Skip to content

Signal: read-receipt and typing settings can be changed here (local scope) - #897

Open
unRekable wants to merge 1 commit into
mainfrom
draft/738-signal-privacy-settings
Open

unRekable wants to merge 1 commit into
mainfrom
draft/738-signal-privacy-settings

Conversation

@unRekable

Copy link
Copy Markdown
Collaborator

What changed and why

Signal's Settings › Privacy rows no longer say "Set on your phone": typing and read receipts get real per-account switches that take effect locally. Off stops tinyclient from sending the typing signal or the receipt frame; on defers to the phone's setting, which still decides above the switch — nothing is ever written to the phone, and nothing stored can loosen what the account keeps off.

Type

  • feature
  • fix
  • perf
  • refactor
  • chore
  • docs

Platforms

  • windows
  • macos
  • linux
  • android
  • ios

Proof

Scenario: e2e/scenarios/signal-privacy-local

Lab run:

Compatibility

No wire or protocol change. The settings table grows by rows it already knows how to store: typing and receipts in the existing account:signal:<aci> scope, values on/off, read by the same Values::typing / Values::receipts readers Teams and Discord already use. Older builds ignore the rows (unknown key in a scope they never read), and a build without this change keeps working against a store that holds them. Defaults are unchanged, so a person who never touches the switches sees exactly today's behavior.

Scope honesty: this is local-only. The switches cap what this computer sends; the account's real read-receipt and typing settings stay owned by the phone, are still learned only from its configuration sync, and are still not shown or changeable here. Changing the account is #745's storage-service work, and showing the account's actual state ("On · set on your phone") is deliberately not attempted in this patch — the UI copy says "Your phone's setting decides too", and it does.

Risk and rollback

The adapter's own gate (phone-reported settings, unknown = off) is untouched, so the new switches can only make tinyclient send less, never more; a wrong turn cannot send receipts when the account has them off. One real behavioral note: receipts off on Signal also skips the read sync to the account's other devices, the same way Teams' switch has always worked — the row's second line says so. Rollback is a plain revert; a stored off row left behind is inert for older builds and honored again once the revert lands.

Release note

Signal: typing and read receipts can now be turned off per account in Settings › Privacy — a local stop only; your phone's setting still decides.

@unRekable unRekable left a comment

Copy link
Copy Markdown
Collaborator Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Code Review

🟢 No findings.

The diff turns Signal's fixed "Set on your phone" P1/P2 rows into local stop switches: settings validation takes an off value in a Signal account scope, and because chatd's mark_read, mark_all_read and Typing paths already gate on settings().receipts()/typing() per account, the switch takes effect with no adapter change. The phone-reported setting is untouched in the adapter and still decides above the switch, defaults are unchanged (nothing stored means on), and the UI captions say honestly that this stops only tinyclient. The schema change is pinned by unit tests and a through-the-socket integration test whose Set request main refuses, and the e2e driver follows the established signal-own-profile prove pattern, failing on main by timeout. I found nothing wrong: no wire change, no data loosened, no new host, no blocking call.

Residual risk

  • A downgrade to a build before this change refuses the stored Signal off rows at read time, so the local stop silently disappears and sending returns to the phone's level; the user gets no hint.
  • Reads made while receipts are off are flushed to the account's other devices only if a later read happens in the same chatd session; after a restart the in-memory unread queue is gone and that gap never syncs.
  • Signal's P2 off now skips the read sync to own devices, which is stricter than the same switch's Teams semantics; future code that mirrors P2 handling across networks could assume the two behave alike.

@unRekable
unRekable marked this pull request as ready for review November 18, 2026 16:13
@unRekable
unRekable requested a review from itz4blitz as a code owner November 18, 2026 16:13
@unRekable

Copy link
Copy Markdown
Collaborator Author

Note on the red ios (libs + unsigned archive) check: the job never ran — the runner annotations say The job was not acquired by Runner of type hosted even after multiple attempts / Due to capacity constraints, jobs targeting macOS arm64 runners may experience longer queue times. That is the documented hosted-runner capacity problem (docs/quality-gate.md records the same on test (macos-latest)), not this diff: every check that actually executed is green (all ubuntu shards, android, doc tests + mobile check, pr-gate), the review found no defects, and the iOS build compiles the same crates it did before this change (the diff touches desktop integration, packaging assets and a mobile-gated module — nothing iOS links differently).

@github-actions

github-actions Bot commented Nov 18, 2026 •

Copy link
Copy Markdown

Tiny Code Review

⚠️ Code review unavailable · run

the AI provider rate-limited the request

This is advice, not a gate. Re-run with the agent:review label or a /review comment.

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