Drafts were always created in the primary account, so a message composed as
a group identity sat in the member's own Drafts while its Sent copy (since the
previous commit) went to the group.
- createDraft takes { accountId, previousDraftAccountId }: the draft goes to
the account that owns the From identity (resolveSendAccountId, now public
and briefly caching the identity lists for autosave), and the replaced
version is destroyed in the account it lives in.
- The composer tracks draftAccountId next to draftId (save, re-resolve of
attachment blobs, send, discard, stashed state), since draft ids are only
unique within one account.
- sendEmail destroys the sent draft in options.draftAccountId; discarding
(inline and Pro tab) deletes it there, through the identity's login.
- Re-opening a draft from a group/shared Drafts folder (or a stamped
aggregate row) keeps its account.
- resolveSendAccountId is optional on IJMAPClient; without it drafts stay in
the primary account.
- Tests: client unit tests for draft placement / replacement / send cleanup,
E2E spec 18 draft cases (autosave + send, discard).
The demo user becomes Nora Lindqvist, engineering lead at Meridian Labs,
in the week of a product launch. The fixtures tell one story across mail,
contacts, calendar, tasks and files, and exercise the features the
screenshots and the public demo should show:
- a second, mail-only login (her personal domain) and the shared Support
mailbox delegated to the work account, so the account switcher, the
shared section and the merged inbox have real content
- raw headers the way Stalwart writes them: the sender check fails on
the phishing mail, newsletters carry a one-click unsubscribe, one mail
asks for a read receipt, and spam scores show in the details
- an iMIP invitation with a served .ics, an event waiting for the answer
on the calendar, and RSVP patches applied in the demo client
- @mentions as mailto links in a received mail, a verification code
twelve minutes old, nested folders, named tags on the mail, contact
groups, a scheduled message and tasks switched on
- every attachment and file backed by real bytes under public/demo,
generated by `scripts/demo-assets/generate.mjs`; contact photos are
bundled portraits instead of requests to randomuser.me
- folder counts computed from the mail instead of hand-kept numbers
Demo login seeds the tags, templates and merged-inbox settings once per
device and switches the cross-account gate on for the session; "Reset"
restores them. `bulwark-demo-simulate=off` in localStorage stops the
simulated incoming mail for screenshot runs.
Stalwart's default folder set has no archive-role mailbox, so on a fresh
account Archive did nothing but raise the not-found toast (#578). Both
archive paths now go through ensureArchiveMailbox(), which creates
"Archive" with the archive role in the account the action targets (own,
the selected shared folder's owner, or the pinned account in unified
view), reloads the mailbox list and continues with the store's own
object. createMailbox() takes an optional role for this. A refused
creation or a folder that does not come back still surfaces as an error.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Attachment thumbnails were object URLs typed with the sender's MIME
type. Object URLs share the webmail origin, so an `image/svg+xml`
attachment opened from its thumbnail ("Open image in new tab", or
dragged to the tab strip) ran the sender's script as the webmail in
Chromium and Firefox. The drag-out URL and Files thumbnails had the
same shape.
`fetchBlobAsObjectUrl` now re-types every blob to an inert type, and
image thumbnails in mail and Files go through `imageBlobUrl`, which
hands an SVG over as a `data:` URL: it still renders in the `<img>`,
but opened on its own it gets an opaque origin.
Someone who writes to a calendar shared with them (read & write) changes
the owner's meeting there on the owner's behalf: Stalwart sends the change
from the addresses of the account the event is in. Since 6d4b6954 Bulwark
decided who organizes by the user's own addresses only, so it took the
sharee for an attendee (live on 1.0 and 0.16.23):
- a drag, a resize or a month-view drop wrote the new time without
scheduling messages: the attendees kept the old time and their answers
while the owner's copy moved; before 6d4b6954 they got "Updated
invitation"
- the editor wrote no participants, hid "Send invitations", showed the
attendees read-only and sent nothing
- "All events" asked nobody again, and "This and following events" ended
the series quietly and made the rest a new series without attendees
`userOrganizes` now also counts the organizer's own copy in another
account: `isOrigin`, which both versions compute from the addresses of
the account the event is in, true for the owner's meeting and every
expanded occurrence of it, false for an invitation the owner got. Such an
invitation is still changed without messages. The editor no longer drops
the user's own address from the attendees it opens with, only the
organizer's, so a sharee who attends the owner's meeting stays on it.
