Merge branch 'develop' of into allow_auto_join_rooms

CHANGES.rst

@@ -1,11 +1,89 @@
-Unreleased
+Changes in synapse v0.27.2 (2018-03-26)
-==========
+=======================================
 
-synctl no longer starts the main synapse when using ``-a`` option with workers.
+Bug fixes:
-A new worker file should be added with ``worker_app: synapse.app.homeserver``.
+
+* Fix bug which broke TCP replication between workers (PR #3015)
+
+
+Changes in synapse v0.27.1 (2018-03-26)
+=======================================
+
+Meta release as v0.27.0 temporarily pointed to the wrong commit
+
+
+Changes in synapse v0.27.0 (2018-03-26)
+=======================================
+
+No changes since v0.27.0-rc2
+
+
+Changes in synapse v0.27.0-rc2 (2018-03-19)
+===========================================
+
+Pulls in v0.26.1
+
+Bug fixes:
+
+* Fix bug introduced in v0.27.0-rc1 that causes much increased memory usage in state cache (PR #3005)
+
+
+Changes in synapse v0.26.1 (2018-03-15)
+=======================================
+
+Bug fixes:
+
+* Fix bug where an invalid event caused server to stop functioning correctly,
+  due to parsing and serializing bugs in ujson library (PR #3008)
+
+
+Changes in synapse v0.27.0-rc1 (2018-03-14)
+===========================================
+
+The common case for running Synapse is not to run separate workers, but for those that do, be aware that synctl no longer starts the main synapse when using ``-a`` option with workers. A new worker file should be added with ``worker_app: synapse.app.homeserver``.
 
 This release also begins the process of renaming a number of the metrics
 reported to prometheus. See `docs/metrics-howto.rst <docs/metrics-howto.rst#block-and-response-metrics-renamed-for-0-27-0>`_.
+Note that the v0.28.0 release will remove the deprecated metric names.
+
+Features:
+
+* Add ability for ASes to override message send time (PR #2754)
+* Add support for custom storage providers for media repository (PR #2867, #2777, #2783, #2789, #2791, #2804, #2812, #2814, #2857, #2868, #2767)
+* Add purge API features, see `docs/admin_api/purge_history_api.rst <docs/admin_api/purge_history_api.rst>`_ for full details (PR #2858, #2867, #2882, #2946, #2962, #2943)
+* Add support for whitelisting 3PIDs that users can register. (PR #2813)
+* Add ``/room/{id}/event/{id}`` API (PR #2766)
+* Add an admin API to get all the media in a room (PR #2818) Thanks to @turt2live!
+* Add ``federation_domain_whitelist`` option (PR #2820, #2821)
+
+
+Changes:
+
+* Continue to factor out processing from main process and into worker processes. See updated `docs/workers.rst <docs/workers.rst>`_ (PR #2892 - #2904, #2913, #2920 - #2926, #2947, #2847, #2854, #2872, #2873, #2874, #2928, #2929, #2934, #2856, #2976 - #2984, #2987 - #2989, #2991 - #2993, #2995, #2784)
+* Ensure state cache is used when persisting events (PR #2864, #2871, #2802, #2835, #2836, #2841, #2842, #2849)
+* Change the default config to bind on both IPv4 and IPv6 on all platforms (PR #2435) Thanks to @silkeh!
+* No longer require a specific version of saml2 (PR #2695) Thanks to @okurz!
+* Remove ``verbosity``/``log_file`` from generated config (PR #2755)
+* Add and improve metrics and logging (PR #2770, #2778, #2785, #2786, #2787, #2793, #2794, #2795, #2809, #2810, #2833, #2834, #2844, #2965, #2927, #2975, #2790, #2796, #2838)
+* When using synctl with workers, don't start the main synapse automatically (PR #2774)
+* Minor performance improvements (PR #2773, #2792)
+* Use a connection pool for non-federation outbound connections (PR #2817)
+* Make it possible to run unit tests against postgres (PR #2829)
+* Update pynacl dependency to 1.2.1 or higher (PR #2888) Thanks to @bachp!
+* Remove ability for AS users to call /events and /sync (PR #2948)
+* Use bcrypt.checkpw (PR #2949) Thanks to @krombel!
+
+Bug fixes:
+
+* Fix broken ``ldap_config`` config option (PR #2683) Thanks to @seckrv!
+* Fix error message when user is not allowed to unban (PR #2761) Thanks to @turt2live!
+* Fix publicised groups GET API (singular) over federation (PR #2772)
+* Fix user directory when using ``user_directory_search_all_users`` config option (PR #2803, #2831)
+* Fix error on ``/publicRooms`` when no rooms exist (PR #2827)
+* Fix bug in quarantine_media (PR #2837)
+* Fix url_previews when no Content-Type is returned from URL (PR #2845)
+* Fix rare race in sync API when joining room (PR #2944)
+* Fix slow event search, switch back from GIST to GIN indexes (PR #2769, #2848)
 
 
 Changes in synapse v0.26.0 (2018-01-05)

CONTRIBUTING.rst

@@ -30,8 +30,12 @@ use github's pull request workflow to review the contribution, and either ask
 you to make any refinements needed or merge it and make them ourselves. The
 changes will then land on master when we next do a release.
 
