sliding-sync/state/to_device_table_test.go

342 lines
11 KiB
Go
Raw Permalink Normal View History

package state
import (
"bytes"
"encoding/json"
"testing"
"github.com/tidwall/gjson"
)
func TestToDeviceTable(t *testing.T) {
db, close := connectToDB(t)
defer close()
table := NewToDeviceTable(db)
2023-05-02 14:34:22 +01:00
sender := "@alice:localhost"
deviceID := "FOO"
var limit int64 = 999
msgs := []json.RawMessage{
json.RawMessage(`{"sender":"alice","type":"something","content":{"foo":"bar"}}`),
json.RawMessage(`{"sender":"bob","type":"something","content":{"foo":"bar2"}}`),
}
var lastPos int64
var err error
2023-05-02 14:34:22 +01:00
if lastPos, err = table.InsertMessages(sender, deviceID, msgs); err != nil {
t.Fatalf("InsertMessages: %s", err)
}
if lastPos != 2 {
t.Fatalf("InsertMessages: bad pos returned, got %d want 2", lastPos)
}
2023-05-02 14:34:22 +01:00
gotMsgs, upTo, err := table.Messages(sender, deviceID, 0, limit)
if err != nil {
t.Fatalf("Messages: %s", err)
}
if upTo != lastPos {
t.Errorf("Message: got up to %d want %d", upTo, lastPos)
}
if len(gotMsgs) != len(msgs) {
t.Fatalf("Messages: got %d messages, want %d", len(gotMsgs), len(msgs))
}
for i := range msgs {
if !bytes.Equal(msgs[i], gotMsgs[i]) {
t.Fatalf("Messages: got %+v want %+v", gotMsgs[i], msgs[i])
}
}
// same to= token, no messages
2023-05-02 14:34:22 +01:00
gotMsgs, upTo, err = table.Messages(sender, deviceID, lastPos, limit)
if err != nil {
t.Fatalf("Messages: %s", err)
}
if upTo != lastPos {
t.Errorf("Message: got up to %d want %d", upTo, lastPos)
}
if len(gotMsgs) > 0 {
t.Fatalf("Messages: got %d messages, want none", len(gotMsgs))
}
// different device ID, no messages
2023-05-02 14:34:22 +01:00
gotMsgs, upTo, err = table.Messages(sender, "OTHER_DEVICE", 0, limit)
if err != nil {
t.Fatalf("Messages: %s", err)
}
add extensions for typing and receipts; bugfixes and additional perf improvements Features: - Add `typing` extension. - Add `receipts` extension. - Add comprehensive prometheus `/metrics` activated via `SYNCV3_PROM`. - Add `SYNCV3_PPROF` support. - Add `by_notification_level` sort order. - Add `include_old_rooms` support. - Add support for `$ME` and `$LAZY`. - Add correct filtering when `*,*` is used as `required_state`. - Add `num_live` to each room response to indicate how many timeline entries are live. Bug fixes: - Use a stricter comparison function on ranges: fixes an issue whereby UTs fail on go1.19 due to change in sorting algorithm. - Send back an `errcode` on HTTP errors (e.g expired sessions). - Remove `unsigned.txn_id` on insertion into the DB. Otherwise other users would see other users txn IDs :( - Improve range delta algorithm: previously it didn't handle cases like `[0,20] -> [20,30]` and would panic. - Send HTTP 400 for invalid range requests. - Don't publish no-op unread counts which just adds extra noise. - Fix leaking DB connections which could eventually consume all available connections. - Ensure we always unblock WaitUntilInitialSync even on invalid access tokens. Other code relies on WaitUntilInitialSync() actually returning at _some_ point e.g on startup we have N workers which bound the number of concurrent pollers made at any one time, we need to not just hog a worker forever. Improvements: - Greatly improve startup times of sync3 handlers by improving `JoinedRoomsTracker`: a modest amount of data would take ~28s to create the handler, now it takes 4s. - Massively improve initial initial v3 sync times, by refactoring `JoinedRoomsTracker`, from ~47s to <1s. - Add `SlidingSyncUntil...` in tests to reduce races. - Tweak the API shape of JoinedUsersForRoom to reduce state block processing time for large rooms from 63s to 39s. - Add trace task for initial syncs. - Include the proxy version in UA strings. - HTTP errors now wait 1s before returning to stop clients tight-looping on error. - Pending event buffer is now 2000. - Index the room ID first to cull the most events when returning timeline entries. Speeds up `SelectLatestEventsBetween` by a factor of 8. - Remove cancelled `m.room_key_requests` from the to-device inbox. Cuts down the amount of events in the inbox by ~94% for very large (20k+) inboxes, ~50% for moderate sized (200 events) inboxes. Adds book-keeping to remember the unacked to-device position for each client.
2022-12-14 18:53:55 +00:00
if upTo != 0 {
t.Errorf("Message: got up to %d want %d", upTo, 0)
}
if len(gotMsgs) > 0 {
t.Fatalf("Messages: got %d messages, want none", len(gotMsgs))
}
// zero limit, no messages
2023-05-02 14:34:22 +01:00
gotMsgs, upTo, err = table.Messages(sender, deviceID, 0, 0)
if err != nil {
t.Fatalf("Messages: %s", err)
}
add extensions for typing and receipts; bugfixes and additional perf improvements Features: - Add `typing` extension. - Add `receipts` extension. - Add comprehensive prometheus `/metrics` activated via `SYNCV3_PROM`. - Add `SYNCV3_PPROF` support. - Add `by_notification_level` sort order. - Add `include_old_rooms` support. - Add support for `$ME` and `$LAZY`. - Add correct filtering when `*,*` is used as `required_state`. - Add `num_live` to each room response to indicate how many timeline entries are live. Bug fixes: - Use a stricter comparison function on ranges: fixes an issue whereby UTs fail on go1.19 due to change in sorting algorithm. - Send back an `errcode` on HTTP errors (e.g expired sessions). - Remove `unsigned.txn_id` on insertion into the DB. Otherwise other users would see other users txn IDs :( - Improve range delta algorithm: previously it didn't handle cases like `[0,20] -> [20,30]` and would panic. - Send HTTP 400 for invalid range requests. - Don't publish no-op unread counts which just adds extra noise. - Fix leaking DB connections which could eventually consume all available connections. - Ensure we always unblock WaitUntilInitialSync even on invalid access tokens. Other code relies on WaitUntilInitialSync() actually returning at _some_ point e.g on startup we have N workers which bound the number of concurrent pollers made at any one time, we need to not just hog a worker forever. Improvements: - Greatly improve startup times of sync3 handlers by improving `JoinedRoomsTracker`: a modest amount of data would take ~28s to create the handler, now it takes 4s. - Massively improve initial initial v3 sync times, by refactoring `JoinedRoomsTracker`, from ~47s to <1s. - Add `SlidingSyncUntil...` in tests to reduce races. - Tweak the API shape of JoinedUsersForRoom to reduce state block processing time for large rooms from 63s to 39s. - Add trace task for initial syncs. - Include the proxy version in UA strings. - HTTP errors now wait 1s before returning to stop clients tight-looping on error. - Pending event buffer is now 2000. - Index the room ID first to cull the most events when returning timeline entries. Speeds up `SelectLatestEventsBetween` by a factor of 8. - Remove cancelled `m.room_key_requests` from the to-device inbox. Cuts down the amount of events in the inbox by ~94% for very large (20k+) inboxes, ~50% for moderate sized (200 events) inboxes. Adds book-keeping to remember the unacked to-device position for each client.