The demo's invitation from Alice now says it is not the demo user's copy.
Follow-ups to the review of the push, keyword and identity work:
- A send that names no sender, like the List-Unsubscribe mailto: of the
unified view, fell back to the first identity in server order, which
on Stalwart 1.0 is a group's. It now takes the login's own identity,
with addresses compared regardless of case.
- Re-register creates the replacement subscription before it destroys
the old one, so at Stalwart's limit (15 per account by default) it
failed with `overQuota`. It now frees the old record's place and
tries once more.
- On 1.0 the push settings note goes by the emailpush capability alone
and ignores refusals remembered under 0.16. On 0.16 a group the user
belongs to is recognised by its identity, so a member without
calendars, contacts or files keeps the group's junk filter and is no
longer told the group sends no notifications.
- Plugin imports can no longer set Stalwart 1.0's attachment keywords,
and `keywords.discover` leaves them out.
- Batch mark-as-read showed messages the server refused as read. It now
updates only what was written and reports the rest, and thread
expansion and category-tab moves keep what a partly refused batch
did write.
- Refused read and star changes from shortcuts, the list, the viewer
and Pro tabs show "Failed to update email" instead of an unhandled
rejection. The automatic mark-as-read on opening stays quiet.
- A tag whose id predates the 128-byte limit can have its colour
changed again.
- The state-regression guard notes that a multi-node 1.0 cluster can
trip it harmlessly, at the cost of one extra full fetch.
Stalwart 1.0 reads a text filter value that starts and ends with a
quote as one exact phrase, and a value ending in `*` as one prefix over
the whole string, not split into words and capped at 16 bytes. Every
search path sent the typed input as one `text` value, so on 1.0
`"annual report" 2024 "draft"` became a single four-word phrase and
`project budg*` a prefix containing a space: both found nothing.
`"Q3 planning" budget` lost its phrase on every version.
`splitSearchText` (`lib/jmap/search-utils.ts`) now sends each quoted
phrase and each `word*` as a condition of its own, ANDed with the rest.
The remaining words stay one condition, so a query without quotes or
stars goes out as before. Outer single quotes are stripped from that
part, since 1.0 reads them as a phrase too. The split applies to:
- the search bar, tag views, cross-account and all-folders search;
- advanced search, including the from/to/subject/body fields;
- contact search;
- global search for mail, contacts and calendar.
`inMailbox` and the calendar bounds sit next to the text conditions in
one AND, with one key per condition: Stalwart rejects an operator that
also carries property keys, and 1.0 allows only one AND group when it
expands recurrences.
SearchSnippet/get switches the whole request to exact matching once a
term is quoted, which marks nothing for a phrase ANDed with other
words, so the snippet filter unquotes phrases.
On 1.0 and on external search backends each phrase is now a real
phrase. The 0.16 built-in index reads it as an AND of its words,
without stemming. The dev mock and the demo client now evaluate AND
filters and quoted phrases, so they keep working with the new shapes.
Stalwart refuses a push subscription's whole emailPush map as
`forbidden` when it names a mailbox shared by ACL (0.16.16 and later,
and 1.0). Bulwark put every account with mailboxes in the map, so
enabling push showed raw JSON, and the background re-sync destroyed the
working subscription after its refresh was refused and dropped its key:
push stopped without a word and could not be turned back on while the
share existed. A network error, a timeout or rate limiting during the
refresh ended it the same way.
- The map holds the primary account and the accounts that list
`urn:ietf:params:jmap:emailpush`, which 1.0 leaves off ACL shares.
0.16 lists it everywhere, so a refused map narrows to the accounts
that look owned, then to the primary account, as the mobile app does.
