[codeGlaze]

Description:

Two pieces, in separate commits so they review apart.

1. A shared picks namespace (439593dd, b33c765a, 909655cf, 453aff1d)

One engine over a character's stored picks, for two pieces of planned work:

• the character-rescue work, which must list and remove picks on the server without the app;
• the hidden-pick work, which sets picks aside and puts them back.

The KB spec is character-rescue.md, "Layer 1 spec", on feature/grant-rows.

• Moved, not rewritten. walk-entries, walk-picks, picks-of and relink-picks leave the browser-only content_reconciliation.cljs for src/cljc/orcpub/dnd/e5/picks.cljc, as walk (now public), walk-typed, keys-of and relink.
  • The bodies are token-identical to the originals, comments stripped and strings masked.
  • The two callers are updated, with no aliases left behind.
  • Their test moves to test/cljc, so it also runs in the JVM suite that CI gates on. It's also listed in the browser test runner.
• New: remove-at and put-at.
  • A pick is addressed by [selection key, entry key] pairs (the app's option-path format), never by index, so removing one pick can't move another.
  • :removed records :multiselect? (the template's own word) because an emptied or absent selection no longer shows its shape. An emptied multiselect keeps [], the same state unticking leaves, and saving strips it (entity/remove-empty-fields).
  • put-at writes a pick back only into an empty place. If a multiselect already holds the key, or a one-pick selection already holds any pick, it writes nothing and returns nil, as it does when the path is gone. A pick the player made while the old one was set aside is never overwritten or duplicated; nil tells the caller the held copy is retired. It checks no rule.
  • Tests fail if remove-at drops by position, if put-at loses a vector's metadata, if it duplicates a re-picked key, or if it overwrites a one-pick selection (each checked by removing the guard).
• Not touched: entity/flatten-options (the build) and extract-content-keys (Missing Content).

2. Lint is clean (ad3c612b, 29e9e2b6)

The 50 warnings and 3 infos on integration are fixed, not suppressed. No behaviour change.

• Locals that shadowed core or library vars are renamed for what they hold.
  • old/new pairs become before/after.
  • after becomes plugins-after; var becomes env-var; get becomes lookup; integer? becomes integer-setting?; import becomes imported-pack; choice-tags becomes tags-by-selection; plus test locals.
  • Map keys are unchanged ({env-var :var …}).
• Unused refers and imports are removed. The seven in equipment_subs were used only by the commented-out ORPHANED remote-item block; they come back with it.
• Smaller fixes: three nested lets merged, two redundant str calls and a duplicate require dropped.
• One more from the merge: the locale-safety merge brought in one more warning, an unused hostile-locales in pedestal_test. It's removed too.

integration (locale safety, #46) is merged in (8877678f). The only conflict was CHANGELOG.md; both blocks are kept.

A follow-up is recorded in the shared TODO on agents/develop: declare which config settings expect a whole number, rather than inferring it from the capacity group.

Checklist:

• The code change is tested and works locally. On the head: lein lint 0 errors, 0 warnings; lein test 571 tests / 0 failures; cljs suite 546 tests / 0 failures (run by hand); lein fig:build clean.
• I have commented my code, particularly in hard-to-understand areas
• I have made corresponding changes to the documentation if necessary
• There is no commented out code in this PR.
• My changes generate no new warnings (check the console)
• docs/branch-changelog.md is updated in the house style (docs/branch-changelog.template.md), and I decided whether this branch earns a ## Highlights. No Highlights (refactor and cleanup); folded into Summer Patch (3cbe12df).

[codeGlaze]

@greptileai review

[greptile-apps]

Confidence Score: 5/5

[Medium risk] Refactors pick-handling code and fixes lint warnings.

The PR appears safe to merge based on the reviewed changes and resolved findings.

Summary

The PR moves pick traversal into a shared namespace, adds pick removal and restoration helpers, and cleans up lint warnings. The follow-up change makes restoration return nil rather than overwrite a selection or duplicate an existing pick.

Diagram

%%{init: {'theme': 'neutral'}}%%
flowchart LR
  A[Stored character picks] --> B[remove-at]
  B --> C[Character without pick]
  B --> D[Removed pick and address]
  C --> E[put-at]
  D --> E
  E --> F{Address available and slot eligible?}
  F -->|Yes| G[Restored character]
  F -->|No| H[nil; nothing written]

Loading

_{Reviews (4) · Last reviewed commit: "picks/put-at: never overwrite a one-pick..."}

[codeGlaze]

Both review findings are addressed in 909655cf.

1. Duplicate multiselect picks: valid, and fixed. put-at now leaves a multiselect vector unchanged when it already holds the pick's key, so it never appends a second copy. A new test covers the re-picked-meanwhile case, and it fails without the check.
2. Browser suite skips pick tests: half right. picks-test is now listed in orcpub.test-runner, so the per-test reporter mode runs it. But the full browser suite (test/e2e/cljs-harness.js) already ran it: it picks up every compiled test namespace. Its count went from 2827 to 2828 assertions with the new test.

Checks on 909655cf: lint 0 warnings, JVM 542 tests, browser 546 tests, all passing.

[codeGlaze]

@greptileai review

[codeGlaze]

@greptileai review

[codeGlaze]

@greptileai review