-We use Jenkins for continuous integration (http://matrix.org/jenkins), and
+We use `Jenkins <http://matrix.org/jenkins>`_ and 
-typically all pull requests get automatically tested Jenkins: if your change breaks the build, Jenkins will yell about it in #matrix-dev:matrix.org so please lurk there and keep an eye open.
+`Travis <https://travis-ci.org/matrix-org/synapse>`_ for continuous
+integration. All pull requests to synapse get automatically tested by Travis; 
+the Jenkins builds require an adminstrator to start them. If your change 
+breaks the build, this will be shown in github, so please keep an eye on the 
+pull request for feedback.
 
 Code style
 ~~~~~~~~~~
@@ -115,4 +119,4 @@ can't be accepted. Git makes this trivial - just use the -s flag when you do
 Conclusion
 ~~~~~~~~~~
 
-That's it!  Matrix is a very open and collaborative project as you might expect given our obsession with open communication.  If we're going to successfully matrix together all the fragmented communication technologies out there we are reliant on contributions and collaboration from the community to do so.  So please get involved - and we hope you have as much fun hacking on Matrix as we do!
+That's it!  Matrix is a very open and collaborative project as you might expect given our obsession with open communication.  If we're going to successfully matrix together all the fragmented communication technologies out there we are reliant on contributions and collaboration from the community to do so.  So please get involved - and we hope you have as much fun hacking on Matrix as we do!

README.rst

@@ -354,6 +354,10 @@ https://matrix.org/docs/projects/try-matrix-now.html (or build your own with one
 Fedora
 ------
 
+Synapse is in the Fedora repositories as ``matrix-synapse``::
+
+    sudo dnf install matrix-synapse
+
 Oleg Girko provides Fedora RPMs at
 https://obs.infoserver.lv/project/monitor/matrix-synapse
 
@@ -890,6 +894,17 @@ This should end with a 'PASSED' result::
 
     PASSED (successes=143)
 
+Running the Integration Tests
+=============================
+
+Synapse is accompanied by `SyTest <https://github.com/matrix-org/sytest>`_,
+a Matrix homeserver integration testing suite, which uses HTTP requests to
+access the API as a Matrix client would. It is able to run Synapse directly from
+the source tree, so installation of the server is not required.
+
+Testing with SyTest is recommended for verifying that changes related to the
+Client-Server API are functioning correctly. See the `installation instructions
+<https://github.com/matrix-org/sytest#installing>`_ for details.
 
 Building Internal API Documentation
 ===================================

docs/admin_api/purge_history_api.rst

@@ -16,9 +16,11 @@ including an ``access_token`` of a server admin.
 
 By default, events sent by local users are not deleted, as they may represent
 the only copies of this content in existence. (Events sent by remote users are
-deleted, and room state data before the cutoff is always removed).
+deleted.)
 
-To delete local events as well, set ``delete_local_events`` in the body:
+Room state data (such as joins, leaves, topic) is always preserved.
+
+To delete local message events as well, set ``delete_local_events`` in the body:
 
 .. code:: json
 

docs/workers.rst

@@ -230,7 +230,7 @@ file. For example::
 ``synapse.app.event_creator``
 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 
-Handles non-state event creation. It can handle REST endpoints matching:
+Handles non-state event creation. It can handle REST endpoints matching::
 
     ^/_matrix/client/(api/v1|r0|unstable)/rooms/.*/send
 

synapse/__init__.py

@@ -16,4 +16,4 @@
 """ This is a reference implementation of a Matrix home server.
 """
 
-__version__ = "0.26.0"
+__version__ = "0.27.2"

synapse/api/filtering.py

@@ -17,7 +17,7 @@ from synapse.storage.presence import UserPresenceState
 from synapse.types import UserID, RoomID
 from twisted.internet import defer
 
-import ujson as json
+import simplejson as json
 import jsonschema
 from jsonschema import FormatChecker
 

synapse/app/appservice.py

@@ -36,7 +36,7 @@ from synapse.util.logcontext import LoggingContext, preserve_fn
 from synapse.util.manhole import manhole
 from synapse.util.versionstring import get_version_string
 from twisted.internet import reactor
-from twisted.web.resource import Resource
+from twisted.web.resource import NoResource
 
 logger = logging.getLogger("synapse.app.appservice")
 
@@ -64,7 +64,7 @@ class AppserviceServer(HomeServer):
                 if name == "metrics":
                     resources[METRICS_PREFIX] = MetricsResource(self)
 
-        root_resource = create_resource_tree(resources, Resource())
+        root_resource = create_resource_tree(resources, NoResource())
 
         _base.listen_tcp(
             bind_addresses,

synapse/app/client_reader.py

@@ -44,7 +44,7 @@ from synapse.util.logcontext import LoggingContext
 from synapse.util.manhole import manhole
 from synapse.util.versionstring import get_version_string
 from twisted.internet import reactor
-from twisted.web.resource import Resource
+from twisted.web.resource import NoResource
 
 logger = logging.getLogger("synapse.app.client_reader")
 
@@ -88,7 +88,7 @@ class ClientReaderServer(HomeServer):
                         "/_matrix/client/api/v1": resource,
                     })
 
-        root_resource = create_resource_tree(resources, Resource())
+        root_resource = create_resource_tree(resources, NoResource())
 
