Phase 4: sync

Implements PLAN.md Phase 4: the read model. /sync, /messages and /context,
the filters that narrow them, and the streams a response carries that are
not room events — typing, receipts, presence and to-device messages. With
this a client can be used against Warp, which is milestone M1.

The one architectural choice is that the syncapi has no copy of the rooms:
it reads them through the roomserver. Dendrite gives its syncapi its own
tables and feeds them from the output stream, which is right for a design
where the two can be separate processes, and Dendrite's could be. Warp is
one process by decision, and there the copy buys nothing and costs the
hardest class of bug a homeserver has — a read model that disagrees with
the room, where a client is shown a message that was rejected and the only
way to find out is to compare two tables. So the roomserver grew a read
model (stream.go, roomsync.go, visibility.go) rather than exporting its
store: reading an event means parsing it under its room's version, and a
second copy of the version registry, the parser and the client projection
would be a second place for them to disagree.

syncapi: stream tokens naming a position in each of five independent
streams, so a client that has seen receipt 7 but only event 4 can say so in
one opaque string. A token with fewer positions than this build writes is
accepted and the rest read as zero, which is what lets a token stored
before a stream existed go on working. The PDU position comes from
MAX(stream_ordering) and not from the counter row the other four use: the
counter is advanced inside the writing transaction, so a token taken from
it would name a position whose event no reader can see yet, and the event
would be skipped for good once that transaction committed.

Long polling goes through a notifier that wakes by room and by user rather
than globally — an event wakes the listeners of its room, a membership
event also wakes the user it names, so an invite reaches a client that is
not in the room yet. A wake-up is only ever a hint: the listener re-reads
the streams and decides for itself what changed, which is what makes it
safe for the notifier to hold no state about what each client has seen. The
listener registers before the read, so an event landing in between leaves
the channel signalled and the sync reads again rather than blocking on a
change it just missed. EventPersisted must not block, because it is called
with the room's write lock still held.

"Something happened" is not "a token moved". A stream can move without this
user being concerned by it, and returning on every such move would turn a
long poll back into a busy loop. The same trap caught presence twice over:
a client says it is still online by syncing, so ?set_presence= makes every
sync a presence write, and a stream that moved each time would wake every
peer, which would sync, which would move it again. SetPresence therefore
reports whether anything a peer would notice actually changed, and a write
that only refreshes the last-active timestamp leaves the stream alone.

Timeline windows are named by their two ends and a direction, Low exclusive
and High inclusive in both directions, rather than by "from" and "to" as
the wire format spells them. Each of those would otherwise be the low bound
in one direction and the high bound in the other, which is exactly the
off-by-one every pagination bug is made of; the handlers translate once and
nothing below them thinks about it again. A sync's timeline is read
backwards and reported forwards, because a client wants the newest events
of the window it missed — a forward read with a limit hands it the oldest
and then calls the rest a gap.

visibility: history visibility is now a span rather than a yes or no, which
is what Phase 3 deferred here. A user who has left a room is served the
room as it was when they left — their state reads, their members and their
timeline — instead of being refused, and is bounded above at their own
membership event so a kicked client cannot go on streaming. Their span is
found from their own member events, so a user who joined, left and rejoined
sees their latest span. ?at= on /members is honoured for the same reason.
What is left is that the span is computed from the room's current
history_visibility where the spec judges each event by the one in force
when it was sent; that is a state read per event and belongs with open
question 1.

Typing is held in memory and nowhere else: notifications expire in seconds,
are worthless across a restart, and arrive one per keystroke-burst per
room, so persisting them would be the busiest table in the database holding
the least valuable data in it. Receipts only ever advance, compared on the
position of the event acknowledged rather than on arrival, because two of a
user's clients read at different speeds and the slower one must not drag
the marker backwards. To-device queues are per device and drained by the
next sync rather than by the one that delivered them, so a client that
crashes in between gets the batch again — the right way round, since a lost
key share leaves a message permanently undecryptable and a duplicate is
ignored. A wildcard addressee is expanded when sent, not stored, because
"every device" means the ones that existed then.

Presence is off by default, which answers PLAN.md open question 3. It is
the noisiest thing in Matrix and the one genuinely fan-out notification
here; a private homeserver gains little from it. With it off the endpoints
refuse rather than disappearing, because a client that gets a 404 from
/presence concludes the server is too old to have it and stops asking.

storage: one migration per driver adding the receipt, to-device, presence
and filter tables with a counter row each, and a stream position on the
existing account data table rather than a parallel one — there is exactly
one current value per (user, room, type), and a second table would be a
second thing to keep in step. Account data is reported in an event list but
is not an event, so it gets its own response type: a room event with an
empty sender and no timestamp would make a client decide what that meant.

