[#446 STEP 2] Tier 1 — idaptik blockers (10 bindings)

[hyperpolymath]

Part of #446. STEP 2 of the top-50 bindings roadmap.

Goal

File 10 sub-issues for Tier 1 bindings (idaptik blockers). Each becomes its own scope-bounded work item that can be picked up independently.

Tier 1 contents

1. PixiJS core (~Application + Container + Sprite minimal)
2. PixiJS sound
3. PixiJS UI
4. Motion (Spine / tween)
5. WASM-exports calling pattern — recommended kickoff; <1 day
6. Phoenix Channels / WebSocket
7. DOM (gated on affinescript#255 codegen bug)
8. WebGL / Canvas2D
9. IPC / postMessage for Gossamer
10. Ajv coverage audit

Acceptance

• 10 sub-issues filed with Refs hyperpolymath/affinescript#446 (or Part of #446 — Tier 1)
• Each carries tier-1 label (label may need creating)
• Each on project board Add Node-target backend + vscode-API bindings to enable .affine VS Code extensions #35
• Top recommendation (WASM-exports calling pattern, item 5) flagged for kickoff

Notes

The DOM binding (item 7) is blocked on affinescript#255 (wasm-codegen for-in/while bug); that's tracked separately.

[hyperpolymath]

Tier 1 #10 — Ajv coverage audit (analysis-only)

stdlib/Ajv.affine ships 2 extern types + 4 extern fns (16 lines). Audit against AJV v8 + ajv-formats v3 mainline surface.

Currently bound

AS extern	AJV v8 binding
ajv_new()	new Ajv({allErrors:true, strict:false}) + addFormats(ajv)
ajv_compile(a, schema)	ajv.compile(JSON.parse(schema))
ajv_validate(v, value)	validator(JSON.parse(value)) // Bool
ajv_errors_json(v)	JSON.stringify(validator.errors)

Sufficient for the named consumer

The row note flags idaptik UMS pipeline (level config validation) as the canonical consumer. That pipeline shape is:

init   → ajv_new + ajv_compile(level_schema)
runtime → for each level: ajv_validate + (if !ok) ajv_errors_json

The 4-extern surface covers this 100%. Treating row #10 status ◑ as "consumer-complete-but-narrow" rather than "broken-incomplete".

Gaps surfaced for broader estate use

For the wider estate use-cases mentioned in the row note (a2ml + schema work), the missing surface clusters around three axes:

Multi-schema registry — for repos validating many schemas (hyperpolymath/standards, multi-doc workflow_audit, the a2ml suite):

• ajv.addSchema(schema, key) — register named schema
• ajv.getSchema(key) — retrieve compiled validator by key
• ajv.removeSchema(key) — remove named schema
• ajv.validate(schemaOrRef, data) — one-shot validate against a registered schema by ref

Custom format / keyword extension — for a2ml's domain-specific format identifiers (a2ml:semver, a2ml:spdx-id, etc.) and the AI-MANIFEST shape:

• ajv.addFormat(name, regex | fn) — register custom format
• ajv.addKeyword({keyword, ...}) — register custom keyword

Async compile (for $ref resolution) — for distributed schemas with HTTP-loaded $refs:

• ajv.compileAsync(schema) — Promise

Human-readable errors:

• ajv.errorsText(errors, opts) — formatted error string (versus the JSON dump from ajv_errors_json)

Estimate (if/when broader use lands)

• Multi-schema registry: 4 externs + 4 codegen entries
• Custom format/keyword: 2 externs (registering native AS lambdas as format/keyword callbacks needs the same opaque-Json-handler pattern as pixiContainerOn / messagePortOnMessage)
• Async compile: 1 extern with / { Async } effect row
• errorsText: 1 extern

That's ~8 additional externs — modest, ship-when-needed. The current ◑ partial is accurate; the question for any future expansion is which consumer surfaces the need (drive by demand, not by pre-emptive coverage).

Recommendation

• Leave row ci: Bump erlef/setup-beam from 1.17.5 to 1.24.0 #10 at ◑ partial with the audit-result note that the consumer-complete-but-narrow state is intentional.
• Open expansion sub-issues only when a specific estate consumer surfaces a need (a2ml custom keyword work would be the natural trigger).

Refs #446 (Tier 1 umbrella), stdlib/Ajv.affine.