         _base.listen_tcp(
             bind_addresses,

synapse/app/event_creator.py

@@ -52,7 +52,7 @@ from synapse.util.logcontext import LoggingContext
 from synapse.util.manhole import manhole
 from synapse.util.versionstring import get_version_string
 from twisted.internet import reactor
-from twisted.web.resource import Resource
+from twisted.web.resource import NoResource
 
 logger = logging.getLogger("synapse.app.event_creator")
 
@@ -104,7 +104,7 @@ class EventCreatorServer(HomeServer):
                         "/_matrix/client/api/v1": resource,
                     })
 
-        root_resource = create_resource_tree(resources, Resource())
+        root_resource = create_resource_tree(resources, NoResource())
 
         _base.listen_tcp(
             bind_addresses,

synapse/app/federation_reader.py

@@ -41,7 +41,7 @@ from synapse.util.logcontext import LoggingContext
 from synapse.util.manhole import manhole
 from synapse.util.versionstring import get_version_string
 from twisted.internet import reactor
-from twisted.web.resource import Resource
+from twisted.web.resource import NoResource
 
 logger = logging.getLogger("synapse.app.federation_reader")
 
@@ -77,7 +77,7 @@ class FederationReaderServer(HomeServer):
                         FEDERATION_PREFIX: TransportLayerServer(self),
                     })
 
-        root_resource = create_resource_tree(resources, Resource())
+        root_resource = create_resource_tree(resources, NoResource())
 
         _base.listen_tcp(
             bind_addresses,

synapse/app/federation_sender.py

@@ -42,7 +42,7 @@ from synapse.util.logcontext import LoggingContext, preserve_fn
 from synapse.util.manhole import manhole
 from synapse.util.versionstring import get_version_string
 from twisted.internet import defer, reactor
-from twisted.web.resource import Resource
+from twisted.web.resource import NoResource
 
 logger = logging.getLogger("synapse.app.federation_sender")
 
@@ -91,7 +91,7 @@ class FederationSenderServer(HomeServer):
                 if name == "metrics":
                     resources[METRICS_PREFIX] = MetricsResource(self)
 
-        root_resource = create_resource_tree(resources, Resource())
+        root_resource = create_resource_tree(resources, NoResource())
 
         _base.listen_tcp(
             bind_addresses,

synapse/app/frontend_proxy.py

@@ -44,7 +44,7 @@ from synapse.util.logcontext import LoggingContext
 from synapse.util.manhole import manhole
 from synapse.util.versionstring import get_version_string
 from twisted.internet import defer, reactor
-from twisted.web.resource import Resource
+from twisted.web.resource import NoResource
 
 logger = logging.getLogger("synapse.app.frontend_proxy")
 
@@ -142,7 +142,7 @@ class FrontendProxyServer(HomeServer):
                         "/_matrix/client/api/v1": resource,
                     })
 
-        root_resource = create_resource_tree(resources, Resource())
+        root_resource = create_resource_tree(resources, NoResource())
 
         _base.listen_tcp(
             bind_addresses,

synapse/app/homeserver.py

@@ -56,7 +56,7 @@ from synapse.util.rlimit import change_resource_limit
 from synapse.util.versionstring import get_version_string
 from twisted.application import service
 from twisted.internet import defer, reactor
-from twisted.web.resource import EncodingResourceWrapper, Resource
+from twisted.web.resource import EncodingResourceWrapper, NoResource
 from twisted.web.server import GzipEncoderFactory
 from twisted.web.static import File
 
@@ -126,7 +126,7 @@ class SynapseHomeServer(HomeServer):
         if WEB_CLIENT_PREFIX in resources:
             root_resource = RootRedirect(WEB_CLIENT_PREFIX)
         else:
-            root_resource = Resource()
+            root_resource = NoResource()
 
         root_resource = create_resource_tree(resources, root_resource)
 

synapse/app/media_repository.py

@@ -43,7 +43,7 @@ from synapse.util.logcontext import LoggingContext
 from synapse.util.manhole import manhole
 from synapse.util.versionstring import get_version_string
 from twisted.internet import reactor
-from twisted.web.resource import Resource
+from twisted.web.resource import NoResource
 
 logger = logging.getLogger("synapse.app.media_repository")
 
@@ -84,7 +84,7 @@ class MediaRepositoryServer(HomeServer):
                         ),
                     })
 
-        root_resource = create_resource_tree(resources, Resource())
+        root_resource = create_resource_tree(resources, NoResource())
 
         _base.listen_tcp(
             bind_addresses,

synapse/app/pusher.py

@@ -37,7 +37,7 @@ from synapse.util.logcontext import LoggingContext, preserve_fn
 from synapse.util.manhole import manhole
 from synapse.util.versionstring import get_version_string
 from twisted.internet import defer, reactor
-from twisted.web.resource import Resource
+from twisted.web.resource import NoResource
 
 logger = logging.getLogger("synapse.app.pusher")
 
@@ -94,7 +94,7 @@ class PusherServer(HomeServer):
                 if name == "metrics":
                     resources[METRICS_PREFIX] = MetricsResource(self)
 
-        root_resource = create_resource_tree(resources, Resource())
+        root_resource = create_resource_tree(resources, NoResource())
 
         _base.listen_tcp(
             bind_addresses,

synapse/app/synchrotron.py

@@ -56,7 +56,7 @@ from synapse.util.manhole import manhole
 from synapse.util.stringutils import random_string
 from synapse.util.versionstring import get_version_string
 from twisted.internet import defer, reactor
-from twisted.web.resource import Resource
+from twisted.web.resource import NoResource
 
 logger = logging.getLogger("synapse.app.synchrotron")
 
