Files
archy/Release & Testing/Detailed Plans/11-Mesh-Investigation-and-Acceptance.md

4.6 KiB
Raw Permalink Blame History

Sanitized detailed source transcription S11, captured 2026-10-24. Original planning steps, acceptance criteria and historical evidence are retained rather than condensed. Read current context and precedence first: dated holds, approvals, test counts and “latest” labels below belong to their original checkpoint and do not grant current execution/publication authority. Privacy replacements preserve roles and test semantics. See section coverage.

Mesh public text, sender attribution and icon-only voice replay

Status: user-requested fix, not reproduced. Source inspected read-only; the reported device's actual public messages were NOT accessed. No backend/frontend change made. Requirement C10; T28 investigation followed by T18 protocol/hardware integration.

Observed source

  • neode-ui/src/views/Mesh.vue imports public/channel helpers and renders msg.plaintext directly for ordinary bubbles.
  • senderLabelFor resolves a friendly live name by msg.peer_contact_id, then snapshot peer_name, then Unknown sender. isPlaceholderName handles selected generic names. A channel contact ID is not necessarily the actual sender identity; trace mappings before changing the fallback.
  • Voice attachments already have an <audio controls> path. Source voice recording uses MediaRecorder Opus/WebM. Do not implement a second audio engine or invoke a paid provider just to replay a clip.
  • neode-ui/src/views/mesh/channelMessages.ts, mesh-styles.css, associated tests and core/archipelago/src/mesh/listener/frames.rs are next source entry points. Find the actual driver/frame/parser path and DB fields by tracing production code.

Investigation procedure

  1. Obtain consented read-only access to reported public messages, or ask for 2–3 examples/screenshots. Keep node addresses, keys, identities and raw private traffic in restricted local evidence only. Public-channel text still may contain personal data.
  2. Capture minimal redacted bytes/length/frame type/protocol version plus stored text and rendered output, sender identifier, channel metadata and device-name resolution state. Do not send a test radio message or flash firmware without authorization.
  3. Compare official protocol framing and standard client display. Differentiate binary payload/type/header offset problems, decode-after-decrypt errors, UTF-8 truncation/replacement, mojibake, protocol delimiter handling and genuine Unicode/emoji. Do not remove non-ASCII characters as a "fix".
  4. Add failing fixtures covering real sanitized example, ASCII, accented/multilingual/emoji text, combining characters, invalid UTF-8, truncated/malformed frames, delimiter-containing body, missing/late NodeInfo, channel and DM, local/remote duplicates.
  5. Resolve sender label from trustworthy sender/device metadata; update live when device info arrives, preserve sensible synthetic ID when no name exists. Display names are untrusted labels, not cryptographic verification.
  6. Remove a sender prefix from the visual body only if the parsed protocol representation identifies it reliably. Never split arbitrary name: text or strip a matching word from user-authored content. Preserve raw stored message/history; migration needs independent review and backup.
  7. Clarify requested replay: interpreted as inline voice-message replay, but wording may refer to reading text aloud. For voice: visible audio/speaker replay icon, no visible text, accessible name/tooltip, disabled/loading/error/end states, one player, explicit user click, no autoplay. If TTS is intended, agree local engine/voice/capability/permission and no-cost behavior before adding anything.

Acceptance

  • Reported example renders correctly and retains all legitimate text; root cause evidenced at decode vs display boundary.
  • Correct device name in channel sender header, no duplicate protocol prefix in body; unknown remains honest when unavailable. Names update on late metadata and rename without rewriting history or conflating different senders.
  • No XSS/control-character injection, arbitrary body edits, cross-channel attribution, message duplication or broken DM behavior.
  • Icon-only replay is keyboard/screen-reader accessible, works on supported codecs/phone, handles unavailable/deleted attachments, preserves session/auth and does not auto-spend or start external inference.
  • Source fixtures are not standard-client RF acceptance. Authorized physical bidirectional/community-channel/reconnect cases remain C08/T18 gates.