Complete editor behavior parity across the supported content pack

[zoeyrose]

Outcome

Burn down the M1 capability inventory in editor#15 until the released MIT editor/toolkit/renderer workflow provides equivalent supported authoring functionality across the complete baseline Atrinik content pack.

This is the editor-owned M5 parity gate. Packaging, operational hardening, documentation cutover, and Gridarta/archive retirement remain M6 work in editor#13 and atrinik#259.

Scope and invariants

• Review every required inventory row against its final editor, content-toolkit, renderer, or wrapper owner and shipped versioned contract.
• Add focused implementation issues for uncovered behavior rather than implementing unrelated domains in this epic.
• Exercise project discovery, document lifecycle, lossless transactions, map and non-map authoring, catalog/search, inspectors, validation/diagnostics, shared-renderer preview/picking, automation, recovery, packaging inputs, and isolated playtest.
• Run no-op round trips over the entire supported corpus and representative targeted edits for every syntax/content/workflow class.
• Verify that unknown fields, comments, multiline text, nesting, object order, stable identities, links, attribution, and untouched bytes survive as required.
• Test clean and incremental indexing/builds, external edits, stale preconditions, malformed/adversarial files, cancellation, disk/permission/device failure, crash recovery, and concurrent read/build behavior.
• Use released content-toolkit and renderer APIs. Do not add a private parser, writer, validator, renderer, pixel-selection path, server manager, or client dependency to close a gap.
• Record approved modernized UX differences separately from authored-data, safety, and workflow differences. No supported content/gameplay design choice may disappear silently.

Whole-pack verification

• Pin exact content@main, toolkit, renderer, resources, and schema/package coordinates.
• Prove every supported source file is discoverable, inspectable, validatable, compilable, and represented by an owning schema/workflow; zero paths may be silently skipped.
• Open and no-op-save a representative file from every grammar/content class, plus randomized/adversarial samples and the largest projects.
• Perform add/remove/move/copy/set/unset/header/link operations, multi-file transactions, undo/redo, dry-run/diff, save/reopen, recovery, preview, and isolated playtest with deterministic assertions.
• Validate ordinary, multipart, tiled, linked-depth, lighting/fog/cutaway, animation, inventory/nesting, quest/interface/dialogue, NPC/service, treasure/artifact/faction, and attribution workflows.
• Publish a machine-readable matrix result consumable by atrinik#279 and the final atrinik#280 aggregate gate.

Acceptance criteria

• editor#15 contains zero unowned, unimplemented, unverified, or ambiguous required rows.
• Every exclusion or approved difference has an explicit product decision and cannot change authored semantics, attribution, losslessness, or safe-write guarantees accidentally.
• Whole-corpus no-op processing is byte-identical where required; targeted edits create bounded semantic/text diffs and never churn unrelated content.
• Human UI and versioned automation/CLI paths exercise the same toolkit transactions and renderer scene semantics.
• Linux and Windows runs cover representative input/display/backend combinations, accessibility, recovery, and deterministic headless/offscreen evidence where applicable.
• The supported edit/validate/preview/playtest workflow requires no Java, Gridarta, old Python tools, classic binaries, the atrinik/classic source checkout, or writes into mutable runtime/generated source paths.
• The parity report contains exact release coordinates, fixtures, provenance/licenses, reproduction commands, evidence, and owning issues for any remaining non-cutover work.
• atrinik#280 can consume the report without a second editor-specific comparison schema.

Dependencies and parallelization

Depends on the completed inventory in editor#15, M2 toolkit/renderer contracts, the M4 map editor/automation workflow (editor#3–#6, #10–#12), and required M5 non-map panels in editor#7.

Map, structured-content, validation/automation, preview/rendering, playtest, platform/accessibility, and failure/recovery rows can be implemented and verified in parallel. Whole-corpus no-op checks should run continuously; final closure waits for every required domain report, not M6 classic archival.

Licensing

Editor code/tests are MIT. Verified original work from an approved MIT provenance grantor may be copied, migrated, or translated only after the complete-history audit and third-party review required by the merged grant policy. Authored content/assets and fixtures retain their exact per-file licenses and attribution.