@@ -269,7 +269,7 @@ class SynchrotronServer(HomeServer):
                         "/_matrix/client/api/v1": resource,
                     })
 
-        root_resource = create_resource_tree(resources, Resource())
+        root_resource = create_resource_tree(resources, NoResource())
 
         _base.listen_tcp(
             bind_addresses,

synapse/app/user_dir.py

@@ -43,7 +43,7 @@ from synapse.util.logcontext import LoggingContext, preserve_fn
 from synapse.util.manhole import manhole
 from synapse.util.versionstring import get_version_string
 from twisted.internet import reactor
-from twisted.web.resource import Resource
+from twisted.web.resource import NoResource
 
 logger = logging.getLogger("synapse.app.user_dir")
 
@@ -116,7 +116,7 @@ class UserDirectoryServer(HomeServer):
                         "/_matrix/client/api/v1": resource,
                     })
 
-        root_resource = create_resource_tree(resources, Resource())
+        root_resource = create_resource_tree(resources, NoResource())
 
         _base.listen_tcp(
             bind_addresses,

synapse/config/registration.py

@@ -77,7 +77,9 @@ class RegistrationConfig(Config):
 
         # Set the number of bcrypt rounds used to generate password hash.
         # Larger numbers increase the work factor needed to generate the hash.
-        # The default number of rounds is 12.
+        # The default number is 12 (which equates to 2^12 rounds).
+        # N.B. that increasing this will exponentially increase the time required
+        # to register or login - e.g. 24 => 2^24 rounds which will take >20 mins.
         bcrypt_rounds: 12
 
         # Allows users to register as guests without a password/email/etc, and

synapse/handlers/e2e_keys.py

@@ -13,7 +13,7 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-import ujson as json
+import simplejson as json
 import logging
 
 from canonicaljson import encode_canonical_json

synapse/handlers/message.py

@@ -38,7 +38,7 @@ from canonicaljson import encode_canonical_json
 
 import logging
 import random
-import ujson
+import simplejson
 
 logger = logging.getLogger(__name__)
 
@@ -678,8 +678,8 @@ class EventCreationHandler(object):
 
         # Ensure that we can round trip before trying to persist in db
         try:
-            dump = ujson.dumps(unfreeze(event.content))
+            dump = simplejson.dumps(unfreeze(event.content))
-            ujson.loads(dump)
+            simplejson.loads(dump)
         except Exception:
             logger.exception("Failed to encode content: %r", event.content)
             raise

synapse/handlers/register.py

@@ -24,7 +24,7 @@ from synapse.api.errors import (
 from synapse.http.client import CaptchaServerHttpClient
 from synapse import types
 from synapse.types import UserID, create_requester, RoomID, RoomAlias
-from synapse.util.async import run_on_reactor
+from synapse.util.async import run_on_reactor, Linearizer
 from synapse.util.threepids import check_3pid_allowed
 from ._base import BaseHandler
 
@@ -46,6 +46,10 @@ class RegistrationHandler(BaseHandler):
 
         self.macaroon_gen = hs.get_macaroon_generator()
 
+        self._generate_user_id_linearizer = Linearizer(
+            name="_generate_user_id_linearizer",
+        )
+
     @defer.inlineCallbacks
     def check_username(self, localpart, guest_access_token=None,
                        assigned_user_id=None):
@@ -352,9 +356,11 @@ class RegistrationHandler(BaseHandler):
     @defer.inlineCallbacks
     def _generate_user_id(self, reseed=False):
         if reseed or self._next_generated_user_id is None:
-            self._next_generated_user_id = (
+            with (yield self._generate_user_id_linearizer.queue(())):
-                yield self.store.find_next_generated_user_id_localpart()
+                if reseed or self._next_generated_user_id is None:
-            )
+                    self._next_generated_user_id = (
+                        yield self.store.find_next_generated_user_id_localpart()
+                    )
 
         id = self._next_generated_user_id
         self._next_generated_user_id += 1

synapse/http/server.py

@@ -37,7 +37,7 @@ from twisted.web.util import redirectTo
 import collections
 import logging
 import urllib
-import ujson
+import simplejson
 
 logger = logging.getLogger(__name__)
 
@@ -461,8 +461,7 @@ def respond_with_json(request, code, json_object, send_cors=False,
         if canonical_json or synapse.events.USE_FROZEN_DICTS:
             json_bytes = encode_canonical_json(json_object)
         else:
-            # ujson doesn't like frozen_dicts.
+            json_bytes = simplejson.dumps(json_object)
-            json_bytes = ujson.dumps(json_object, ensure_ascii=False)
 
     return respond_with_json_bytes(
         request, code, json_bytes,
@@ -489,6 +488,7 @@ def respond_with_json_bytes(request, code, json_bytes, send_cors=False,
     request.setHeader(b"Content-Type", b"application/json")
     request.setHeader(b"Server", version_string)
     request.setHeader(b"Content-Length", b"%d" % (len(json_bytes),))
+    request.setHeader(b"Cache-Control", b"no-cache, no-store, must-revalidate")
 
     if send_cors:
         set_cors_headers(request)