2022-12-14 18:53:55 +00:00
if upTo != 0 {
t.Errorf("Message: got up to %d want %d", upTo, 0)
}
if len(gotMsgs) > 0 {
t.Fatalf("Messages: got %d messages, want none", len(gotMsgs))
}
// lower limit, cap out
var wantLimit int64 = 1
2023-05-02 14:34:22 +01:00
gotMsgs, upTo, err = table.Messages(sender, deviceID, 0, wantLimit)
if err != nil {
t.Fatalf("Messages: %s", err)
}
// we inserted 2 messages, and request a limit of 1 so the position should be one before
if upTo != (lastPos - 1) {
t.Errorf("Message: got up to %d want %d", upTo, lastPos-1)
}
if int64(len(gotMsgs)) != wantLimit {
t.Fatalf("Messages: got %d messages, want %d", len(gotMsgs), wantLimit)
}
// delete the first message, requerying only gives 1 message
2023-05-02 14:34:22 +01:00
if err := table.DeleteMessagesUpToAndIncluding(sender, deviceID, lastPos-1); err != nil {
t.Fatalf("DeleteMessagesUpTo: %s", err)
}
2023-05-02 14:34:22 +01:00
gotMsgs, upTo, err = table.Messages(sender, deviceID, 0, limit)
if err != nil {
t.Fatalf("Messages: %s", err)
}
if upTo != lastPos {
t.Errorf("Message: got up to %d want %d", upTo, lastPos)
}
if len(gotMsgs) != 1 {
t.Fatalf("Messages: got %d messages, want %d", len(gotMsgs), 1)
}
want, err := json.Marshal(msgs[1])
if err != nil {
t.Fatalf("failed to marshal msg: %s", err)
}
if !bytes.Equal(gotMsgs[0], want) {
t.Fatalf("Messages: deleted message but unexpected message left: got %s want %s", string(gotMsgs[0]), string(want))
}
// delete everything and check it works
2023-05-02 14:34:22 +01:00
err = table.DeleteAllMessagesForDevice(sender, deviceID)
assertNoError(t, err)
2023-05-02 14:34:22 +01:00
msgs, _, err = table.Messages(sender, deviceID, -1, 10)
assertNoError(t, err)
assertVal(t, "wanted 0 msgs", len(msgs), 0)
}
add extensions for typing and receipts; bugfixes and additional perf improvements Features: - Add `typing` extension. - Add `receipts` extension. - Add comprehensive prometheus `/metrics` activated via `SYNCV3_PROM`. - Add `SYNCV3_PPROF` support. - Add `by_notification_level` sort order. - Add `include_old_rooms` support. - Add support for `$ME` and `$LAZY`. - Add correct filtering when `*,*` is used as `required_state`. - Add `num_live` to each room response to indicate how many timeline entries are live. Bug fixes: - Use a stricter comparison function on ranges: fixes an issue whereby UTs fail on go1.19 due to change in sorting algorithm. - Send back an `errcode` on HTTP errors (e.g expired sessions). - Remove `unsigned.txn_id` on insertion into the DB. Otherwise other users would see other users txn IDs :( - Improve range delta algorithm: previously it didn't handle cases like `[0,20] -> [20,30]` and would panic. - Send HTTP 400 for invalid range requests. - Don't publish no-op unread counts which just adds extra noise. - Fix leaking DB connections which could eventually consume all available connections. - Ensure we always unblock WaitUntilInitialSync even on invalid access tokens. Other code relies on WaitUntilInitialSync() actually returning at _some_ point e.g on startup we have N workers which bound the number of concurrent pollers made at any one time, we need to not just hog a worker forever. Improvements: - Greatly improve startup times of sync3 handlers by improving `JoinedRoomsTracker`: a modest amount of data would take ~28s to create the handler, now it takes 4s. - Massively improve initial initial v3 sync times, by refactoring `JoinedRoomsTracker`, from ~47s to <1s. - Add `SlidingSyncUntil...` in tests to reduce races. - Tweak the API shape of JoinedUsersForRoom to reduce state block processing time for large rooms from 63s to 39s. - Add trace task for initial syncs. - Include the proxy version in UA strings. - HTTP errors now wait 1s before returning to stop clients tight-looping on error. - Pending event buffer is now 2000. - Index the room ID first to cull the most events when returning timeline entries. Speeds up `SelectLatestEventsBetween` by a factor of 8. - Remove cancelled `m.room_key_requests` from the to-device inbox. Cuts down the amount of events in the inbox by ~94% for very large (20k+) inboxes, ~50% for moderate sized (200 events) inboxes. Adds book-keeping to remember the unacked to-device position for each client.
2022-12-14 18:53:55 +00:00
// Test that https://github.com/uhoreg/matrix-doc/blob/drop-stale-to-device/proposals/3944-drop-stale-to-device.md works for m.room_key_request
func TestToDeviceTableDeleteCancels(t *testing.T) {
db, close := connectToDB(t)
defer close()
add extensions for typing and receipts; bugfixes and additional perf improvements Features: - Add `typing` extension. - Add `receipts` extension. - Add comprehensive prometheus `/metrics` activated via `SYNCV3_PROM`. - Add `SYNCV3_PPROF` support. - Add `by_notification_level` sort order. - Add `include_old_rooms` support. - Add support for `$ME` and `$LAZY`. - Add correct filtering when `*,*` is used as `required_state`. - Add `num_live` to each room response to indicate how many timeline entries are live. Bug fixes: - Use a stricter comparison function on ranges: fixes an issue whereby UTs fail on go1.19 due to change in sorting algorithm. - Send back an `errcode` on HTTP errors (e.g expired sessions). - Remove `unsigned.txn_id` on insertion into the DB. Otherwise other users would see other users txn IDs :( - Improve range delta algorithm: previously it didn't handle cases like `[0,20] -> [20,30]` and would panic. - Send HTTP 400 for invalid range requests. - Don't publish no-op unread counts which just adds extra noise. - Fix leaking DB connections which could eventually consume all available connections. - Ensure we always unblock WaitUntilInitialSync even on invalid access tokens. Other code relies on WaitUntilInitialSync() actually returning at _some_ point e.g on startup we have N workers which bound the number of concurrent pollers made at any one time, we need to not just hog a worker forever. Improvements: - Greatly improve startup times of sync3 handlers by improving `JoinedRoomsTracker`: a modest amount of data would take ~28s to create the handler, now it takes 4s. - Massively improve initial initial v3 sync times, by refactoring `JoinedRoomsTracker`, from ~47s to <1s. - Add `SlidingSyncUntil...` in tests to reduce races. - Tweak the API shape of JoinedUsersForRoom to reduce state block processing time for large rooms from 63s to 39s. - Add trace task for initial syncs. - Include the proxy version in UA strings. - HTTP errors now wait 1s before returning to stop clients tight-looping on error. - Pending event buffer is now 2000. - Index the room ID first to cull the most events when returning timeline entries. Speeds up `SelectLatestEventsBetween` by a factor of 8. - Remove cancelled `m.room_key_requests` from the to-device inbox. Cuts down the amount of events in the inbox by ~94% for very large (20k+) inboxes, ~50% for moderate sized (200 events) inboxes. Adds book-keeping to remember the unacked to-device position for each client.
2022-12-14 18:53:55 +00:00
sender := "SENDER"
destination := "DEST"
table := NewToDeviceTable(db)
// insert 2 requests
reqEv1 := newRoomKeyEvent(t, "request", "1", sender, map[string]interface{}{
"foo": "bar",
})
2023-05-02 14:34:22 +01:00
_, err := table.InsertMessages(sender, destination, []json.RawMessage{reqEv1})
add extensions for typing and receipts; bugfixes and additional perf improvements Features: - Add `typing` extension. - Add `receipts` extension. - Add comprehensive prometheus `/metrics` activated via `SYNCV3_PROM`. - Add `SYNCV3_PPROF` support. - Add `by_notification_level` sort order. - Add `include_old_rooms` support. - Add support for `$ME` and `$LAZY`. - Add correct filtering when `*,*` is used as `required_state`. - Add `num_live` to each room response to indicate how many timeline entries are live. Bug fixes: - Use a stricter comparison function on ranges: fixes an issue whereby UTs fail on go1.19 due to change in sorting algorithm. - Send back an `errcode` on HTTP errors (e.g expired sessions). - Remove `unsigned.txn_id` on insertion into the DB. Otherwise other users would see other users txn IDs :( - Improve range delta algorithm: previously it didn't handle cases like `[0,20] -> [20,30]` and would panic. - Send HTTP 400 for invalid range requests. - Don't publish no-op unread counts which just adds extra noise. - Fix leaking DB connections which could eventually consume all available connections. - Ensure we always unblock WaitUntilInitialSync even on invalid access tokens. Other code relies on WaitUntilInitialSync() actually returning at _some_ point e.g on startup we have N workers which bound the number of concurrent pollers made at any one time, we need to not just hog a worker forever. Improvements: - Greatly improve startup times of sync3 handlers by improving `JoinedRoomsTracker`: a modest amount of data would take ~28s to create the handler, now it takes 4s. - Massively improve initial initial v3 sync times, by refactoring `JoinedRoomsTracker`, from ~47s to <1s. - Add `SlidingSyncUntil...` in tests to reduce races. - Tweak the API shape of JoinedUsersForRoom to reduce state block processing time for large rooms from 63s to 39s. - Add trace task for initial syncs. - Include the proxy version in UA strings. - HTTP errors now wait 1s before returning to stop clients tight-looping on error. - Pending event buffer is now 2000. - Index the room ID first to cull the most events when returning timeline entries. Speeds up `SelectLatestEventsBetween` by a factor of 8. - Remove cancelled `m.room_key_requests` from the to-device inbox. Cuts down the amount of events in the inbox by ~94% for very large (20k+) inboxes, ~50% for moderate sized (200 events) inboxes. Adds book-keeping to remember the unacked to-device position for each client.
2022-12-14 18:53:55 +00:00
assertNoError(t, err)
2023-05-02 14:34:22 +01:00
gotMsgs, _, err := table.Messages(sender, destination, 0, 10)
add extensions for typing and receipts; bugfixes and additional perf improvements Features: - Add `typing` extension. - Add `receipts` extension. - Add comprehensive prometheus `/metrics` activated via `SYNCV3_PROM`. - Add `SYNCV3_PPROF` support. - Add `by_notification_level` sort order. - Add `include_old_rooms` support. - Add support for `$ME` and `$LAZY`. - Add correct filtering when `*,*` is used as `required_state`. - Add `num_live` to each room response to indicate how many timeline entries are live. Bug fixes: - Use a stricter comparison function on ranges: fixes an issue whereby UTs fail on go1.19 due to change in sorting algorithm. - Send back an `errcode` on HTTP errors (e.g expired sessions). - Remove `unsigned.txn_id` on insertion into the DB. Otherwise other users would see other users txn IDs :( - Improve range delta algorithm: previously it didn't handle cases like `[0,20] -> [20,30]` and would panic. - Send HTTP 400 for invalid range requests. - Don't publish no-op unread counts which just adds extra noise. - Fix leaking DB connections which could eventually consume all available connections. - Ensure we always unblock WaitUntilInitialSync even on invalid access tokens. Other code relies on WaitUntilInitialSync() actually returning at _some_ point e.g on startup we have N workers which bound the number of concurrent pollers made at any one time, we need to not just hog a worker forever. Improvements: - Greatly improve startup times of sync3 handlers by improving `JoinedRoomsTracker`: a modest amount of data would take ~28s to create the handler, now it takes 4s. - Massively improve initial initial v3 sync times, by refactoring `JoinedRoomsTracker`, from ~47s to <1s. - Add `SlidingSyncUntil...` in tests to reduce races. - Tweak the API shape of JoinedUsersForRoom to reduce state block processing time for large rooms from 63s to 39s. - Add trace task for initial syncs. - Include the proxy version in UA strings. - HTTP errors now wait 1s before returning to stop clients tight-looping on error. - Pending event buffer is now 2000. - Index the room ID first to cull the most events when returning timeline entries. Speeds up `SelectLatestEventsBetween` by a factor of 8. - Remove cancelled `m.room_key_requests` from the to-device inbox. Cuts down the amount of events in the inbox by ~94% for very large (20k+) inboxes, ~50% for moderate sized (200 events) inboxes. Adds book-keeping to remember the unacked to-device position for each client.
