Skip to content

Bounded global-rule array local storage - #56

Draft
Teakowa wants to merge 4 commits into
mainfrom
codex/pr45-rebuild
Draft

Bounded global-rule array local storage#56
Teakowa wants to merge 4 commits into
mainfrom
codex/pr45-rebuild

Conversation

@Teakowa

@Teakowa Teakowa commented Aug 24, 2026

Copy link
Copy Markdown
Contributor

No description provided.

Extend deterministic rule-local global storage to nested arrays of scalar WIR values and local array foreach collections, while failing closed for object/reference and re-entrant semantics.

Refs #31
Assert exact local-array foreach action ordering, generated helper identities, provenance, and an unsupported vector-array diagnostic.

Refs #31
Verify exact source spans for materialized array locals and foreach collection/index helpers.

Refs #31
@Teakowa

Teakowa commented Aug 24, 2026

Copy link
Copy Markdown
Contributor Author

Independent QA disposition: NOT VERIFIED for merge. All del-rs implementation, safety, provenance, ownership, full-suite, corpus, matrix, support, and hosted CI gates pass. The remaining blocker is the authoritative workshop-rs canonical contract: emitted WIR Value::Array reparses as Call("array", ...), so roundtrip::equivalent cannot prove runtime semantic equivalence. This PR intentionally does not bypass or duplicate that workshop-rs behavior. Retain the #45 goal and re-run QA after the owning workshop-rs fix.

@Teakowa
Teakowa marked this pull request as draft August 24, 2026 19:05
@Teakowa

Teakowa commented Aug 25, 2026

Copy link
Copy Markdown
Contributor Author

Status refresh (2026-08-25)

The earlier workshop-rs canonical Array roundtrip concern was rechecked against current workshop-rs main. Indexed array/member writes and the bounded array composition now have positive canonical parse/validate/emit evidence, and del-rs current main passes the focused lowering suite.

This does not expand PR #56 scope: it remains a bounded global-rule array-local slice, not object/reference/generation runtime lowering and not #31 completion. Keep the PR Draft until fresh QA reviews the exact head; the remaining object/runtime evidence belongs to DEL #31 and live Workshop behavior remains unverified.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant