A review found that an open page kept sending to a DSN after consent was
taken back, that a state file or config file could name where telemetry
goes, and that `BULWARK_TELEMETRY=off` did nothing on the desktop build.
A due heartbeat now waits a random 30 s to 10 min, a refusal from the
collector isn't retried, crash files go at sign-out and account switch,
and server builds parse stacks as they did.
Existing assertions changed in the desktop telemetry tests: the catch-up
wait is now a range, the endpoint comes from the shell, and the platform
is `desktop` against the collector's patched branch.
The timers used to delete everything an account kept, unsent mail included.
They now follow the refused-login rule: sign out and hide, the next sign-in
brings it back, otherwise it goes after 30 days. A login without "Remember
me" that ends with the browser, an admin wipe and a wipe from settings don't
change.
- `wipeOnRefusedLogin` is now `wipeOnForcedSignOut`, since it also covers the
limits. The server's `SESSION_TOO_OLD` goes the same way, as `session`.
- A refused account beside one that stays signed in shows in the switcher as
needing sign-in, with the count and the date, and gets one toast.
Existing tests changed because the rule did: the session timer tests expected
a plain `logoutAll`, and the auth store tests expected no hook at the session
limit.
A refused login used to delete the account's data, and with it everyone's
S/MIME and OpenPGP keys, also with offline off. The usual cause is a password
changed on another device, which Stalwart 0.16 and 1.0 both answer with
`invalid_grant`, so that lost unsent mail for little protection.
- A refused login signs out at once and hides the data. A new password brings
it back, unsent mail included. Otherwise it goes after 30 days. The sign-in
page gives a count and a date. The 24-hour timer is gone.
- `organisation.devices.wipeOnRefusedLogin`, off by default, asks for deletion
at once, of that account's own data only.
- The session limit no longer counts as a refusal on any route. It hides the
data and says nothing.
- A wipe runs under a Web Lock, tells other windows when it starts and ends,
and queues its report before deleting anything. It reports only once the
deletion worked and no plugin database is left.
- More goes with a wipe: Pro tabs, search history, Files and WebDAV state,
templates, pop-out drafts, shown notifications, push state, and Web Push
with the last account.
- The unsent count is per account, and unknown rather than 0 when the store
can't say. A new row is checked against the account that asked, and a
sign-out drops its row.
Existing tests changed because the rule did: `auth-store-keep-local.test.ts`
expected a wipe on `invalid_grant` and at the session limit, and
`device-wipe.test.ts` expected shared keys to go on a refused login.
Until now a lost laptop kept its offline mail for as long as anyone could
open it, and an account whose login the server had ended kept its copy too.
- `lib/devices`, `app/api/devices`: browsers with offline on and desktop
apps register, in a durable collection with no expiry. A check-in goes
by a per-device secret, so a revoked device still hears its wipe.
- `app/api/admin/devices` and the org devices page: wipe one device or all
of a user's, with audit entries for each request and each confirmation.
- `lib/offline/device-wipe.ts`: a start finishes any wipe cut short, then
waits up to a second for the instance before showing the local copy. A
wipe takes the store, keys, plugin databases and handoffs, signs out,
and reports back, again at the next contact if the report failed.
- `lib/jmap/client.ts`, `stores/auth-store.ts`: only `invalid_grant`, the
session limit, or a 401 lasting past a re-auth count as refused. A 500,
a proxy page or an outage never do.
- Settings: "Remove everything from this device", with a confirmation.
`lib/shared-store/__tests__/import-data.test.ts` changes its assertion
because it lists every collection the import copies, and `devices` is new.
The lock is off by default, and an OS policy can require it and set the
timeout. It locks after a time without use counted across all windows and
tabs, on OS screen lock in the desktop app, and from Lock now.
Locking saves open composers, stops the outbox at its next message, closes
the store and stops its worker with every key, then reloads the page, so
nothing of the old page stays in reach. The new page holds the app back at a
Suspense boundary until someone unlocks. Pop-outs get no session while
locked, push notifications name nobody, and subjects leave plain storage.
One random lock key wraps each account's key as a layer. The PIN wraps the
lock key with PBKDF2 at 600,000 iterations, with a wait after wrong tries
that a reload doesn't reset. A passkey with PRF and the desktop's OS storage
wrap it too. Details and limits are in the decisions file.
A persisted account id with no subject yet read as decided: Everyone's
noOffline wiped mail a group scope allowed, and a wider scope went out at
every reload. A public, partial policy counted as the organisation's word,
and turned on the desktop default in a forbidding organisation.
- an organisation's cap on one file now binds a pin
- Clear and the off switch read what would be lost when asked, messages,
changes and drafts, and Clear waits for a connection
- turning one account off survives the last one going off, and off then on
no longer warns of an eviction
- a tab that takes over the store sends no scope until it has heard the policy
- a dropped account is judged by what applies to anyone
- a choice reaches the store once it settles; progress announces phases only
- restored Pro tabs ask for their titles
A forced value could be got around from inside the app. Only a forced
value may now decide what's allowed: the keys that do are forced-only,
so a recommended one no longer acts as forced. With a plugin list a
package can name any id, so uploads are refused and only managed
plugins can match. A server taken off AllowedJmapServers stops its
accounts without signing out: client, push and sync stop, the switcher
drops them, restore and refresh answer 403 with the cookie kept, and
listing it again resumes with no sign-in. OAuth and SSO starts check
the list too.
- `policy.onWipeOfflineData` is gone from the bridge: main runs the wipe
- `KeystoreAllowedOrigins` added, and `POLICY_CONSUMERS` names who acts
on each key or who still owes it
- a Bulwark server's name must be plain after parsing, a wildcard in the
auth lists needs a domain of two labels
Eviction now goes by the base's signal: `storeLost` and the `store-reset`
notice, with `bulwark-store-opened` as the marker and `prune-files` for files
with no key. The storage page and the lost count hang off that. What a user
picks is stored under `bulwark-offline-scope`, the key `offlineScopeFor`
reads, and `email-snapshot` follows the base's rule. `OfflineEnabled` is read
from the desktop bridge when it's there.
The token route now answers 503 when the provider can't be reached, in every build, and a refused renewal for a background account now removes it as evicted, which keeps its data unless the server said invalid_grant. Two assertions from the desktop and pop-out branches change for that. Also drops a duplicate import from the same merge.
Review of the pop-out work found ways to lose text, a wrong sign-out and
changes reaching ordinary browser tabs.
- Drafts: a closing window hands text over only to a main window holding
its login, and asks instead when the last save failed, the browser is
offline, a save is running or plugins could still have a say. A failed
rescue goes to "Continue draft". The opening window lets go only once
the composer is on screen, also for Edit draft in a message window.
Send options travel with the draft. Closing during a send asks first.
- Sessions: answers are encrypted for the window that asked, and only a
window holding the login answers. A renewal queued on the lock can be
called off by a sign-out, signs nobody out once its account is gone, and
takes the cached token after waiting (Rauthy `nbf`, #552).
- Ordinary tabs run none of the host. No lock is held for life.
- A message window hides after a refused session and waits for its
folders; a composer stays on screen with its text. Another tab dropping a
login it couldn't restore no longer closes pop-outs.
- Plain-text mail stays selectable in the desktop app. Folder save renames
reserved and shell names and skips one refused name. All display-mode
checks know the overlay. The attachment bar keeps to one row.
The earlier test that accepted text being gone after a failed rescue now
asserts that it reaches the "Continue draft" slot.
An installed app had no title bar of its own, no way to save attachments
except a zip, and "pop out" stayed inside the page. A new window was a
second copy of the app that had to restore the session itself, and on
start-up that evicts any login made without "Remember me" from the list
every window shares.
- Title bar: `display_override` with `window-controls-overlay`, one
`--titlebar-inset` variable at the page roots and skeletons, and a drag
strip with the window title. 0 without an overlay, so nothing moves.
- Save all to a folder: `showDirectoryPicker`, asking the device policy
before the picker and before each file, with the zip as the fallback.
- Pop-out routes `/popout/message/<id>` and `/popout/compose`, on offer in
an installed app or when the desktop bridge has windows. Hidden in Lite.
- A pop-out borrows the main window's session and never evicts or starts a
token timer. Every renewal now takes a Web Lock, so two windows can't
send the same refresh token, which Stalwart 1.0 may treat as a replay.
- A draft moves to its window under the same id. A closing window hands
unsaved text to the main window to save, since its own request wouldn't
survive the close.
- Read state, drafts, theme and language follow across windows, and
`html[data-desktop]` gates the app-like behaviour in `app/desktop.css`.
Review of the calendar store found writes that went missing, went out
wrong, or appeared twice.
- Writes with different scheduling flags stay separate and go out in
order. Merging them collapsed the two steps of an occurrence change, so
attendees who had answered got no mail, and "don't notify" went out to
everyone with a later RSVP.
- An edit made after a create's lost response is sent as an update. An
import of a uid the server has updates what differs.
- `createCalendarEvent`, `batchCreateCalendarEvents` and
`createCalendarTask` choose a uid before the first send, so a lost
answer online doesn't duplicate.
- The client resolves stand-in ids before the network, which covers an
open dialog once the server is back.
- A stale whole `recurrenceOverrides` map no longer erases overrides.
- A 4xx refuses that write and the queue goes on. `serverFail`,
`rateLimit` or an answer with no verdict retry instead of deleting.
- "This and all following" is refused offline with a clear message.
- A full list only drops what the server says it doesn't have. Sync adopts
a stand-in by uid. Reminder keys survive a stand-in becoming real.
Accounts the session no longer lists are forgotten.
Changed tests: the one asserting a sticky scheduling flag is gone, it
locked in the bug. Reminder keys now use the account and uid. Two flush
assertions drop `sent: 1`, because sync reads a lost write back first. The
two-account id test uses a listed account, as unlisted ones are removed.
Linus Rath