2022-12-14 18:53:55 +00:00
assertNoError(t, err)
bytesEqual(t, gotMsgs[0], reqEv1)
reqEv2 := newRoomKeyEvent(t, "request", "2", sender, map[string]interface{}{
"foo": "baz",
})
2023-05-02 14:34:22 +01:00
_, err = table.InsertMessages(sender, destination, []json.RawMessage{reqEv2})
add extensions for typing and receipts; bugfixes and additional perf improvements Features: - Add `typing` extension. - Add `receipts` extension. - Add comprehensive prometheus `/metrics` activated via `SYNCV3_PROM`. - Add `SYNCV3_PPROF` support. - Add `by_notification_level` sort order. - Add `include_old_rooms` support. - Add support for `$ME` and `$LAZY`. - Add correct filtering when `*,*` is used as `required_state`. - Add `num_live` to each room response to indicate how many timeline entries are live. Bug fixes: - Use a stricter comparison function on ranges: fixes an issue whereby UTs fail on go1.19 due to change in sorting algorithm. - Send back an `errcode` on HTTP errors (e.g expired sessions). - Remove `unsigned.txn_id` on insertion into the DB. Otherwise other users would see other users txn IDs :( - Improve range delta algorithm: previously it didn't handle cases like `[0,20] -> [20,30]` and would panic. - Send HTTP 400 for invalid range requests. - Don't publish no-op unread counts which just adds extra noise. - Fix leaking DB connections which could eventually consume all available connections. - Ensure we always unblock WaitUntilInitialSync even on invalid access tokens. Other code relies on WaitUntilInitialSync() actually returning at _some_ point e.g on startup we have N workers which bound the number of concurrent pollers made at any one time, we need to not just hog a worker forever. Improvements: - Greatly improve startup times of sync3 handlers by improving `JoinedRoomsTracker`: a modest amount of data would take ~28s to create the handler, now it takes 4s. - Massively improve initial initial v3 sync times, by refactoring `JoinedRoomsTracker`, from ~47s to <1s. - Add `SlidingSyncUntil...` in tests to reduce races. - Tweak the API shape of JoinedUsersForRoom to reduce state block processing time for large rooms from 63s to 39s. - Add trace task for initial syncs. - Include the proxy version in UA strings. - HTTP errors now wait 1s before returning to stop clients tight-looping on error. - Pending event buffer is now 2000. - Index the room ID first to cull the most events when returning timeline entries. Speeds up `SelectLatestEventsBetween` by a factor of 8. - Remove cancelled `m.room_key_requests` from the to-device inbox. Cuts down the amount of events in the inbox by ~94% for very large (20k+) inboxes, ~50% for moderate sized (200 events) inboxes. Adds book-keeping to remember the unacked to-device position for each client.
2022-12-14 18:53:55 +00:00
assertNoError(t, err)
2023-05-02 14:34:22 +01:00
gotMsgs, _, err = table.Messages(sender, destination, 0, 10)
add extensions for typing and receipts; bugfixes and additional perf improvements Features: - Add `typing` extension. - Add `receipts` extension. - Add comprehensive prometheus `/metrics` activated via `SYNCV3_PROM`. - Add `SYNCV3_PPROF` support. - Add `by_notification_level` sort order. - Add `include_old_rooms` support. - Add support for `$ME` and `$LAZY`. - Add correct filtering when `*,*` is used as `required_state`. - Add `num_live` to each room response to indicate how many timeline entries are live. Bug fixes: - Use a stricter comparison function on ranges: fixes an issue whereby UTs fail on go1.19 due to change in sorting algorithm. - Send back an `errcode` on HTTP errors (e.g expired sessions). - Remove `unsigned.txn_id` on insertion into the DB. Otherwise other users would see other users txn IDs :( - Improve range delta algorithm: previously it didn't handle cases like `[0,20] -> [20,30]` and would panic. - Send HTTP 400 for invalid range requests. - Don't publish no-op unread counts which just adds extra noise. - Fix leaking DB connections which could eventually consume all available connections. - Ensure we always unblock WaitUntilInitialSync even on invalid access tokens. Other code relies on WaitUntilInitialSync() actually returning at _some_ point e.g on startup we have N workers which bound the number of concurrent pollers made at any one time, we need to not just hog a worker forever. Improvements: - Greatly improve startup times of sync3 handlers by improving `JoinedRoomsTracker`: a modest amount of data would take ~28s to create the handler, now it takes 4s. - Massively improve initial initial v3 sync times, by refactoring `JoinedRoomsTracker`, from ~47s to <1s. - Add `SlidingSyncUntil...` in tests to reduce races. - Tweak the API shape of JoinedUsersForRoom to reduce state block processing time for large rooms from 63s to 39s. - Add trace task for initial syncs. - Include the proxy version in UA strings. - HTTP errors now wait 1s before returning to stop clients tight-looping on error. - Pending event buffer is now 2000. - Index the room ID first to cull the most events when returning timeline entries. Speeds up `SelectLatestEventsBetween` by a factor of 8. - Remove cancelled `m.room_key_requests` from the to-device inbox. Cuts down the amount of events in the inbox by ~94% for very large (20k+) inboxes, ~50% for moderate sized (200 events) inboxes. Adds book-keeping to remember the unacked to-device position for each client.
2022-12-14 18:53:55 +00:00
assertNoError(t, err)
bytesEqual(t, gotMsgs[1], reqEv2)
// now delete 1
cancelEv1 := newRoomKeyEvent(t, "request_cancellation", "1", sender, nil)
2023-05-02 14:34:22 +01:00
_, err = table.InsertMessages(sender, destination, []json.RawMessage{cancelEv1})
add extensions for typing and receipts; bugfixes and additional perf improvements Features: - Add `typing` extension. - Add `receipts` extension. - Add comprehensive prometheus `/metrics` activated via `SYNCV3_PROM`. - Add `SYNCV3_PPROF` support. - Add `by_notification_level` sort order. - Add `include_old_rooms` support. - Add support for `$ME` and `$LAZY`. - Add correct filtering when `*,*` is used as `required_state`. - Add `num_live` to each room response to indicate how many timeline entries are live. Bug fixes: - Use a stricter comparison function on ranges: fixes an issue whereby UTs fail on go1.19 due to change in sorting algorithm. - Send back an `errcode` on HTTP errors (e.g expired sessions). - Remove `unsigned.txn_id` on insertion into the DB. Otherwise other users would see other users txn IDs :( - Improve range delta algorithm: previously it didn't handle cases like `[0,20] -> [20,30]` and would panic. - Send HTTP 400 for invalid range requests. - Don't publish no-op unread counts which just adds extra noise. - Fix leaking DB connections which could eventually consume all available connections. - Ensure we always unblock WaitUntilInitialSync even on invalid access tokens. Other code relies on WaitUntilInitialSync() actually returning at _some_ point e.g on startup we have N workers which bound the number of concurrent pollers made at any one time, we need to not just hog a worker forever. Improvements: - Greatly improve startup times of sync3 handlers by improving `JoinedRoomsTracker`: a modest amount of data would take ~28s to create the handler, now it takes 4s. - Massively improve initial initial v3 sync times, by refactoring `JoinedRoomsTracker`, from ~47s to <1s. - Add `SlidingSyncUntil...` in tests to reduce races. - Tweak the API shape of JoinedUsersForRoom to reduce state block processing time for large rooms from 63s to 39s. - Add trace task for initial syncs. - Include the proxy version in UA strings. - HTTP errors now wait 1s before returning to stop clients tight-looping on error. - Pending event buffer is now 2000. - Index the room ID first to cull the most events when returning timeline entries. Speeds up `SelectLatestEventsBetween` by a factor of 8. - Remove cancelled `m.room_key_requests` from the to-device inbox. Cuts down the amount of events in the inbox by ~94% for very large (20k+) inboxes, ~50% for moderate sized (200 events) inboxes. Adds book-keeping to remember the unacked to-device position for each client.
