patch/jvm_jar.rs", "patch/sidecars/maven.rs", "utils/group_commit.rs", ...]

Any PR that merges main inherits this red test (ubuntu/macos/windows), test-release and coverage. The same ratchet also failed in the other direction on the last two scheduled main CI runs at 9c43dfc: 37578193876 (2026-10-07) failed test-release, coverage, test (windows-latest) and test (macos-latest), and 37420351192 (2026-10-06) also failed. Those runs showed the three files computing digests inline while missing from the list.

Root cause

This is a semantic merge race between two PRs that each passed on their own:

• Fix scan/get --json dropping apply failures (#424) #955 (cd7dd92) added crawlers/gradle_cache.rs, patch/jvm_jar.rs and patch/sidecars/maven.rs to PENDING_INLINE_DIGESTS, because those files still computed digests inline on its base.
• sbt, Mill and scala-cli support in agent, hosted and vendored modes #690 (5a96c52) moved the same three files onto the utils::digest::*_of helpers.

After both merged, the list has three stale entries. The test is designed to fail on stale entries.

Fix

Remove the three stale entries. The change is 3 deleted lines in crates/socket-patch-core/src/utils/digest.rs. No assertion is weakened, and the ratchet still catches new inline digests and stale entries.

Proof

• Before: cargo test -p socket-patch-core --lib production_digests_go_through_the_helpers fails locally on origin/main with the diff shown above.
• After: that test passes. The full cargo test -p socket-patch-core --lib run gave 5558 passed and 4 failed. The 4 failures only happen because the sandbox runs as root, where chmod-0o555 read-only-dir tests cannot fail as they expect: copy_tree::relax_loop_must_not_traverse_symlinked_root, vlt_heal::an_unremovable_hidden_lock_keeps_every_store_entry, pypi_poetry::wire_write_failure_… and pypi_requirements::wire_failure_rolls_back_…. They are unrelated to this change, and the CI runner (non-root) passed them in run 37578193876.
• rustfmt --check is clean on the touched file.

Note for open PRs

#876 and #889 each add crawlers/gradle_cache.rs back to this list. They reintroduce inline digests in that file, so that part of their change stays correct after this lands.

Where tests run

No tests were removed or moved.

🤖 Generated with Claude Code

https://claude.ai/code/session_01Qq8uwZ9NTCZXZRygn7woy7

---

Generated by Claude Code

---

Note

Low Risk
Test-only list maintenance in digest.rs; no production digest behavior changes.

Overview
Fixes main CI by syncing PENDING_INLINE_DIGESTS in utils/digest.rs with reality after two PRs merged in opposite directions: crawlers/gradle_cache.rs, patch/jvm_jar.rs, and patch/sidecars/maven.rs now use utils::digest::*_of helpers, so they are removed from the pending list.

The ratchet test production_digests_go_through_the_helpers is unchanged—it still fails on new inline digest copies and on stale list entries; only the three obsolete paths are dropped.

^{Reviewed by Cursor Bugbot for commit 6cb46c0. Configure here.}

---

Generated by Claude Code

[Mikola Lysenko (mikolalysenko)]

bugbot run

---

Generated by Claude Code

[cursor]

✅ Bugbot reviewed your changes and found no new issues!

Comment @cursor review or bugbot run to trigger another review on this PR

^{Reviewed by Cursor Bugbot for commit 6cb46c0. Configure here.}