[KonstantinMirin]

What this changes

The draft links line items in two ways, and neither keeps a line item about itself. linked_execution has one value, all, so it works as a Boolean for the whole proposal. requires lets a line item name others, so their state reaches it:

• A line item can become non-committable because a line item it names sells out.
• Its visibility follows theirs.
• After agreed, cancelling a named line item invokes the dependent line item's cancellation_policy.
• If A requires B and B requires C, buying A requires C, unnamed.

This PR puts what spans line items one level up, on a package that holds them.

• § 4.1: a proposal holds line_items[] or packages[], not both.
• § 4.1.4: a Package holds line_items[], sold only together, and add_ons[], sold only with all of them once its min_commitments[] are met. A line item sold alone is a package of one, and a proposal sold only whole is one package.
• § 4.1.4: a commit that names part of a package, or an add-on without its package, returns not_committable.
• § 5.6.3: a bonus line item is an add-on, priced on its own.
• § 5.6.2: the holds on a package's line items move as a unit, and § 2.3 counts them once against hold_limits.
• § 4.1.4: a buyer cancels an agreed package's line items only together. A notice on only some of them exercises no option.
• § 2.1 and § 2.3: ProposalSummary lists packages in place of has_requires.
• § 4.1.1 and § 5.8: the binding record and a response keep a line item's package.
• requires and linked_execution are removed.

Related issue

Refs #23 and #6 (OQ-5). Refs #50, #86, #103, and #106.

Builds on

This PR builds on our formats PR, the last of our stack, so applying all 17 in order gives the final text. It replaces the requires and linked_execution of #70 by @Sirajmx, which our PRs carry; our comment on #70 says why. Review the commits "Replace requires and linked_execution with packages", "Count a package's holds once against hold limits", and "Cancel an agreed package's line items only together".

Type

• Editorial — no normative effect
• Normative — changes what the spec requires
• Breaking — would break an implementation already built to this draft
• Examples / schema / docs only

Checklist

• The body section is updated — § 2, § 2.1, § 2.3, § 2.4, § 4.1, § 4.1.1 to § 4.1.4, § 5.6, § 5.6.2, § 5.6.3, § 5.8, § 6.1, § 6.5, § 6.6, and § 8 question 5, with OQ-5 in docs/open-questions.md
• Appendix A (field index) matches the body. linked_execution and requires go, and packages[] comes. scripts/check_appendix.py passes: 96 field names, 98 rows
• Appendix B (fields by marker, and the counts) matches the body — seller-set 46 → 45; 99 → 98 fields; 27 touched by the buyer
• Any affected example in examples/ is updated and still parses. examples/proposal-package.yaml holds the CTV and audio line items of proposal.yaml as a package, with the CTV bonus as its add-on, and the bonus leaves proposal.yaml. manifest.json is regenerated, and neither record_digest changes. scripts/check_examples.py checks one shape per proposal, unique ids, the Package keys, thresholds on the package's own line items, and that a line item in both proposals is the same. All three checks pass
• RFC 2119 keywords used deliberately — MUST / SHOULD / MAY are not synonyms
• The no-inheritance rule (§ 2) still holds — no line item names another, and a package holds only its line items and thresholds
• Anything selectable declares available[] and included[]; anything settable declares bounds (§ 3.1, § 3.2) — no new selectable or settable fields
• No confidential or commercially sensitive content

Knock-on effects

• Breaking. A record with requires or linked_execution no longer conforms.
• Changes from our earlier filings. This replaces our suggestions on § 2, § 4.1, § 5.6 — Line-item dependencies and bonus lines (#23) #70 and [Change] § 2, § 4.1, § 5.6 — Line-item dependencies and bonus lines (linked_execution, requires) #23 that built on requires, such as dropping the cycle ban. any_of has no counterpart: no filing gave a case for it, and an add-on in each of two packages covers a bonus either one unlocks.
• Filings this answers.
  • @premesh's bundles bought whole (#6): full-line transit, market showings, and station dominations. They place per-line terms inside the package on #146.
  • @clerongi's indivisible packages (#50) and cross-media packages ([Comment] § 2, § 5.2, § 5.6 — Support coordinated cross-media campaign packages #86).
  • @bhillman-TelevisaUnivision's must-buy package, with added-value lines and holds that move together.
  • BCN's point, through @matthew-iabtechlab, that binding terms between line items "would be hard to map into publisher order systems" ([Change] § 5.6 — Ordered pricing[].adjustments[] and new cost_method values (per_insertion, per_mm, per_thousand_copies) #103, [Comment] OQ-5, OQ-3 — Print evidence for min_line_items and circulation-based guarantees #106).

Open questions

1. What does cancelling one line item of an agreed package do to the rest? @premesh answered on #6: OOH usually cancels a package whole. § 4.1.4 now cancels its line items only together, each under its own cancellation_policy, since a package holds no terms (question 2). § 8 question 5 asks only whether a package covers every bundle sellers sell.
2. Should a package carry terms its line items share? @premesh suggested one exclusivity term for a domination on [Change] § 5.9, § 5.6, § 2.1 — Add delivery_type: guaranteed, share_of_voice, non_guaranteed #93, then preferred a term on each line item, inside the package, because static and digital faces often take different terms. This PR adds none.

---

By opening this pull request I agree my contribution may be published under this
repository's licence, as set out in CONTRIBUTING.md.

Analyzed and generated with the help of Claude, fully reviewed by me.