2022-12-14 18:53:55 +00:00
assertNoError(t, err)
// selecting messages now returns only reqEv2
2023-05-02 14:34:22 +01:00
gotMsgs, _, err = table.Messages(sender, destination, 0, 10)
add extensions for typing and receipts; bugfixes and additional perf improvements Features: - Add `typing` extension. - Add `receipts` extension. - Add comprehensive prometheus `/metrics` activated via `SYNCV3_PROM`. - Add `SYNCV3_PPROF` support. - Add `by_notification_level` sort order. - Add `include_old_rooms` support. - Add support for `$ME` and `$LAZY`. - Add correct filtering when `*,*` is used as `required_state`. - Add `num_live` to each room response to indicate how many timeline entries are live. Bug fixes: - Use a stricter comparison function on ranges: fixes an issue whereby UTs fail on go1.19 due to change in sorting algorithm. - Send back an `errcode` on HTTP errors (e.g expired sessions). - Remove `unsigned.txn_id` on insertion into the DB. Otherwise other users would see other users txn IDs :( - Improve range delta algorithm: previously it didn't handle cases like `[0,20] -> [20,30]` and would panic. - Send HTTP 400 for invalid range requests. - Don't publish no-op unread counts which just adds extra noise. - Fix leaking DB connections which could eventually consume all available connections. - Ensure we always unblock WaitUntilInitialSync even on invalid access tokens. Other code relies on WaitUntilInitialSync() actually returning at _some_ point e.g on startup we have N workers which bound the number of concurrent pollers made at any one time, we need to not just hog a worker forever. Improvements: - Greatly improve startup times of sync3 handlers by improving `JoinedRoomsTracker`: a modest amount of data would take ~28s to create the handler, now it takes 4s. - Massively improve initial initial v3 sync times, by refactoring `JoinedRoomsTracker`, from ~47s to <1s. - Add `SlidingSyncUntil...` in tests to reduce races. - Tweak the API shape of JoinedUsersForRoom to reduce state block processing time for large rooms from 63s to 39s. - Add trace task for initial syncs. - Include the proxy version in UA strings. - HTTP errors now wait 1s before returning to stop clients tight-looping on error. - Pending event buffer is now 2000. - Index the room ID first to cull the most events when returning timeline entries. Speeds up `SelectLatestEventsBetween` by a factor of 8. - Remove cancelled `m.room_key_requests` from the to-device inbox. Cuts down the amount of events in the inbox by ~94% for very large (20k+) inboxes, ~50% for moderate sized (200 events) inboxes. Adds book-keeping to remember the unacked to-device position for each client.
2022-12-14 18:53:55 +00:00
assertNoError(t, err)
bytesEqual(t, gotMsgs[0], reqEv2)
// now do lots of close but not quite cancellation requests that should not match reqEv2
2023-05-02 14:34:22 +01:00
_, err = table.InsertMessages(sender, destination, []json.RawMessage{
add extensions for typing and receipts; bugfixes and additional perf improvements Features: - Add `typing` extension. - Add `receipts` extension. - Add comprehensive prometheus `/metrics` activated via `SYNCV3_PROM`. - Add `SYNCV3_PPROF` support. - Add `by_notification_level` sort order. - Add `include_old_rooms` support. - Add support for `$ME` and `$LAZY`. - Add correct filtering when `*,*` is used as `required_state`. - Add `num_live` to each room response to indicate how many timeline entries are live. Bug fixes: - Use a stricter comparison function on ranges: fixes an issue whereby UTs fail on go1.19 due to change in sorting algorithm. - Send back an `errcode` on HTTP errors (e.g expired sessions). - Remove `unsigned.txn_id` on insertion into the DB. Otherwise other users would see other users txn IDs :( - Improve range delta algorithm: previously it didn't handle cases like `[0,20] -> [20,30]` and would panic. - Send HTTP 400 for invalid range requests. - Don't publish no-op unread counts which just adds extra noise. - Fix leaking DB connections which could eventually consume all available connections. - Ensure we always unblock WaitUntilInitialSync even on invalid access tokens. Other code relies on WaitUntilInitialSync() actually returning at _some_ point e.g on startup we have N workers which bound the number of concurrent pollers made at any one time, we need to not just hog a worker forever. Improvements: - Greatly improve startup times of sync3 handlers by improving `JoinedRoomsTracker`: a modest amount of data would take ~28s to create the handler, now it takes 4s. - Massively improve initial initial v3 sync times, by refactoring `JoinedRoomsTracker`, from ~47s to <1s. - Add `SlidingSyncUntil...` in tests to reduce races. - Tweak the API shape of JoinedUsersForRoom to reduce state block processing time for large rooms from 63s to 39s. - Add trace task for initial syncs. - Include the proxy version in UA strings. - HTTP errors now wait 1s before returning to stop clients tight-looping on error. - Pending event buffer is now 2000. - Index the room ID first to cull the most events when returning timeline entries. Speeds up `SelectLatestEventsBetween` by a factor of 8. - Remove cancelled `m.room_key_requests` from the to-device inbox. Cuts down the amount of events in the inbox by ~94% for very large (20k+) inboxes, ~50% for moderate sized (200 events) inboxes. Adds book-keeping to remember the unacked to-device position for each client.
2022-12-14 18:53:55 +00:00
newRoomKeyEvent(t, "cancellation", "2", sender, nil), // wrong action
newRoomKeyEvent(t, "request_cancellation", "22", sender, nil), // wrong request ID
newRoomKeyEvent(t, "request_cancellation", "2", "not_who_you_think", nil), // wrong req device id
})
assertNoError(t, err)
2023-05-02 14:34:22 +01:00
_, err = table.InsertMessages(sender, "wrong_destination", []json.RawMessage{ // wrong destination
add extensions for typing and receipts; bugfixes and additional perf improvements Features: - Add `typing` extension. - Add `receipts` extension. - Add comprehensive prometheus `/metrics` activated via `SYNCV3_PROM`. - Add `SYNCV3_PPROF` support. - Add `by_notification_level` sort order. - Add `include_old_rooms` support. - Add support for `$ME` and `$LAZY`. - Add correct filtering when `*,*` is used as `required_state`. - Add `num_live` to each room response to indicate how many timeline entries are live. Bug fixes: - Use a stricter comparison function on ranges: fixes an issue whereby UTs fail on go1.19 due to change in sorting algorithm. - Send back an `errcode` on HTTP errors (e.g expired sessions). - Remove `unsigned.txn_id` on insertion into the DB. Otherwise other users would see other users txn IDs :( - Improve range delta algorithm: previously it didn't handle cases like `[0,20] -> [20,30]` and would panic. - Send HTTP 400 for invalid range requests. - Don't publish no-op unread counts which just adds extra noise. - Fix leaking DB connections which could eventually consume all available connections. - Ensure we always unblock WaitUntilInitialSync even on invalid access tokens. Other code relies on WaitUntilInitialSync() actually returning at _some_ point e.g on startup we have N workers which bound the number of concurrent pollers made at any one time, we need to not just hog a worker forever. Improvements: - Greatly improve startup times of sync3 handlers by improving `JoinedRoomsTracker`: a modest amount of data would take ~28s to create the handler, now it takes 4s. - Massively improve initial initial v3 sync times, by refactoring `JoinedRoomsTracker`, from ~47s to <1s. - Add `SlidingSyncUntil...` in tests to reduce races. - Tweak the API shape of JoinedUsersForRoom to reduce state block processing time for large rooms from 63s to 39s. - Add trace task for initial syncs. - Include the proxy version in UA strings. - HTTP errors now wait 1s before returning to stop clients tight-looping on error. - Pending event buffer is now 2000. - Index the room ID first to cull the most events when returning timeline entries. Speeds up `SelectLatestEventsBetween` by a factor of 8. - Remove cancelled `m.room_key_requests` from the to-device inbox. Cuts down the amount of events in the inbox by ~94% for very large (20k+) inboxes, ~50% for moderate sized (200 events) inboxes. Adds book-keeping to remember the unacked to-device position for each client.
2022-12-14 18:53:55 +00:00
newRoomKeyEvent(t, "request_cancellation", "2", sender, nil),
})
assertNoError(t, err)
2023-05-02 14:34:22 +01:00
gotMsgs, _, err = table.Messages(sender, destination, 0, 10)
add extensions for typing and receipts; bugfixes and additional perf improvements Features: - Add `typing` extension. - Add `receipts` extension. - Add comprehensive prometheus `/metrics` activated via `SYNCV3_PROM`. - Add `SYNCV3_PPROF` support. - Add `by_notification_level` sort order. - Add `include_old_rooms` support. - Add support for `$ME` and `$LAZY`. - Add correct filtering when `*,*` is used as `required_state`. - Add `num_live` to each room response to indicate how many timeline entries are live. Bug fixes: - Use a stricter comparison function on ranges: fixes an issue whereby UTs fail on go1.19 due to change in sorting algorithm. - Send back an `errcode` on HTTP errors (e.g expired sessions). - Remove `unsigned.txn_id` on insertion into the DB. Otherwise other users would see other users txn IDs :( - Improve range delta algorithm: previously it didn't handle cases like `[0,20] -> [20,30]` and would panic. - Send HTTP 400 for invalid range requests. - Don't publish no-op unread counts which just adds extra noise. - Fix leaking DB connections which could eventually consume all available connections. - Ensure we always unblock WaitUntilInitialSync even on invalid access tokens. Other code relies on WaitUntilInitialSync() actually returning at _some_ point e.g on startup we have N workers which bound the number of concurrent pollers made at any one time, we need to not just hog a worker forever. Improvements: - Greatly improve startup times of sync3 handlers by improving `JoinedRoomsTracker`: a modest amount of data would take ~28s to create the handler, now it takes 4s. - Massively improve initial initial v3 sync times, by refactoring `JoinedRoomsTracker`, from ~47s to <1s. - Add `SlidingSyncUntil...` in tests to reduce races. - Tweak the API shape of JoinedUsersForRoom to reduce state block processing time for large rooms from 63s to 39s. - Add trace task for initial syncs. - Include the proxy version in UA strings. - HTTP errors now wait 1s before returning to stop clients tight-looping on error. - Pending event buffer is now 2000. - Index the room ID first to cull the most events when returning timeline entries. Speeds up `SelectLatestEventsBetween` by a factor of 8. - Remove cancelled `m.room_key_requests` from the to-device inbox. Cuts down the amount of events in the inbox by ~94% for very large (20k+) inboxes, ~50% for moderate sized (200 events) inboxes. Adds book-keeping to remember the unacked to-device position for each client.
