Phase 5: federation

Implements PLAN.md Phase 5: the Server-Server API. Warp joins rooms on
other servers and lets their users join its rooms, invites across servers
with the invitee's server countersigning, rejects remote invites (locally
if the inviter is unreachable), and exchanges events, typing, receipts and
to-device messages with other homeservers — all behind the runtime switch,
which now starts and stops real outbound traffic.

The protocol lives in internal/federation with no knowledge of rooms:
X-Matrix signing (fuzzed parser), the spec's discovery algorithm keeping
the dial address and the TLS name apart, a signed client that refuses
every request while the switch is off or the policy denies the
destination, and a key ring that caches other servers' keys for at most a
week, with optional notaries.

The roomserver takes events from elsewhere through the spec's intake:
signatures and hashes (a hash mismatch keeps the redacted form), the rules
against the event's own auth events, against the state before it — with
/get_missing_events for small gaps and /state_ids for wide ones — and
against the current state, where a failure is a soft failure. Events held
without their state are outliers: a column on roomserver_events, because
the current state of a room joined over federation is almost all
outliers, and excluded from every timeline query. Forks are resolved at
the room's head. Membership across servers is make_join/send_join,
make_leave/send_leave and countersigned invites, in both directions;
m.room.server_acl is enforced on everything another server sends.
History before a remote join is backfilled when /messages pages past it,
stored at negative stream positions.

internal/federationapi is the running component: the inbound endpoints,
and an outbound queue that persists what it owes each destination, sends
it in transactions of 50 PDUs and 100 EDUs, backs off from ten seconds to
six hours, survives restarts, and drops events persisted while federation
is off rather than delivering a backlog in one burst. Its Module is what
the switch starts and stops.

Also: a TLS listener for port 8448, trusted key servers, key validity and
an outbound CA bundle in the config; per-destination health and backoff
reset in the admin API; remote profiles, aliases and directories in the
Client-Server API; a Complement federation profile. See
docs/federation.md, including its known gaps (no partial-state joins, no
key rotation).

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

server/test/complement/run.sh

@@ -1,8 +1,9 @@
 #!/usr/bin/env bash
 # Run Complement against a freshly built Warp image.
 #
-#   server/test/complement/run.sh                 # the test set for the phases that are done
-#   server/test/complement/run.sh TestLogin       # one test
+#   server/test/complement/run.sh                          # the C-S test set, federation off
+#   COMPLEMENT_PROFILE=federation server/test/complement/run.sh   # the federation test set
+#   server/test/complement/run.sh TestLogin                # one test
 #   COMPLEMENT_TESTS='Test.*' server/test/complement/run.sh
 #
 # Complement is a separate Go module, so it has to be checked out and run on
@@ -16,17 +17,42 @@ REPO_ROOT="$(cd "$(dirname "${BASH_SOURCE[0]}")/../../.." && pwd)"
 : "${COMPLEMENT_REF:=main}"
 : "${COMPLEMENT_DIR:=${REPO_ROOT}/.complement}"
 : "${COMPLEMENT_IMAGE:=complement-warp}"
+: "${COMPLEMENT_PROFILE:=csapi}"
 