Deliberately missing, each named in docs/sync.md: notification counts (they
need push rules, Phase 8), device lists (the keyserver, Phase 7), anything
remote (Phase 5), threads (Phase 10). device_lists is sent as two empty
lists because a client reads a missing one as malformed; the notification
counts are left out entirely, because a zero there is a claim and an absent
field is not.

Tested through HTTP rather than against the builder, because what is worth
pinning down is the response a client sees — which section a room lands in,
whether a token handed back can be handed in again, whether paging
terminates and loses nothing. The storage tests run against both SQL
drivers; the notifier has its own tests for the concurrent case. A live
server was driven end to end, which is what caught the two bugs the unit
tests had missed: a backwards /messages page must be in reverse
chronological order, and `start` must name where the page began rather than
position zero when ?from= was omitted.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>

server/internal/syncapi/syncapi.go

@@ -0,0 +1,216 @@
+// Package syncapi is the read model: it answers "what has happened that I
+// have not seen", which is the question a Matrix client spends its whole
+// life asking.
+//
+// Everything a client learns after it has logged in arrives through here —
+// /sync for the live stream, /messages for a room's history, /context for
+// one event and what surrounds it. The component writes almost nothing:
+// read receipts, to-device messages and presence are its own, and
+// everything else it serves is the roomserver's, read through the roomserver
+// rather than copied. That is the one real architectural choice in this
+// package, and it is a departure from Dendrite, which gives its syncapi its
+// own tables and feeds them from the output stream. In one process the copy
+// buys nothing and costs the hardest bug in a homeserver: a read model that
+// disagrees with the room.
+//
+// What this package owns instead is everything the roomserver has no
+// opinion about:
+//
+//   - Stream tokens: a position in each of the five streams a response
+//     carries, in one opaque string (token.go).
+//   - The notifier: waking a client blocked in a long poll the moment
+//     something concerns it, rather than when its timeout expires
+//     (notifier.go).
+//   - Filters: what a client asked to be left out (filter.go).
+//   - The ephemeral and device streams: typing, receipts, presence and
+//     to-device messages, none of which are events in a room.
+//
+// Phase 4 of PLAN.md. What it does not do yet is notification counts, which
+// need push rules (Phase 8), and device list tracking, which needs the
+// keyserver (Phase 7); both are named in docs/sync.md as the gaps they are.
+package syncapi
+
+import (
+	"context"
+	"errors"
+	"log/slog"
+	"time"
+
+	"github.com/dariuszjaniak/warp/server/internal/roomserver"
+	"github.com/dariuszjaniak/warp/server/internal/storage"
+)
+
+// Errors callers have to tell apart. The storage and roomserver sentinels
+// are re-exported rather than translated, so that errors.Is works whichever
+// package a caller reaches for.
+var (
+	ErrNoSuchRoom  = storage.ErrNoSuchRoom
+	ErrNoSuchEvent = storage.ErrNoSuchEvent
+
+	// ErrForbidden is a read the caller is not entitled to: a room they are
+	// not in, or somebody else's filter.
+	ErrForbidden = roomserver.ErrForbidden
+
+	// ErrPresenceDisabled is a presence operation on a server that has
+	// presence switched off.
+	ErrPresenceDisabled = errors.New("presence is disabled on this server")
+
+	// ErrInvalidParam is a request parameter this server will not accept:
+	// a receipt type it does not know, a presence state that is not one.
+	ErrInvalidParam = errors.New("invalid parameter")
+
+	// ErrInvalidToken is a ?since= or ?from= that is not a sync token.
+	ErrInvalidToken = errors.New("invalid sync token")
+)
+
+// Accounts is the part of the userapi the sync API needs: account data to
+// put in a response, and the devices of a user to address a to-device
+// message to. *userapi.API satisfies it as it stands.
+type Accounts interface {
+	AccountData(ctx context.Context, localpart, roomID, dataType string) ([]byte, error)
+	SetAccountData(ctx context.Context, localpart, roomID, dataType string, content []byte) error
+	Devices(ctx context.Context, localpart string) ([]storage.Device, error)
+}
+
+// API is the syncapi component.
+type API struct {
+	serverName string
+	rooms      *roomserver.API
+	accounts   Accounts