2022-12-14 18:53:55 +00:00
assertNoError(t, err)
bytesEqual(t, gotMsgs[0], reqEv2) // the request lives on
if len(gotMsgs) != 4 { // the cancellations live on too, but not the one sent to the wrong dest
t.Errorf("got %d msgs, want 4", len(gotMsgs))
}
// request + cancel in one go => nothing inserted
destination2 := "DEST2"
2023-05-02 14:34:22 +01:00
_, err = table.InsertMessages(sender, destination2, []json.RawMessage{
add extensions for typing and receipts; bugfixes and additional perf improvements Features: - Add `typing` extension. - Add `receipts` extension. - Add comprehensive prometheus `/metrics` activated via `SYNCV3_PROM`. - Add `SYNCV3_PPROF` support. - Add `by_notification_level` sort order. - Add `include_old_rooms` support. - Add support for `$ME` and `$LAZY`. - Add correct filtering when `*,*` is used as `required_state`. - Add `num_live` to each room response to indicate how many timeline entries are live. Bug fixes: - Use a stricter comparison function on ranges: fixes an issue whereby UTs fail on go1.19 due to change in sorting algorithm. - Send back an `errcode` on HTTP errors (e.g expired sessions). - Remove `unsigned.txn_id` on insertion into the DB. Otherwise other users would see other users txn IDs :( - Improve range delta algorithm: previously it didn't handle cases like `[0,20] -> [20,30]` and would panic. - Send HTTP 400 for invalid range requests. - Don't publish no-op unread counts which just adds extra noise. - Fix leaking DB connections which could eventually consume all available connections. - Ensure we always unblock WaitUntilInitialSync even on invalid access tokens. Other code relies on WaitUntilInitialSync() actually returning at _some_ point e.g on startup we have N workers which bound the number of concurrent pollers made at any one time, we need to not just hog a worker forever. Improvements: - Greatly improve startup times of sync3 handlers by improving `JoinedRoomsTracker`: a modest amount of data would take ~28s to create the handler, now it takes 4s. - Massively improve initial initial v3 sync times, by refactoring `JoinedRoomsTracker`, from ~47s to <1s. - Add `SlidingSyncUntil...` in tests to reduce races. - Tweak the API shape of JoinedUsersForRoom to reduce state block processing time for large rooms from 63s to 39s. - Add trace task for initial syncs. - Include the proxy version in UA strings. - HTTP errors now wait 1s before returning to stop clients tight-looping on error. - Pending event buffer is now 2000. - Index the room ID first to cull the most events when returning timeline entries. Speeds up `SelectLatestEventsBetween` by a factor of 8. - Remove cancelled `m.room_key_requests` from the to-device inbox. Cuts down the amount of events in the inbox by ~94% for very large (20k+) inboxes, ~50% for moderate sized (200 events) inboxes. Adds book-keeping to remember the unacked to-device position for each client.
2022-12-14 18:53:55 +00:00
newRoomKeyEvent(t, "request", "A", sender, map[string]interface{}{
"foo": "baz",
}),
newRoomKeyEvent(t, "request_cancellation", "A", sender, nil),
})
assertNoError(t, err)
2023-05-02 14:34:22 +01:00
gotMsgs, _, err = table.Messages(sender, destination2, 0, 10)
add extensions for typing and receipts; bugfixes and additional perf improvements Features: - Add `typing` extension. - Add `receipts` extension. - Add comprehensive prometheus `/metrics` activated via `SYNCV3_PROM`. - Add `SYNCV3_PPROF` support. - Add `by_notification_level` sort order. - Add `include_old_rooms` support. - Add support for `$ME` and `$LAZY`. - Add correct filtering when `*,*` is used as `required_state`. - Add `num_live` to each room response to indicate how many timeline entries are live. Bug fixes: - Use a stricter comparison function on ranges: fixes an issue whereby UTs fail on go1.19 due to change in sorting algorithm. - Send back an `errcode` on HTTP errors (e.g expired sessions). - Remove `unsigned.txn_id` on insertion into the DB. Otherwise other users would see other users txn IDs :( - Improve range delta algorithm: previously it didn't handle cases like `[0,20] -> [20,30]` and would panic. - Send HTTP 400 for invalid range requests. - Don't publish no-op unread counts which just adds extra noise. - Fix leaking DB connections which could eventually consume all available connections. - Ensure we always unblock WaitUntilInitialSync even on invalid access tokens. Other code relies on WaitUntilInitialSync() actually returning at _some_ point e.g on startup we have N workers which bound the number of concurrent pollers made at any one time, we need to not just hog a worker forever. Improvements: - Greatly improve startup times of sync3 handlers by improving `JoinedRoomsTracker`: a modest amount of data would take ~28s to create the handler, now it takes 4s. - Massively improve initial initial v3 sync times, by refactoring `JoinedRoomsTracker`, from ~47s to <1s. - Add `SlidingSyncUntil...` in tests to reduce races. - Tweak the API shape of JoinedUsersForRoom to reduce state block processing time for large rooms from 63s to 39s. - Add trace task for initial syncs. - Include the proxy version in UA strings. - HTTP errors now wait 1s before returning to stop clients tight-looping on error. - Pending event buffer is now 2000. - Index the room ID first to cull the most events when returning timeline entries. Speeds up `SelectLatestEventsBetween` by a factor of 8. - Remove cancelled `m.room_key_requests` from the to-device inbox. Cuts down the amount of events in the inbox by ~94% for very large (20k+) inboxes, ~50% for moderate sized (200 events) inboxes. Adds book-keeping to remember the unacked to-device position for each client.
2022-12-14 18:53:55 +00:00
assertNoError(t, err)
if len(gotMsgs) > 0 {
t.Errorf("Got %+v want nothing", jsonArrStr(gotMsgs))
}
}
// Test that unacked events are safe from deletion
func TestToDeviceTableNoDeleteUnacks(t *testing.T) {
db, close := connectToDB(t)
defer close()
add extensions for typing and receipts; bugfixes and additional perf improvements Features: - Add `typing` extension. - Add `receipts` extension. - Add comprehensive prometheus `/metrics` activated via `SYNCV3_PROM`. - Add `SYNCV3_PPROF` support. - Add `by_notification_level` sort order. - Add `include_old_rooms` support. - Add support for `$ME` and `$LAZY`. - Add correct filtering when `*,*` is used as `required_state`. - Add `num_live` to each room response to indicate how many timeline entries are live. Bug fixes: - Use a stricter comparison function on ranges: fixes an issue whereby UTs fail on go1.19 due to change in sorting algorithm. - Send back an `errcode` on HTTP errors (e.g expired sessions). - Remove `unsigned.txn_id` on insertion into the DB. Otherwise other users would see other users txn IDs :( - Improve range delta algorithm: previously it didn't handle cases like `[0,20] -> [20,30]` and would panic. - Send HTTP 400 for invalid range requests. - Don't publish no-op unread counts which just adds extra noise. - Fix leaking DB connections which could eventually consume all available connections. - Ensure we always unblock WaitUntilInitialSync even on invalid access tokens. Other code relies on WaitUntilInitialSync() actually returning at _some_ point e.g on startup we have N workers which bound the number of concurrent pollers made at any one time, we need to not just hog a worker forever. Improvements: - Greatly improve startup times of sync3 handlers by improving `JoinedRoomsTracker`: a modest amount of data would take ~28s to create the handler, now it takes 4s. - Massively improve initial initial v3 sync times, by refactoring `JoinedRoomsTracker`, from ~47s to <1s. - Add `SlidingSyncUntil...` in tests to reduce races. - Tweak the API shape of JoinedUsersForRoom to reduce state block processing time for large rooms from 63s to 39s. - Add trace task for initial syncs. - Include the proxy version in UA strings. - HTTP errors now wait 1s before returning to stop clients tight-looping on error. - Pending event buffer is now 2000. - Index the room ID first to cull the most events when returning timeline entries. Speeds up `SelectLatestEventsBetween` by a factor of 8. - Remove cancelled `m.room_key_requests` from the to-device inbox. Cuts down the amount of events in the inbox by ~94% for very large (20k+) inboxes, ~50% for moderate sized (200 events) inboxes. Adds book-keeping to remember the unacked to-device position for each client.