synapse/python_dependencies.py

@@ -34,7 +34,6 @@ REQUIREMENTS = {
     "bcrypt": ["bcrypt>=3.1.0"],
     "pillow": ["PIL"],
     "pydenticon": ["pydenticon"],
-    "ujson": ["ujson"],
     "blist": ["blist"],
     "pysaml2>=3.0.0": ["saml2>=3.0.0"],
     "pymacaroons-pynacl": ["pymacaroons"],

synapse/replication/tcp/commands.py

@@ -19,7 +19,7 @@ allowed to be sent by which side.
 """
 
 import logging
-import ujson as json
+import simplejson
 
 
 logger = logging.getLogger(__name__)
@@ -100,14 +100,14 @@ class RdataCommand(Command):
         return cls(
             stream_name,
             None if token == "batch" else int(token),
-            json.loads(row_json)
+            simplejson.loads(row_json)
         )
 
     def to_line(self):
         return " ".join((
             self.stream_name,
             str(self.token) if self.token is not None else "batch",
-            json.dumps(self.row),
+            simplejson.dumps(self.row, namedtuple_as_object=False),
         ))
 
 
@@ -298,10 +298,12 @@ class InvalidateCacheCommand(Command):
     def from_line(cls, line):
         cache_func, keys_json = line.split(" ", 1)
 
-        return cls(cache_func, json.loads(keys_json))
+        return cls(cache_func, simplejson.loads(keys_json))
 
     def to_line(self):
-        return " ".join((self.cache_func, json.dumps(self.keys)))
+        return " ".join((
+            self.cache_func, simplejson.dumps(self.keys, namedtuple_as_object=False)
+        ))
 
 
 class UserIpCommand(Command):
@@ -325,14 +327,14 @@ class UserIpCommand(Command):
     def from_line(cls, line):
         user_id, jsn = line.split(" ", 1)
 
-        access_token, ip, user_agent, device_id, last_seen = json.loads(jsn)
+        access_token, ip, user_agent, device_id, last_seen = simplejson.loads(jsn)
 
         return cls(
             user_id, access_token, ip, user_agent, device_id, last_seen
         )
 
     def to_line(self):
-        return self.user_id + " " + json.dumps((
+        return self.user_id + " " + simplejson.dumps((
             self.access_token, self.ip, self.user_agent, self.device_id,
             self.last_seen,
         ))

synapse/rest/client/v1/room.py

@@ -30,7 +30,7 @@ from synapse.http.servlet import (
 
 import logging
 import urllib
-import ujson as json
+import simplejson as json
 
 logger = logging.getLogger(__name__)
 

synapse/rest/client/v2_alpha/sync.py

@@ -33,7 +33,7 @@ from ._base import set_timeline_upper_limit
 import itertools
 import logging
 
-import ujson as json
+import simplejson as json
 
 logger = logging.getLogger(__name__)
 

synapse/rest/media/v1/preview_url_resource.py

@@ -23,7 +23,7 @@ import re
 import shutil
 import sys
 import traceback
-import ujson as json
+import simplejson as json
 import urlparse
 
 from twisted.web.server import NOT_DONE_YET

synapse/state.py

@@ -132,7 +132,7 @@ class StateHandler(object):
 
         state_map = yield self.store.get_events(state.values(), get_prev_content=False)
         state = {
-            key: state_map[e_id] for key, e_id in state.items() if e_id in state_map
+            key: state_map[e_id] for key, e_id in state.iteritems() if e_id in state_map
         }
 
         defer.returnValue(state)
@@ -378,7 +378,7 @@ class StateHandler(object):
             new_state = resolve_events_with_state_map(state_set_ids, state_map)
 
         new_state = {
-            key: state_map[ev_id] for key, ev_id in new_state.items()
+            key: state_map[ev_id] for key, ev_id in new_state.iteritems()
         }
 
         return new_state
@@ -458,15 +458,15 @@ class StateResolutionHandler(object):
             # build a map from state key to the event_ids which set that state.
             # dict[(str, str), set[str])
             state = {}
-            for st in state_groups_ids.values():
+            for st in state_groups_ids.itervalues():
-                for key, e_id in st.items():
+                for key, e_id in st.iteritems():
                     state.setdefault(key, set()).add(e_id)
 
             # build a map from state key to the event_ids which set that state,
             # including only those where there are state keys in conflict.
             conflicted_state = {
                 k: list(v)
-                for k, v in state.items()
+                for k, v in state.iteritems()
                 if len(v) > 1
             }
 
@@ -480,7 +480,7 @@ class StateResolutionHandler(object):
                     )
             else:
                 new_state = {
-                    key: e_ids.pop() for key, e_ids in state.items()
+                    key: e_ids.pop() for key, e_ids in state.iteritems()
                 }
 
             # if the new state matches any of the input state groups, we can
@@ -488,8 +488,8 @@ class StateResolutionHandler(object):
             # which will be used as a cache key for future resolutions, but
             # not get persisted.
             state_group = None
-            new_state_event_ids = frozenset(new_state.values())
+            new_state_event_ids = frozenset(new_state.itervalues())
-            for sg, events in state_groups_ids.items():
+            for sg, events in state_groups_ids.iteritems():
                 if new_state_event_ids == frozenset(e_id for e_id in events):
                     state_group = sg
                     break
@@ -702,7 +702,7 @@ def _resolve_with_state(unconflicted_state_ids, conflicted_state_ds, auth_event_
 
     auth_events = {
         key: state_map[ev_id]
-        for key, ev_id in auth_event_ids.items()
+        for key, ev_id in auth_event_ids.iteritems()
         if ev_id in state_map
     }
 