Refused accounts are remembered under
`bulwark.push.emailPushRefused.v1.<id>`; Re-register and disable
clear the list.
- A subscription ends only when a verified replacement exists, the user
turns push off, or the server no longer has it. The old record is
destroyed after the new one is verified. A failed listing, a network
error or a refused method call leave it and its key alone.
- PushSubscription/set failures throw typed errors (`jmapSetError` for
a refused object, `jmapError` for a refused call) instead of JSON in
the message or `false`, and the client exposes `getAccountInfo`.
- Settings says that mailboxes others share don't send background
notifications, and `forbidden` and `overQuota` read as plain text in
settings and the onboarding prompt.
Stalwart 1.0 counts every EmailSubmission record of an account against
`maxSubmissions` (500 by default), whatever its status or age, and its
own purge of old records never runs. Bulwark never destroyed a
submission, so after about 500 JMAP sends an account could not send at
all: every send, read receipt and reschedule failed with a raw "There
are too many email submissions" error, and each failed attempt left one
more copy of the message in Drafts.
Bulwark now removes the records that can no longer be cancelled.
Destroying a record never stops its delivery, but a pending one could
no longer be cancelled, so pending records are never touched:
- an immediate send (no hold, no undo window) destroys its own record
right after the send, riding along with the old-draft cleanup; only
where `urn:stalwart:jmap` is advertised, and never decided by the
created `undoStatus`, which 0.16.10-0.16.16 report as final for held
sends
- a throttled background prune destroys canceled records and final
records whose `sendAt` has passed, going by `EmailSubmission/get`, in
every submission account; it runs at startup, after sends, after
undo and on EmailSubmission state changes
- a confirmed cancel (undo, cancel, reschedule) destroys the canceled
record in a request of its own
- an `overQuota` on a submission create prunes at once, on Stalwart
walks the change log for records older than the query's three-day
window, retries once, and otherwise throws `SubmissionQuotaError`,
shown as a translated explanation in the composer, quick reply, read
receipts and the scheduled-send actions
Failed sends now clean up after themselves and say so:
- a failed send removes the Drafts copy it created and keeps the
user's draft; the raw send paths remove their import the same way
- a read receipt that did not go out is taken back out of Sent, a
method-level error counts as a failure, `$MDNSent` is only set after
a sent receipt, and the automatic receipt no longer retries on every
refresh of the open message
- quick-reply failures, "Send now", "Reschedule" and a failed undo are
shown instead of only logged
- the scheduled list reads every page of `EmailSubmission/query`; it
stopped after the first because `total` needs `calculateTotal`
Recurring events defaulted skip:"omit"+rscale:"gregorian" on every rule,
which Stalwart serialises into the CalDAV RRULE as RSCALE=GREGORIAN;SKIP=OMIT.
DAVx5 rejects that ("Invalid recurrence rule"), breaking Android sync. Those
RFC 7529 fields only matter for non-Gregorian scales / invalid-date handling,
so stop emitting them for plain Gregorian rules.
- lib/recurrence-rule.ts: createRecurrenceRule() helper with no skip/rscale.
- CalendarRecurrenceRule.rscale/skip are now optional.
- event-modal + recurrence-editor build rules via the helper.
- demo fixture cleaned; unit test asserts no skip/rscale.
Closes#805.
A PatchObject key is a JSON Pointer (RFC 8620 section 5.3), so a slash
inside the member name it points at has to be escaped as `~1`. Keyword
patches were built as `keywords/${keyword}` verbatim, which is fine for
every flat keyword but wrong for a nested tag: `$label:work/clients`
became the pointer `keywords/$label:work/clients`, asking for the
`clients` member of the `$label:work` keyword rather than for the keyword
itself. The patch is not a boolean at that path, so the server rejects or
ignores it and the tag never lands.
Route the keyword patch sites - setKeyword, removeKeyword, migrateKeyword
and the category-tab batch patch - through `keywordPointer`, which escapes
per RFC 6901 the way the calendar store already does for participant ids.