2022-12-14 18:53:55 +00:00
sender := "SENDER2"
destination := "DEST2"
table := NewToDeviceTable(db)
// insert request
reqEv := newRoomKeyEvent(t, "request", "1", sender, map[string]interface{}{
"foo": "bar",
})
2023-05-02 14:34:22 +01:00
pos, err := table.InsertMessages(sender, destination, []json.RawMessage{reqEv})
add extensions for typing and receipts; bugfixes and additional perf improvements Features: - Add `typing` extension. - Add `receipts` extension. - Add comprehensive prometheus `/metrics` activated via `SYNCV3_PROM`. - Add `SYNCV3_PPROF` support. - Add `by_notification_level` sort order. - Add `include_old_rooms` support. - Add support for `$ME` and `$LAZY`. - Add correct filtering when `*,*` is used as `required_state`. - Add `num_live` to each room response to indicate how many timeline entries are live. Bug fixes: - Use a stricter comparison function on ranges: fixes an issue whereby UTs fail on go1.19 due to change in sorting algorithm. - Send back an `errcode` on HTTP errors (e.g expired sessions). - Remove `unsigned.txn_id` on insertion into the DB. Otherwise other users would see other users txn IDs :( - Improve range delta algorithm: previously it didn't handle cases like `[0,20] -> [20,30]` and would panic. - Send HTTP 400 for invalid range requests. - Don't publish no-op unread counts which just adds extra noise. - Fix leaking DB connections which could eventually consume all available connections. - Ensure we always unblock WaitUntilInitialSync even on invalid access tokens. Other code relies on WaitUntilInitialSync() actually returning at _some_ point e.g on startup we have N workers which bound the number of concurrent pollers made at any one time, we need to not just hog a worker forever. Improvements: - Greatly improve startup times of sync3 handlers by improving `JoinedRoomsTracker`: a modest amount of data would take ~28s to create the handler, now it takes 4s. - Massively improve initial initial v3 sync times, by refactoring `JoinedRoomsTracker`, from ~47s to <1s. - Add `SlidingSyncUntil...` in tests to reduce races. - Tweak the API shape of JoinedUsersForRoom to reduce state block processing time for large rooms from 63s to 39s. - Add trace task for initial syncs. - Include the proxy version in UA strings. - HTTP errors now wait 1s before returning to stop clients tight-looping on error. - Pending event buffer is now 2000. - Index the room ID first to cull the most events when returning timeline entries. Speeds up `SelectLatestEventsBetween` by a factor of 8. - Remove cancelled `m.room_key_requests` from the to-device inbox. Cuts down the amount of events in the inbox by ~94% for very large (20k+) inboxes, ~50% for moderate sized (200 events) inboxes. Adds book-keeping to remember the unacked to-device position for each client.
2022-12-14 18:53:55 +00:00
assertNoError(t, err)
// mark this position as unacked: this means the client MAY know about this request so it isn't
// safe to delete it
2023-05-02 14:34:22 +01:00
err = table.SetUnackedPosition(sender, destination, pos)
add extensions for typing and receipts; bugfixes and additional perf improvements Features: - Add `typing` extension. - Add `receipts` extension. - Add comprehensive prometheus `/metrics` activated via `SYNCV3_PROM`. - Add `SYNCV3_PPROF` support. - Add `by_notification_level` sort order. - Add `include_old_rooms` support. - Add support for `$ME` and `$LAZY`. - Add correct filtering when `*,*` is used as `required_state`. - Add `num_live` to each room response to indicate how many timeline entries are live. Bug fixes: - Use a stricter comparison function on ranges: fixes an issue whereby UTs fail on go1.19 due to change in sorting algorithm. - Send back an `errcode` on HTTP errors (e.g expired sessions). - Remove `unsigned.txn_id` on insertion into the DB. Otherwise other users would see other users txn IDs :( - Improve range delta algorithm: previously it didn't handle cases like `[0,20] -> [20,30]` and would panic. - Send HTTP 400 for invalid range requests. - Don't publish no-op unread counts which just adds extra noise. - Fix leaking DB connections which could eventually consume all available connections. - Ensure we always unblock WaitUntilInitialSync even on invalid access tokens. Other code relies on WaitUntilInitialSync() actually returning at _some_ point e.g on startup we have N workers which bound the number of concurrent pollers made at any one time, we need to not just hog a worker forever. Improvements: - Greatly improve startup times of sync3 handlers by improving `JoinedRoomsTracker`: a modest amount of data would take ~28s to create the handler, now it takes 4s. - Massively improve initial initial v3 sync times, by refactoring `JoinedRoomsTracker`, from ~47s to <1s. - Add `SlidingSyncUntil...` in tests to reduce races. - Tweak the API shape of JoinedUsersForRoom to reduce state block processing time for large rooms from 63s to 39s. - Add trace task for initial syncs. - Include the proxy version in UA strings. - HTTP errors now wait 1s before returning to stop clients tight-looping on error. - Pending event buffer is now 2000. - Index the room ID first to cull the most events when returning timeline entries. Speeds up `SelectLatestEventsBetween` by a factor of 8. - Remove cancelled `m.room_key_requests` from the to-device inbox. Cuts down the amount of events in the inbox by ~94% for very large (20k+) inboxes, ~50% for moderate sized (200 events) inboxes. Adds book-keeping to remember the unacked to-device position for each client.
2022-12-14 18:53:55 +00:00
assertNoError(t, err)
// now issue a cancellation: this should NOT result in a cancellation due to protection for unacked events
cancelEv := newRoomKeyEvent(t, "request_cancellation", "1", sender, nil)
2023-05-02 14:34:22 +01:00
_, err = table.InsertMessages(sender, destination, []json.RawMessage{cancelEv})
add extensions for typing and receipts; bugfixes and additional perf improvements Features: - Add `typing` extension. - Add `receipts` extension. - Add comprehensive prometheus `/metrics` activated via `SYNCV3_PROM`. - Add `SYNCV3_PPROF` support. - Add `by_notification_level` sort order. - Add `include_old_rooms` support. - Add support for `$ME` and `$LAZY`. - Add correct filtering when `*,*` is used as `required_state`. - Add `num_live` to each room response to indicate how many timeline entries are live. Bug fixes: - Use a stricter comparison function on ranges: fixes an issue whereby UTs fail on go1.19 due to change in sorting algorithm. - Send back an `errcode` on HTTP errors (e.g expired sessions). - Remove `unsigned.txn_id` on insertion into the DB. Otherwise other users would see other users txn IDs :( - Improve range delta algorithm: previously it didn't handle cases like `[0,20] -> [20,30]` and would panic. - Send HTTP 400 for invalid range requests. - Don't publish no-op unread counts which just adds extra noise. - Fix leaking DB connections which could eventually consume all available connections. - Ensure we always unblock WaitUntilInitialSync even on invalid access tokens. Other code relies on WaitUntilInitialSync() actually returning at _some_ point e.g on startup we have N workers which bound the number of concurrent pollers made at any one time, we need to not just hog a worker forever. Improvements: - Greatly improve startup times of sync3 handlers by improving `JoinedRoomsTracker`: a modest amount of data would take ~28s to create the handler, now it takes 4s. - Massively improve initial initial v3 sync times, by refactoring `JoinedRoomsTracker`, from ~47s to <1s. - Add `SlidingSyncUntil...` in tests to reduce races. - Tweak the API shape of JoinedUsersForRoom to reduce state block processing time for large rooms from 63s to 39s. - Add trace task for initial syncs. - Include the proxy version in UA strings. - HTTP errors now wait 1s before returning to stop clients tight-looping on error. - Pending event buffer is now 2000. - Index the room ID first to cull the most events when returning timeline entries. Speeds up `SelectLatestEventsBetween` by a factor of 8. - Remove cancelled `m.room_key_requests` from the to-device inbox. Cuts down the amount of events in the inbox by ~94% for very large (20k+) inboxes, ~50% for moderate sized (200 events) inboxes. Adds book-keeping to remember the unacked to-device position for each client.
2022-12-14 18:53:55 +00:00
assertNoError(t, err)
// selecting messages returns both events
2023-05-02 14:34:22 +01:00
gotMsgs, _, err := table.Messages(sender, destination, 0, 10)
add extensions for typing and receipts; bugfixes and additional perf improvements Features: - Add `typing` extension. - Add `receipts` extension. - Add comprehensive prometheus `/metrics` activated via `SYNCV3_PROM`. - Add `SYNCV3_PPROF` support. - Add `by_notification_level` sort order. - Add `include_old_rooms` support. - Add support for `$ME` and `$LAZY`. - Add correct filtering when `*,*` is used as `required_state`. - Add `num_live` to each room response to indicate how many timeline entries are live. Bug fixes: - Use a stricter comparison function on ranges: fixes an issue whereby UTs fail on go1.19 due to change in sorting algorithm. - Send back an `errcode` on HTTP errors (e.g expired sessions). - Remove `unsigned.txn_id` on insertion into the DB. Otherwise other users would see other users txn IDs :( - Improve range delta algorithm: previously it didn't handle cases like `[0,20] -> [20,30]` and would panic. - Send HTTP 400 for invalid range requests. - Don't publish no-op unread counts which just adds extra noise. - Fix leaking DB connections which could eventually consume all available connections. - Ensure we always unblock WaitUntilInitialSync even on invalid access tokens. Other code relies on WaitUntilInitialSync() actually returning at _some_ point e.g on startup we have N workers which bound the number of concurrent pollers made at any one time, we need to not just hog a worker forever. Improvements: - Greatly improve startup times of sync3 handlers by improving `JoinedRoomsTracker`: a modest amount of data would take ~28s to create the handler, now it takes 4s. - Massively improve initial initial v3 sync times, by refactoring `JoinedRoomsTracker`, from ~47s to <1s. - Add `SlidingSyncUntil...` in tests to reduce races. - Tweak the API shape of JoinedUsersForRoom to reduce state block processing time for large rooms from 63s to 39s. - Add trace task for initial syncs. - Include the proxy version in UA strings. - HTTP errors now wait 1s before returning to stop clients tight-looping on error. - Pending event buffer is now 2000. - Index the room ID first to cull the most events when returning timeline entries. Speeds up `SelectLatestEventsBetween` by a factor of 8. - Remove cancelled `m.room_key_requests` from the to-device inbox. Cuts down the amount of events in the inbox by ~94% for very large (20k+) inboxes, ~50% for moderate sized (200 events) inboxes. Adds book-keeping to remember the unacked to-device position for each client.