@@ -740,7 +740,7 @@ def _resolve_state_events(conflicted_state, auth_events):
 
     auth_events.update(resolved_state)
 
-    for key, events in conflicted_state.items():
+    for key, events in conflicted_state.iteritems():
         if key[0] == EventTypes.JoinRules:
             logger.debug("Resolving conflicted join rules %r", events)
             resolved_state[key] = _resolve_auth_events(
@@ -750,7 +750,7 @@ def _resolve_state_events(conflicted_state, auth_events):
 
     auth_events.update(resolved_state)
 
-    for key, events in conflicted_state.items():
+    for key, events in conflicted_state.iteritems():
         if key[0] == EventTypes.Member:
             logger.debug("Resolving conflicted member lists %r", events)
             resolved_state[key] = _resolve_auth_events(
@@ -760,7 +760,7 @@ def _resolve_state_events(conflicted_state, auth_events):
 
     auth_events.update(resolved_state)
 
-    for key, events in conflicted_state.items():
+    for key, events in conflicted_state.iteritems():
         if key not in resolved_state:
             logger.debug("Resolving conflicted state %r:%r", key, events)
             resolved_state[key] = _resolve_normal_events(

synapse/storage/account_data.py

@@ -23,7 +23,7 @@ from synapse.util.caches.stream_change_cache import StreamChangeCache
 from synapse.util.caches.descriptors import cached, cachedList, cachedInlineCallbacks
 
 import abc
-import ujson as json
+import simplejson as json
 import logging
 
 logger = logging.getLogger(__name__)

synapse/storage/background_updates.py

@@ -19,7 +19,7 @@ from . import engines
 
 from twisted.internet import defer
 
-import ujson as json
+import simplejson as json
 import logging
 
 logger = logging.getLogger(__name__)

synapse/storage/deviceinbox.py

@@ -14,7 +14,7 @@
 # limitations under the License.
 
 import logging
-import ujson
+import simplejson
 
 from twisted.internet import defer
 
@@ -85,7 +85,7 @@ class DeviceInboxStore(BackgroundUpdateStore):
             )
             rows = []
             for destination, edu in remote_messages_by_destination.items():
-                edu_json = ujson.dumps(edu)
+                edu_json = simplejson.dumps(edu)
                 rows.append((destination, stream_id, now_ms, edu_json))
             txn.executemany(sql, rows)
 
@@ -177,7 +177,7 @@ class DeviceInboxStore(BackgroundUpdateStore):
                     " WHERE user_id = ?"
                 )
                 txn.execute(sql, (user_id,))
-                message_json = ujson.dumps(messages_by_device["*"])
+                message_json = simplejson.dumps(messages_by_device["*"])
                 for row in txn:
                     # Add the message for all devices for this user on this
                     # server.
@@ -199,7 +199,7 @@ class DeviceInboxStore(BackgroundUpdateStore):
                     # Only insert into the local inbox if the device exists on
                     # this server
                     device = row[0]
-                    message_json = ujson.dumps(messages_by_device[device])
+                    message_json = simplejson.dumps(messages_by_device[device])
                     messages_json_for_user[device] = message_json
 
             if messages_json_for_user:
@@ -253,7 +253,7 @@ class DeviceInboxStore(BackgroundUpdateStore):
             messages = []
             for row in txn:
                 stream_pos = row[0]
-                messages.append(ujson.loads(row[1]))
+                messages.append(simplejson.loads(row[1]))
             if len(messages) < limit:
                 stream_pos = current_stream_id
             return (messages, stream_pos)
@@ -389,7 +389,7 @@ class DeviceInboxStore(BackgroundUpdateStore):
             messages = []
             for row in txn:
                 stream_pos = row[0]
-                messages.append(ujson.loads(row[1]))
+                messages.append(simplejson.loads(row[1]))
             if len(messages) < limit:
                 stream_pos = current_stream_id
             return (messages, stream_pos)

synapse/storage/devices.py

@@ -13,7 +13,7 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 import logging
-import ujson as json
+import simplejson as json
 
 from twisted.internet import defer
 

synapse/storage/end_to_end_keys.py

@@ -17,7 +17,7 @@ from twisted.internet import defer
 from synapse.util.caches.descriptors import cached
 
 from canonicaljson import encode_canonical_json
-import ujson as json
+import simplejson as json
 
 from ._base import SQLBaseStore
 

synapse/storage/event_push_actions.py

@@ -22,7 +22,7 @@ from synapse.types import RoomStreamToken
 from .stream import lower_bound
 
 import logging
-import ujson as json
+import simplejson as json
 
 logger = logging.getLogger(__name__)
 

synapse/storage/events.py

@@ -38,7 +38,7 @@ from functools import wraps
 import synapse.metrics
 
 import logging
-import ujson as json
+import simplejson as json
 
 # these are only included to make the type annotations work
 from synapse.events import EventBase    # noqa: F401
@@ -53,10 +53,25 @@ event_counter = metrics.register_counter(
     "persisted_events_sep", labels=["type", "origin_type", "origin_entity"]
 )
 