Flat keywords are untouched, so `$seen`/`$flagged` and friends keep the
exact pointers they had. The dev JMAP mock reads the escaping back, or
nested tags would behave differently against the mock than against a real
server.
migrateKeyword also took any answer as success: a method-level error on
the query read as "no message carries this keyword", one on the write was
ignored outright, and a per-message refusal was invisible - so a rename
that moved nothing still reported a clean migration. It now answers with
{ migrated, refused }, because the work is batched and a failure partway
through cannot undo the batches already applied. A whole batch the server
refuses stops the rest, which it would refuse too; a single message it
refuses does not, the mail behind it being unaffected. A query error
throws, nothing having been written yet, and so does a run that migrated
nothing at all - the case where the caller must not rename a tag over
messages that all kept the old keyword.
The walk over tagged mail was not paginating safely either. It stopped at
the first page shorter than the limit it asked for, but RFC 8620 section
5.5 lets a server clamp that limit, so a clamping server left most of the
mail unmigrated. Positions are also only comparable within one queryState,
which the walk ignored: mail arriving mid-walk shifts every later page. It
now restarts when the state changes under it, and gives up - before any
write, so the caller can simply repeat - once a mailbox has moved under it
three times, rather than mixing ids from states that no longer agree. A
page carrying no id it has not already seen ends the walk, which keeps it
bounded against a server that answers every position with the same page.
Folder management performed on a shared/group account was sent to the
logged-in user's primary JMAP account instead of the mailbox owner's
account. Subfolder creation therefore failed with invalidProperties for
parentId, root folders could be created silently in the personal account,
and rename, delete, role, and reorder targeted the wrong mailbox tree.
Shared mailboxes use namespaced store ids to avoid cross-account
collisions, while Mailbox/set requires the owner's accountId and the bare
server mailbox id. Resolve that mutation context centrally: recover the
mailbox originalId and owner account, use a directly connected owner
client when one exists, otherwise reach delegated/group accounts through
the active client with an explicit accountId.
Thread the account target through create, update, and delete JMAP methods
and all folder-management actions. Scope role replacement to the same
owner account, and expose explicit root-folder creation from shared
account headers without making personal root creation depend on the
currently selected folder.
Add unit coverage for JMAP request scoping, shared store routing, role
isolation, direct-client handling, and reorder. Add a Stalwart integration
scenario that creates root and nested group folders, renames them, and
deletes them through the UI.
If you have a large number of tags, getTagCounts would not be able to get the
unread counts because it did not respect maxCallsInRequest, even though the
value was actually read out, it was just ignored. There are also other places
where the limits were not respected.
Batching is now generalized in a helper that also takes maxObjectsInSet, which
also was ignored, into account and is applied to all functions.
In addition, the dev mock now also advertises and enforces the limits, so these
issues can get picked up during development.
Possible closes#699
Possibly closes#399
Moving a message to a folder in another account (own ↔ delegated/shared) via the
"Move to" context menu was a no-op — the handlers always issued a single-account
Email/set, which can't move between JMAP accounts. Drag-and-drop already routed
these correctly; the context menu never did.
Add moveToMailboxCrossAware: it detects a cross-account destination (own and
shared mailboxes both carry accountId) and routes through the drag-and-drop
crossAccountMoveEmails pipeline, else falls back to the single-account move.
Fix the pipeline for delegated folders too: a client can't stage a blob in a
delegated account (Blob/upload → blobNotFound), so importing into a shared folder
failed. When one client reaches both accounts, use a server-side JMAP Email/copy
(+ destroy original) instead of blob copy+import; the blob path is kept only for
separate cross-server login accounts. Adds client.copyEmailAcrossAccounts.
Unit tests for the dispatch; the two 08-shared-moves specs are un-pinned. Full
docker integration suite green (37 passed).
Linus Rath
Linus Rath
Stefan Hildebrandt
sudandClaude Fable 5.1
KazNIISA IT
Ctaoist
Mathy Vanvoorden