2022-12-14 18:53:55 +00:00
assertNoError(t, err)
if len(gotMsgs) != 2 {
t.Fatalf("got %d msgs, want 2: %v", len(gotMsgs), jsonArrStr(gotMsgs))
}
bytesEqual(t, gotMsgs[0], reqEv)
bytesEqual(t, gotMsgs[1], cancelEv)
// test that injecting another req/cancel does cause them to be deleted
2023-05-02 14:34:22 +01:00
_, err = table.InsertMessages(sender, destination, []json.RawMessage{newRoomKeyEvent(t, "request", "2", sender, map[string]interface{}{
add extensions for typing and receipts; bugfixes and additional perf improvements Features: - Add `typing` extension. - Add `receipts` extension. - Add comprehensive prometheus `/metrics` activated via `SYNCV3_PROM`. - Add `SYNCV3_PPROF` support. - Add `by_notification_level` sort order. - Add `include_old_rooms` support. - Add support for `$ME` and `$LAZY`. - Add correct filtering when `*,*` is used as `required_state`. - Add `num_live` to each room response to indicate how many timeline entries are live. Bug fixes: - Use a stricter comparison function on ranges: fixes an issue whereby UTs fail on go1.19 due to change in sorting algorithm. - Send back an `errcode` on HTTP errors (e.g expired sessions). - Remove `unsigned.txn_id` on insertion into the DB. Otherwise other users would see other users txn IDs :( - Improve range delta algorithm: previously it didn't handle cases like `[0,20] -> [20,30]` and would panic. - Send HTTP 400 for invalid range requests. - Don't publish no-op unread counts which just adds extra noise. - Fix leaking DB connections which could eventually consume all available connections. - Ensure we always unblock WaitUntilInitialSync even on invalid access tokens. Other code relies on WaitUntilInitialSync() actually returning at _some_ point e.g on startup we have N workers which bound the number of concurrent pollers made at any one time, we need to not just hog a worker forever. Improvements: - Greatly improve startup times of sync3 handlers by improving `JoinedRoomsTracker`: a modest amount of data would take ~28s to create the handler, now it takes 4s. - Massively improve initial initial v3 sync times, by refactoring `JoinedRoomsTracker`, from ~47s to <1s. - Add `SlidingSyncUntil...` in tests to reduce races. - Tweak the API shape of JoinedUsersForRoom to reduce state block processing time for large rooms from 63s to 39s. - Add trace task for initial syncs. - Include the proxy version in UA strings. - HTTP errors now wait 1s before returning to stop clients tight-looping on error. - Pending event buffer is now 2000. - Index the room ID first to cull the most events when returning timeline entries. Speeds up `SelectLatestEventsBetween` by a factor of 8. - Remove cancelled `m.room_key_requests` from the to-device inbox. Cuts down the amount of events in the inbox by ~94% for very large (20k+) inboxes, ~50% for moderate sized (200 events) inboxes. Adds book-keeping to remember the unacked to-device position for each client.
2022-12-14 18:53:55 +00:00
"foo": "bar",
})})
assertNoError(t, err)
2023-05-02 14:34:22 +01:00
_, err = table.InsertMessages(sender, destination, []json.RawMessage{newRoomKeyEvent(t, "request_cancellation", "2", sender, nil)})
add extensions for typing and receipts; bugfixes and additional perf improvements Features: - Add `typing` extension. - Add `receipts` extension. - Add comprehensive prometheus `/metrics` activated via `SYNCV3_PROM`. - Add `SYNCV3_PPROF` support. - Add `by_notification_level` sort order. - Add `include_old_rooms` support. - Add support for `$ME` and `$LAZY`. - Add correct filtering when `*,*` is used as `required_state`. - Add `num_live` to each room response to indicate how many timeline entries are live. Bug fixes: - Use a stricter comparison function on ranges: fixes an issue whereby UTs fail on go1.19 due to change in sorting algorithm. - Send back an `errcode` on HTTP errors (e.g expired sessions). - Remove `unsigned.txn_id` on insertion into the DB. Otherwise other users would see other users txn IDs :( - Improve range delta algorithm: previously it didn't handle cases like `[0,20] -> [20,30]` and would panic. - Send HTTP 400 for invalid range requests. - Don't publish no-op unread counts which just adds extra noise. - Fix leaking DB connections which could eventually consume all available connections. - Ensure we always unblock WaitUntilInitialSync even on invalid access tokens. Other code relies on WaitUntilInitialSync() actually returning at _some_ point e.g on startup we have N workers which bound the number of concurrent pollers made at any one time, we need to not just hog a worker forever. Improvements: - Greatly improve startup times of sync3 handlers by improving `JoinedRoomsTracker`: a modest amount of data would take ~28s to create the handler, now it takes 4s. - Massively improve initial initial v3 sync times, by refactoring `JoinedRoomsTracker`, from ~47s to <1s. - Add `SlidingSyncUntil...` in tests to reduce races. - Tweak the API shape of JoinedUsersForRoom to reduce state block processing time for large rooms from 63s to 39s. - Add trace task for initial syncs. - Include the proxy version in UA strings. - HTTP errors now wait 1s before returning to stop clients tight-looping on error. - Pending event buffer is now 2000. - Index the room ID first to cull the most events when returning timeline entries. Speeds up `SelectLatestEventsBetween` by a factor of 8. - Remove cancelled `m.room_key_requests` from the to-device inbox. Cuts down the amount of events in the inbox by ~94% for very large (20k+) inboxes, ~50% for moderate sized (200 events) inboxes. Adds book-keeping to remember the unacked to-device position for each client.
2022-12-14 18:53:55 +00:00
assertNoError(t, err)
// selecting messages returns the same as before
2023-05-02 14:34:22 +01:00
gotMsgs, _, err = table.Messages(sender, destination, 0, 10)
add extensions for typing and receipts; bugfixes and additional perf improvements Features: - Add `typing` extension. - Add `receipts` extension. - Add comprehensive prometheus `/metrics` activated via `SYNCV3_PROM`. - Add `SYNCV3_PPROF` support. - Add `by_notification_level` sort order. - Add `include_old_rooms` support. - Add support for `$ME` and `$LAZY`. - Add correct filtering when `*,*` is used as `required_state`. - Add `num_live` to each room response to indicate how many timeline entries are live. Bug fixes: - Use a stricter comparison function on ranges: fixes an issue whereby UTs fail on go1.19 due to change in sorting algorithm. - Send back an `errcode` on HTTP errors (e.g expired sessions). - Remove `unsigned.txn_id` on insertion into the DB. Otherwise other users would see other users txn IDs :( - Improve range delta algorithm: previously it didn't handle cases like `[0,20] -> [20,30]` and would panic. - Send HTTP 400 for invalid range requests. - Don't publish no-op unread counts which just adds extra noise. - Fix leaking DB connections which could eventually consume all available connections. - Ensure we always unblock WaitUntilInitialSync even on invalid access tokens. Other code relies on WaitUntilInitialSync() actually returning at _some_ point e.g on startup we have N workers which bound the number of concurrent pollers made at any one time, we need to not just hog a worker forever. Improvements: - Greatly improve startup times of sync3 handlers by improving `JoinedRoomsTracker`: a modest amount of data would take ~28s to create the handler, now it takes 4s. - Massively improve initial initial v3 sync times, by refactoring `JoinedRoomsTracker`, from ~47s to <1s. - Add `SlidingSyncUntil...` in tests to reduce races. - Tweak the API shape of JoinedUsersForRoom to reduce state block processing time for large rooms from 63s to 39s. - Add trace task for initial syncs. - Include the proxy version in UA strings. - HTTP errors now wait 1s before returning to stop clients tight-looping on error. - Pending event buffer is now 2000. - Index the room ID first to cull the most events when returning timeline entries. Speeds up `SelectLatestEventsBetween` by a factor of 8. - Remove cancelled `m.room_key_requests` from the to-device inbox. Cuts down the amount of events in the inbox by ~94% for very large (20k+) inboxes, ~50% for moderate sized (200 events) inboxes. Adds book-keeping to remember the unacked to-device position for each client.
2022-12-14 18:53:55 +00:00
assertNoError(t, err)
if len(gotMsgs) != 2 {
t.Fatalf("got %d msgs, want 2: %v", len(gotMsgs), jsonArrStr(gotMsgs))
}
bytesEqual(t, gotMsgs[0], reqEv)
bytesEqual(t, gotMsgs[1], cancelEv)
}
add extensions for typing and receipts; bugfixes and additional perf improvements Features: - Add `typing` extension. - Add `receipts` extension. - Add comprehensive prometheus `/metrics` activated via `SYNCV3_PROM`. - Add `SYNCV3_PPROF` support. - Add `by_notification_level` sort order. - Add `include_old_rooms` support. - Add support for `$ME` and `$LAZY`. - Add correct filtering when `*,*` is used as `required_state`. - Add `num_live` to each room response to indicate how many timeline entries are live. Bug fixes: - Use a stricter comparison function on ranges: fixes an issue whereby UTs fail on go1.19 due to change in sorting algorithm. - Send back an `errcode` on HTTP errors (e.g expired sessions). - Remove `unsigned.txn_id` on insertion into the DB. Otherwise other users would see other users txn IDs :( - Improve range delta algorithm: previously it didn't handle cases like `[0,20] -> [20,30]` and would panic. - Send HTTP 400 for invalid range requests. - Don't publish no-op unread counts which just adds extra noise. - Fix leaking DB connections which could eventually consume all available connections. - Ensure we always unblock WaitUntilInitialSync even on invalid access tokens. Other code relies on WaitUntilInitialSync() actually returning at _some_ point e.g on startup we have N workers which bound the number of concurrent pollers made at any one time, we need to not just hog a worker forever. Improvements: - Greatly improve startup times of sync3 handlers by improving `JoinedRoomsTracker`: a modest amount of data would take ~28s to create the handler, now it takes 4s. - Massively improve initial initial v3 sync times, by refactoring `JoinedRoomsTracker`, from ~47s to <1s. - Add `SlidingSyncUntil...` in tests to reduce races. - Tweak the API shape of JoinedUsersForRoom to reduce state block processing time for large rooms from 63s to 39s. - Add trace task for initial syncs. - Include the proxy version in UA strings. - HTTP errors now wait 1s before returning to stop clients tight-looping on error. - Pending event buffer is now 2000. - Index the room ID first to cull the most events when returning timeline entries. Speeds up `SelectLatestEventsBetween` by a factor of 8. - Remove cancelled `m.room_key_requests` from the to-device inbox. Cuts down the amount of events in the inbox by ~94% for very large (20k+) inboxes, ~50% for moderate sized (200 events) inboxes. Adds book-keeping to remember the unacked to-device position for each client.