+# The number of times we are recalculating the current state
+state_delta_counter = metrics.register_counter(
+    "state_delta",
+)
+# The number of times we are recalculating state when there is only a
+# single forward extremity
+state_delta_single_event_counter = metrics.register_counter(
+    "state_delta_single_event",
+)
+# The number of times we are reculating state when we could have resonably
+# calculated the delta when we calculated the state for an event we were
+# persisting.
+state_delta_reuse_delta_counter = metrics.register_counter(
+    "state_delta_reuse_delta",
+)
+
 
 def encode_json(json_object):
     if USE_FROZEN_DICTS:
-        # ujson doesn't like frozen_dicts
         return encode_canonical_json(json_object)
     else:
         return json.dumps(json_object, ensure_ascii=False)
@@ -369,7 +384,8 @@ class EventsStore(EventsWorkerStore):
                                 room_id, ev_ctx_rm, latest_event_ids
                             )
 
-                            if new_latest_event_ids == set(latest_event_ids):
+                            latest_event_ids = set(latest_event_ids)
+                            if new_latest_event_ids == latest_event_ids:
                                 # No change in extremities, so no change in state
                                 continue
 
@@ -390,6 +406,26 @@ class EventsStore(EventsWorkerStore):
                                 if all_single_prev_not_state:
                                     continue
 
+                            state_delta_counter.inc()
+                            if len(new_latest_event_ids) == 1:
+                                state_delta_single_event_counter.inc()
+
+                                # This is a fairly handwavey check to see if we could
+                                # have guessed what the delta would have been when
+                                # processing one of these events.
+                                # What we're interested in is if the latest extremities
+                                # were the same when we created the event as they are
+                                # now. When this server creates a new event (as opposed
+                                # to receiving it over federation) it will use the
+                                # forward extremities as the prev_events, so we can
+                                # guess this by looking at the prev_events and checking
+                                # if they match the current forward extremities.
+                                for ev, _ in ev_ctx_rm:
+                                    prev_event_ids = set(e for e, _ in ev.prev_events)
+                                    if latest_event_ids == prev_event_ids:
+                                        state_delta_reuse_delta_counter.inc()
+                                        break
+
                             logger.info(
                                 "Calculating state delta for room %s", room_id,
                             )

synapse/storage/events_worker.py

@@ -28,7 +28,7 @@ from synapse.api.errors import SynapseError
 from collections import namedtuple
 
 import logging
-import ujson as json
+import simplejson as json
 
 # these are only included to make the type annotations work
 from synapse.events import EventBase    # noqa: F401

synapse/storage/group_server.py

@@ -19,7 +19,7 @@ from synapse.api.errors import SynapseError
 
 from ._base import SQLBaseStore
 
-import ujson as json
+import simplejson as json
 
 
 # The category ID for the "default" category. We don't store as null in the

synapse/storage/receipts.py

@@ -23,7 +23,7 @@ from twisted.internet import defer
 
 import abc
 import logging
-import ujson as json
+import simplejson as json
 
 
 logger = logging.getLogger(__name__)

synapse/storage/registration.py

@@ -460,14 +460,12 @@ class RegistrationStore(RegistrationWorkerStore,
         """
         def _find_next_generated_user_id(txn):
             txn.execute("SELECT name FROM users")
-            rows = self.cursor_to_dict(txn)
 
             regex = re.compile("^@(\d+):")
 
             found = set()
 
-            for r in rows:
+            for user_id, in txn:
-                user_id = r["name"]
                 match = regex.search(user_id)
                 if match:
                     found.add(int(match.group(1)))

synapse/storage/room.py

@@ -22,7 +22,7 @@ from synapse.util.caches.descriptors import cached, cachedInlineCallbacks
 
 import collections
 import logging
-import ujson as json
+import simplejson as json
 import re
 
 logger = logging.getLogger(__name__)

synapse/storage/roommember.py

@@ -28,7 +28,7 @@ from synapse.api.constants import Membership, EventTypes
 from synapse.types import get_domain_from_id
 
 import logging
-import ujson as json
+import simplejson as json
 
 logger = logging.getLogger(__name__)
 

synapse/storage/schema/delta/25/fts.py

@@ -17,7 +17,7 @@ import logging
 from synapse.storage.prepare_database import get_statements
 from synapse.storage.engines import PostgresEngine, Sqlite3Engine
 
-import ujson
+import simplejson
 
 logger = logging.getLogger(__name__)
 
@@ -66,7 +66,7 @@ def run_create(cur, database_engine, *args, **kwargs):
             "max_stream_id_exclusive": max_stream_id + 1,
             "rows_inserted": 0,
         }
-        progress_json = ujson.dumps(progress)
+        progress_json = simplejson.dumps(progress)
 
         sql = (
             "INSERT into background_updates (update_name, progress_json)"

synapse/storage/schema/delta/27/ts.py

@@ -16,7 +16,7 @@ import logging
 
 from synapse.storage.prepare_database import get_statements
 
-import ujson
+import simplejson
 
 logger = logging.getLogger(__name__)
 
