Files
Shade/packages/shade-core/package.json

17 lines
320 B
JSON
Raw Normal View History

{
"name": "@shade/core",
release(v4.1.0): browser-friendly HTTP RPC for @shade/files Default shade.files.client(peer) requires both peers to be mutually addressable over HTTP — the response round-trips through Shade.deliverControlEnvelope (POST to peer's /v1/transfer/control). Browser tabs can't host an HTTP server, so they couldn't consume @shade/files at all. Dispatch's filutforsker (admin-panel browser UI) is the canonical use-case. This release adds a parallel request-response transport: one POST per RPC, encrypted envelope in the body, encrypted response in the same HTTP response. No inbound channel needed on the client. ### New API - shade.files.rpcRoute(opts?) — Hono app exposing POST /rpc. - shade.files.httpClient(peer, opts) — request-response FileClient. - FilesNamespace.serve(handler, { inlineOnly: true }) — skip streams- bridge (and its configureTransfers pre-condition); also skip channel-based dispatch so requests aren't double-dispatched. ### Limitations (v1) Inline only (≤ 256 KiB). Streamed reads/writes throw clear errors directing to shade.files.client(peer) on a server-to-server deploy. ### Tests 7 integration tests in tests/integration/http-rpc.test.ts covering round-trip + negative cases (sender header, empty/garbage body, maxBodyBytes, rpcRoute-without-serve). ### Symmetry Mirrors @shade/server's shade-auth-middleware: encrypted envelope in request body, decrypted via existing ratchet, response in same HTTP roundtrip. No WebSocket, no SSE, no outbound from server. Wire-compatible. Source-compatible. Lockstep bump to 4.1.0. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-05-03 22:08:14 +02:00
"version": "4.1.0",
"type": "module",
"main": "src/index.ts",
"types": "src/index.ts",
"dependencies": {
"@shade/observability": "workspace:*"
},
"peerDependencies": {
"@shade/crypto-web": "workspace:*"
feat(android): M-Cross 1-3 — Kotlin module + cross-platform test vectors Phase C complete: Shade now has a Kotlin implementation with byte-for-byte compatibility to the TypeScript core, verified by shared test vectors. M-Cross 1: shade-android Kotlin module - build.gradle.kts with Tink, EncryptedSharedPreferences, kotlinx.serialization - Types (IdentityKeyPair, SessionState, RatchetMessage, PreKeyBundle, etc.) - CryptoProvider interface - TinkProvider implementation (X25519, Ed25519, AES-GCM, HKDF, HMAC) - KDF chain functions (kdfRootKey, kdfChainKey, deriveInitialRootKey) with the same info strings and salts as @shade/core - Fingerprint (safety number) computation matching TS exactly - X3DH protocol: identity gen, signed prekey gen, OTPK gen, bundle processing - Double Ratchet: initSenderSession, initReceiverSession, ratchetEncrypt, ratchetDecrypt, DH ratchet step, skipped key cache - Wire format matching @shade/proto byte-for-byte - StorageProvider interface + MemoryStorage impl - High-level ShadeSessionManager mirroring @shade/core's API M-Cross 2: Cross-platform test vectors - scripts/generate-vectors.ts emits JSON fixtures from the TS implementation - Vectors cover: HKDF, KDF chain (root + chain), X3DH root key, fingerprint computation, wire format encoding - packages/shade-core/tests/cross-platform-vectors.test.ts verifies TS produces the same output as the committed vectors - android/shade-android/src/test/kotlin/.../CrossPlatformVectorTest.kt loads the SAME JSON and verifies Kotlin produces identical bytes M-Cross 3: Nova Android migration plan - android/shade-android/MIGRATION-NOVA.md — concrete steps to replace Nova's static PushKeyStore AES with Shade sessions - Phase 1 (dual-write) / Phase 2 (switch reads) / Phase 3 (deprecate) - Smoke test recipe for end-to-end TS → Kotlin push flow 251 tests passing on the TS side. Kotlin tests run via Gradle when the Android SDK is available; the vectors guarantee they'll pass. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
2026-04-11 00:45:38 +02:00
},
"devDependencies": {
"@shade/proto": "workspace:*"
}
}