2022-12-14 18:53:55 +00:00
// Guard against possible message truncation?
func TestToDeviceTableBytesInEqualBytesOut(t *testing.T) {
db, close := connectToDB(t)
defer close()
2023-05-02 14:34:22 +01:00
sender := "@sendymcsendface:localhost"
table := NewToDeviceTable(db)
testCases := []json.RawMessage{
json.RawMessage(`{}`),
json.RawMessage(`{"foo":"bar"}`),
json.RawMessage(`{ "foo": "bar" }`),
json.RawMessage(`{ not even valid json :D }`),
2022-12-21 10:58:57 +00:00
json.RawMessage(`{ "\~./.-$%_!@£?;'\[]= }`),
}
var pos int64
for _, msg := range testCases {
2023-05-02 14:34:22 +01:00
nextPos, err := table.InsertMessages(sender, "A", []json.RawMessage{msg})
if err != nil {
t.Fatalf("InsertMessages: %s", err)
}
2023-05-02 14:34:22 +01:00
got, _, err := table.Messages(sender, "A", pos, 1)
if err != nil {
t.Fatalf("Messages: %s", err)
}
bytesEqual(t, got[0], msg)
pos = nextPos
}
2022-12-21 10:56:25 +00:00
// and all at once
2023-05-02 14:34:22 +01:00
_, err := table.InsertMessages(sender, "B", testCases)
2022-12-21 10:56:25 +00:00
if err != nil {
t.Fatalf("InsertMessages: %s", err)
}
2023-05-02 14:34:22 +01:00
got, _, err := table.Messages(sender, "B", 0, 100)
2022-12-21 10:56:25 +00:00
if err != nil {
t.Fatalf("Messages: %s", err)
}
if len(got) != len(testCases) {
t.Fatalf("got %d messages, want %d", len(got), len(testCases))
}
for i := range testCases {
bytesEqual(t, got[i], testCases[i])
}
add extensions for typing and receipts; bugfixes and additional perf improvements Features: - Add `typing` extension. - Add `receipts` extension. - Add comprehensive prometheus `/metrics` activated via `SYNCV3_PROM`. - Add `SYNCV3_PPROF` support. - Add `by_notification_level` sort order. - Add `include_old_rooms` support. - Add support for `$ME` and `$LAZY`. - Add correct filtering when `*,*` is used as `required_state`. - Add `num_live` to each room response to indicate how many timeline entries are live. Bug fixes: - Use a stricter comparison function on ranges: fixes an issue whereby UTs fail on go1.19 due to change in sorting algorithm. - Send back an `errcode` on HTTP errors (e.g expired sessions). - Remove `unsigned.txn_id` on insertion into the DB. Otherwise other users would see other users txn IDs :( - Improve range delta algorithm: previously it didn't handle cases like `[0,20] -> [20,30]` and would panic. - Send HTTP 400 for invalid range requests. - Don't publish no-op unread counts which just adds extra noise. - Fix leaking DB connections which could eventually consume all available connections. - Ensure we always unblock WaitUntilInitialSync even on invalid access tokens. Other code relies on WaitUntilInitialSync() actually returning at _some_ point e.g on startup we have N workers which bound the number of concurrent pollers made at any one time, we need to not just hog a worker forever. Improvements: - Greatly improve startup times of sync3 handlers by improving `JoinedRoomsTracker`: a modest amount of data would take ~28s to create the handler, now it takes 4s. - Massively improve initial initial v3 sync times, by refactoring `JoinedRoomsTracker`, from ~47s to <1s. - Add `SlidingSyncUntil...` in tests to reduce races. - Tweak the API shape of JoinedUsersForRoom to reduce state block processing time for large rooms from 63s to 39s. - Add trace task for initial syncs. - Include the proxy version in UA strings. - HTTP errors now wait 1s before returning to stop clients tight-looping on error. - Pending event buffer is now 2000. - Index the room ID first to cull the most events when returning timeline entries. Speeds up `SelectLatestEventsBetween` by a factor of 8. - Remove cancelled `m.room_key_requests` from the to-device inbox. Cuts down the amount of events in the inbox by ~94% for very large (20k+) inboxes, ~50% for moderate sized (200 events) inboxes. Adds book-keeping to remember the unacked to-device position for each client.
2022-12-14 18:53:55 +00:00
}
func TestMsgID(t *testing.T) {
data := json.RawMessage(`{
"content": {
"algorithm": "m.olm.v1.curve25519-aes-sha2",
"ciphertext": {
"gMObR+/4dqL5T4DisRRRYBJpn+OjzFnkyCFOktP6Eyw": {
"body": "AwogrdbTbG8VCW....slqU",
"type": 0
}
},
"org.matrix.msgid": "6390a372-fd3c-4f56-b0d5-2f2ce39f2d56",
"sender_key": "EWnYTm/yIQ1lStSIqO6fdVYvS69OfU2DzrX+q+1d+w8"
},
"type": "m.room.encrypted",
"sender": "@sample:localhost:8480"
}`)
m := gjson.ParseBytes(data)
got := m.Get(`content.org\.matrix\.msgid`).Str
want := "6390a372-fd3c-4f56-b0d5-2f2ce39f2d56"
if got != want {
t.Fatalf("got %v want %v", got, want)
}
}
add extensions for typing and receipts; bugfixes and additional perf improvements Features: - Add `typing` extension. - Add `receipts` extension. - Add comprehensive prometheus `/metrics` activated via `SYNCV3_PROM`. - Add `SYNCV3_PPROF` support. - Add `by_notification_level` sort order. - Add `include_old_rooms` support. - Add support for `$ME` and `$LAZY`. - Add correct filtering when `*,*` is used as `required_state`. - Add `num_live` to each room response to indicate how many timeline entries are live. Bug fixes: - Use a stricter comparison function on ranges: fixes an issue whereby UTs fail on go1.19 due to change in sorting algorithm. - Send back an `errcode` on HTTP errors (e.g expired sessions). - Remove `unsigned.txn_id` on insertion into the DB. Otherwise other users would see other users txn IDs :( - Improve range delta algorithm: previously it didn't handle cases like `[0,20] -> [20,30]` and would panic. - Send HTTP 400 for invalid range requests. - Don't publish no-op unread counts which just adds extra noise. - Fix leaking DB connections which could eventually consume all available connections. - Ensure we always unblock WaitUntilInitialSync even on invalid access tokens. Other code relies on WaitUntilInitialSync() actually returning at _some_ point e.g on startup we have N workers which bound the number of concurrent pollers made at any one time, we need to not just hog a worker forever. Improvements: - Greatly improve startup times of sync3 handlers by improving `JoinedRoomsTracker`: a modest amount of data would take ~28s to create the handler, now it takes 4s. - Massively improve initial initial v3 sync times, by refactoring `JoinedRoomsTracker`, from ~47s to <1s. - Add `SlidingSyncUntil...` in tests to reduce races. - Tweak the API shape of JoinedUsersForRoom to reduce state block processing time for large rooms from 63s to 39s. - Add trace task for initial syncs. - Include the proxy version in UA strings. - HTTP errors now wait 1s before returning to stop clients tight-looping on error. - Pending event buffer is now 2000. - Index the room ID first to cull the most events when returning timeline entries. Speeds up `SelectLatestEventsBetween` by a factor of 8. - Remove cancelled `m.room_key_requests` from the to-device inbox. Cuts down the amount of events in the inbox by ~94% for very large (20k+) inboxes, ~50% for moderate sized (200 events) inboxes. Adds book-keeping to remember the unacked to-device position for each client.
2022-12-14 18:53:55 +00:00
func bytesEqual(t *testing.T, got, want json.RawMessage) {
t.Helper()
if !bytes.Equal(got, want) {
t.Fatalf("bytesEqual: \ngot %s\n want %s", string(got), string(want))
}
}
type roomKeyRequest struct {
Type string `json:"type"`
Content roomKeyRequestContent `json:"content"`
}
type roomKeyRequestContent struct {
Action string `json:"action"`
RequestID string `json:"request_id"`
RequestingDeviceID string `json:"requesting_device_id"`
Body map[string]interface{} `json:"body,omitempty"`
}
func newRoomKeyEvent(t *testing.T, action, reqID, reqDeviceID string, body map[string]interface{}) json.RawMessage {
rkr := roomKeyRequest{
Type: "m.room_key_request",
Content: roomKeyRequestContent{
Action: action,
RequestID: reqID,
RequestingDeviceID: reqDeviceID,
Body: body,
},
}
b, err := json.Marshal(rkr)
if err != nil {
t.Fatalf("newRoomKeyEvent: %s", err)
}
return json.RawMessage(b)
}