@@ -45,7 +45,7 @@ def run_create(cur, database_engine, *args, **kwargs):
             "max_stream_id_exclusive": max_stream_id + 1,
             "rows_inserted": 0,
         }
-        progress_json = ujson.dumps(progress)
+        progress_json = simplejson.dumps(progress)
 
         sql = (
             "INSERT into background_updates (update_name, progress_json)"

synapse/storage/schema/delta/31/search_update.py

@@ -16,7 +16,7 @@ from synapse.storage.engines import PostgresEngine
 from synapse.storage.prepare_database import get_statements
 
 import logging
-import ujson
+import simplejson
 
 logger = logging.getLogger(__name__)
 
@@ -49,7 +49,7 @@ def run_create(cur, database_engine, *args, **kwargs):
             "rows_inserted": 0,
             "have_added_indexes": False,
         }
-        progress_json = ujson.dumps(progress)
+        progress_json = simplejson.dumps(progress)
 
         sql = (
             "INSERT into background_updates (update_name, progress_json)"

synapse/storage/schema/delta/33/event_fields.py

@@ -15,7 +15,7 @@
 from synapse.storage.prepare_database import get_statements
 
 import logging
-import ujson
+import simplejson
 
 logger = logging.getLogger(__name__)
 
@@ -44,7 +44,7 @@ def run_create(cur, database_engine, *args, **kwargs):
             "max_stream_id_exclusive": max_stream_id + 1,
             "rows_inserted": 0,
         }
-        progress_json = ujson.dumps(progress)
+        progress_json = simplejson.dumps(progress)
 
         sql = (
             "INSERT into background_updates (update_name, progress_json)"

synapse/storage/search.py

@@ -16,7 +16,7 @@
 from collections import namedtuple
 import logging
 import re
-import ujson as json
+import simplejson as json
 
 from twisted.internet import defer
 

synapse/storage/tags.py

@@ -19,7 +19,7 @@ from synapse.storage.account_data import AccountDataWorkerStore
 from synapse.util.caches.descriptors import cached
 from twisted.internet import defer
 
-import ujson as json
+import simplejson as json
 import logging
 
 logger = logging.getLogger(__name__)

synapse/storage/transactions.py

@@ -23,7 +23,7 @@ from canonicaljson import encode_canonical_json
 from collections import namedtuple
 
 import logging
-import ujson as json
+import simplejson as json
 
 logger = logging.getLogger(__name__)
 

synapse/storage/user_directory.py

@@ -667,7 +667,7 @@ class UserDirectoryStore(SQLBaseStore):
             # The array of numbers are the weights for the various part of the
             # search: (domain, _, display name, localpart)
             sql = """
-                SELECT d.user_id, display_name, avatar_url
+                SELECT d.user_id AS user_id, display_name, avatar_url
                 FROM user_directory_search
                 INNER JOIN user_directory AS d USING (user_id)
                 %s
@@ -702,7 +702,7 @@ class UserDirectoryStore(SQLBaseStore):
             search_query = _parse_query_sqlite(search_term)
 
             sql = """
-                SELECT d.user_id, display_name, avatar_url
+                SELECT d.user_id AS user_id, display_name, avatar_url
                 FROM user_directory_search
                 INNER JOIN user_directory AS d USING (user_id)
                 %s

synapse/util/caches/dictionary_cache.py

@@ -132,9 +132,13 @@ class DictionaryCache(object):
                 self._update_or_insert(key, value, known_absent)
 
     def _update_or_insert(self, key, value, known_absent):
-        entry = self.cache.setdefault(key, DictionaryEntry(False, set(), {}))
+        # We pop and reinsert as we need to tell the cache the size may have
+        # changed
+
+        entry = self.cache.pop(key, DictionaryEntry(False, set(), {}))
         entry.value.update(value)
         entry.known_absent.update(known_absent)
+        self.cache[key] = entry
 
     def _insert(self, key, value, known_absent):
         self.cache[key] = DictionaryEntry(True, known_absent, value)

synapse/util/caches/lrucache.py

@@ -154,14 +154,21 @@ class LruCache(object):
         def cache_set(key, value, callbacks=[]):
             node = cache.get(key, None)
             if node is not None:
-                if value != node.value:
+                # We sometimes store large objects, e.g. dicts, which cause
+                # the inequality check to take a long time. So let's only do
+                # the check if we have some callbacks to call.
+                if node.callbacks and value != node.value:
                     for cb in node.callbacks:
                         cb()
                     node.callbacks.clear()
 
-                    if size_callback:
+                # We don't bother to protect this by value != node.value as
-                        cached_cache_len[0] -= size_callback(node.value)
+                # generally size_callback will be cheap compared with equality
-                        cached_cache_len[0] += size_callback(value)
+                # checks. (For example, taking the size of two dicts is quicker
+                # than comparing them for equality.)
+                if size_callback:
+                    cached_cache_len[0] -= size_callback(node.value)
+                    cached_cache_len[0] += size_callback(value)
 
                 node.callbacks.update(callbacks)
 

synapse/util/httpresourcetree.py

@@ -12,7 +12,7 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-from twisted.web.resource import Resource
+from twisted.web.resource import NoResource
 
 import logging
 
@@ -45,7 +45,7 @@ def create_resource_tree(desired_tree, root_resource):
         for path_seg in full_path.split('/')[1:-1]:
             if path_seg not in last_resource.listNames():
                 # resource doesn't exist, so make a "dummy resource"
-                child_resource = Resource()
+                child_resource = NoResource()
                 last_resource.putChild(path_seg, child_resource)
                 res_id = _resource_id(last_resource, path_seg)
                 resource_mappings[res_id] = child_resource