-# The target set for the phases that are done, from PLAN.md section 7. Add
-# to it as phases land rather than switching to the whole suite: a pass
+# The target sets for the phases that are done, from PLAN.md section 7. Add
+# to them as phases land rather than switching to the whole suite: a pass
 # count that goes down because a later phase is missing is noise.
 #
-# Phase 4 adds the stream tests: /sync in all its forms, /messages,
-# /context, filters, typing, receipts and to-device messaging. What is still
-# out is anything needing a later phase — the notification counts of the
-# push rules (Phase 8), the device lists of the keyserver (Phase 7), and
-# the whole of federation (Phase 5).
-: "${COMPLEMENT_TESTS:=TestRegistration|TestLogin|TestDeviceManagement|TestRoomCreate|TestRoomState|TestRoomAlias|TestRoomForget|TestMembership.*|TestRedaction.*|TestJoinRoom.*|TestLeaveRoom.*|TestKickRoom.*|TestBanRoom.*|TestPublicRoom.*|TestSync.*|TestMessages.*|TestContext.*|TestFilter.*|TestTyping.*|TestReceipt.*|TestToDevice.*|TestRoomCreationReportsEventsToMyself|TestLazyLoading.*|TestWriteMDirectAccountData}"
+# csapi: the Client-Server tests, with federation off. Phase 4 added the
+# stream tests: /sync in all its forms, /messages, /context, filters,
+# typing, receipts and to-device messaging. What is still out is anything
+# needing a later phase — the notification counts of the push rules
+# (Phase 8) and the device lists of the keyserver (Phase 7).
+CSAPI_TESTS='TestRegistration|TestLogin|TestDeviceManagement|TestRoomCreate|TestRoomState|TestRoomAlias|TestRoomForget|TestMembership.*|TestRedaction.*|TestJoinRoom.*|TestLeaveRoom.*|TestKickRoom.*|TestBanRoom.*|TestPublicRoom.*|TestSync.*|TestMessages.*|TestContext.*|TestFilter.*|TestTyping.*|TestReceipt.*|TestToDevice.*|TestRoomCreationReportsEventsToMyself|TestLazyLoading.*|TestWriteMDirectAccountData'
+
+# federation: Phase 5. Joins, leaves and invites across servers, keys,
+# transactions, missing events, server ACLs and the queries. Partial-state
+# joins (MSC3706, TestPartialStateJoin*) are not implemented and are left
+# out; so is anything needing an application service (Phase 9) or device
+# lists (Phase 7).
+FEDERATION_TESTS='TestFederation.*|TestJoinViaRoomIDAndServerName|TestJoinFederatedRoom.*|TestBannedUserCannotSendJoin|TestCannotSendNon.*Via.*|TestCannotSendKnockViaSend.*|TestOutboundFederation.*|TestInboundFederation.*|TestInboundCanReturnMissingEvents|TestRemoteAliasRequestsUnderstandUnicode|TestNetworkPartitionOrdering|TestACLs|TestGetMissingEventsGapFilling|TestEventAuth|TestUnbanViaInvite|TestJoinRoomWithFederation.*|TestRemote.*'
+
+case "$COMPLEMENT_PROFILE" in
+	csapi)
+		: "${COMPLEMENT_TESTS:=$CSAPI_TESTS}"
+		PACKAGES=(./tests/csapi/...)
+		FEDERATION_ENABLED=false
+		;;
+	federation)
+		: "${COMPLEMENT_TESTS:=$FEDERATION_TESTS}"
+		PACKAGES=(./tests/... )
+		FEDERATION_ENABLED=true
+		;;
+	*)
+		echo "complement: unknown COMPLEMENT_PROFILE '$COMPLEMENT_PROFILE' (want csapi or federation)" >&2
+		exit 2
+		;;
+esac
 
 if [[ $# -gt 0 ]]; then
 	COMPLEMENT_TESTS="$(IFS='|'; echo "$*")"
@@ -49,12 +75,18 @@ else
 	git -C "$COMPLEMENT_DIR" checkout -q FETCH_HEAD
 fi
 
-echo "complement: running '${COMPLEMENT_TESTS}' with federation off"
+echo "complement: running '${COMPLEMENT_TESTS}' (profile ${COMPLEMENT_PROFILE}, federation ${FEDERATION_ENABLED})"
 cd "$COMPLEMENT_DIR"
 
 # COMPLEMENT_SPAWN_HS_TIMEOUT_SECS is generous because the first container
 # of a run has to apply the migrations on a cold SQLite file.
+#
+# COMPLEMENT_SHARE_ENV_PREFIX hands every host variable starting with
+# PASS_ to the homeserver containers, prefix removed: it is how the profile
+# reaches the entrypoint's federation switch.
 COMPLEMENT_BASE_IMAGE="$COMPLEMENT_IMAGE" \
 	COMPLEMENT_SPAWN_HS_TIMEOUT_SECS="${COMPLEMENT_SPAWN_HS_TIMEOUT_SECS:-60}" \
 	COMPLEMENT_DEBUG="${COMPLEMENT_DEBUG:-0}" \
-	go test -count=1 -v -timeout 20m -run "^(${COMPLEMENT_TESTS})$" ./tests/csapi/...
+	COMPLEMENT_SHARE_ENV_PREFIX=PASS_ \
+	PASS_WARP_FEDERATION_ENABLED="$FEDERATION_ENABLED" \
+	go test -count=1 -v -timeout 30m -run "^(${COMPLEMENT_TESTS})$" "${PACKAGES[@]}"