+	store      storage.SyncStore
+	notifier   *Notifier
+	typing     *typingTracker
+	log        *slog.Logger
+	now        func() time.Time
+
+	// presenceEnabled is the config switch of PLAN.md open question 3.
+	// With presence off the endpoints answer M_FORBIDDEN and the presence
+	// section of a sync is empty, rather than the endpoints disappearing:
+	// a client that finds no /presence assumes an old server and stops
+	// asking, where a clear refusal tells it this one has it turned off.
+	presenceEnabled bool
+
+	// longPollTimeout caps how long a /sync may block, whatever ?timeout=
+	// the client asked for. A client asking for ten minutes would hold a
+	// connection and a database handle for ten minutes.
+	longPollTimeout time.Duration
+}
+
+// Options configure the component.
+type Options struct {
+	// ServerName is the Matrix server name.
+	ServerName string
+
+	// Rooms is the roomserver, which every room read goes through.
+	Rooms *roomserver.API
+
+	// Accounts is the userapi.
+	Accounts Accounts
+
+	// Store is the sync half of storage.
+	Store storage.SyncStore
+
+	// Notifier wakes blocked syncs. It is the same object the roomserver
+	// was given as its Notifier; nil builds one, which is only useful in a
+	// test where nothing is being written.
+	Notifier *Notifier
+
+	// PresenceEnabled turns the presence stream on.
+	PresenceEnabled bool
+
+	// TypingTimeout is how long a typing notification lasts without being
+	// renewed, whatever timeout the client asked for.
+	TypingTimeout time.Duration
+
+	// LongPollTimeout caps a /sync's ?timeout=.
+	LongPollTimeout time.Duration
+
+	Logger *slog.Logger
+
+	// Now is the clock, so tests can expire a typing notification without
+	// waiting for it.
+	Now func() time.Time
+}
+
+// Defaults for the timeouts.
+const (
+	// DefaultTypingTimeout is the spec's own suggestion, and what every
+	// client sends.
+	DefaultTypingTimeout = 30 * time.Second
+
+	// DefaultLongPollTimeout is the timeout clients ask for. Holding a
+	// connection longer buys nothing: the notifier wakes a sync the moment
+	// anything happens, so the timeout only decides how often an idle
+	// client re-establishes its request.
+	DefaultLongPollTimeout = 30 * time.Second
+
+	// MaxTypingTimeout caps what a client may ask for. A typing
+	// notification that outlived its sender's session would leave a room
+	// permanently claiming somebody was typing.
+	MaxTypingTimeout = 2 * time.Minute
+)
+
+// New builds the component.
+func New(opts Options) (*API, error) {
+	switch {
+	case opts.ServerName == "":
+		return nil, errors.New("syncapi: ServerName must not be empty")
+	case opts.Rooms == nil:
+		return nil, errors.New("syncapi: Rooms must not be nil")
+	case opts.Store == nil:
+		return nil, errors.New("syncapi: Store must not be nil")
+	case opts.Accounts == nil:
+		return nil, errors.New("syncapi: Accounts must not be nil")
+	}
+
+	if opts.Notifier == nil {
+		opts.Notifier = NewNotifier()
+	}
+	if opts.TypingTimeout <= 0 {
+		opts.TypingTimeout = DefaultTypingTimeout
+	}
+	if opts.TypingTimeout > MaxTypingTimeout {
+		opts.TypingTimeout = MaxTypingTimeout
+	}
+	if opts.LongPollTimeout <= 0 {
+		opts.LongPollTimeout = DefaultLongPollTimeout
+	}
+	if opts.Logger == nil {
+		opts.Logger = slog.Default()
+	}
+	if opts.Now == nil {
+		opts.Now = time.Now
+	}
+
+	return &API{
+		serverName:      opts.ServerName,
+		rooms:           opts.Rooms,
+		accounts:        opts.Accounts,
+		store:           opts.Store,
+		notifier:        opts.Notifier,
+		typing:          newTypingTracker(opts.TypingTimeout, opts.Notifier, opts.Now),
+		log:             opts.Logger,
+		now:             opts.Now,
+		presenceEnabled: opts.PresenceEnabled,
+		longPollTimeout: opts.LongPollTimeout,
+	}, nil
+}
+
+// Notifier returns the notifier, which the roomserver is wired to as its
+// emit stage.
+func (s *API) Notifier() *Notifier { return s.notifier }
+
+// PresenceEnabled reports whether the presence stream is on, which
+// /capabilities and the admin API both report.
+func (s *API) PresenceEnabled() bool { return s.presenceEnabled }
+
+// LongPollTimeout is the cap a /sync's ?timeout= is held to.
+func (s *API) LongPollTimeout() time.Duration { return s.longPollTimeout }
+
+// nowMS is the current time in the unit the API uses on the wire.
+func (s *API) nowMS() int64 { return s.now().UnixMilli() }
+
+// Close releases what the component started. The typing tracker holds a
+// timer per typing user, and a server shutting down should not wait for
+// them.
+func (s *API) Close() { s.typing.close() }