-
-
Notifications
You must be signed in to change notification settings - Fork 2.1k
mypy plugin to check @cached
return types
#14911
Conversation
Can we return |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
You grabbed my interest. Seems reasonable but I have a couple thoughts!
(Thanks for your thoughts, Patrick. I'll need to give this and them more thought, when I get some spare time to stare at this again) |
Yep, I think we should return a Mapping to make it clear that the caller can't mutate dictionaries that are kept as cache values. |
@@ -710,7 +710,7 @@ async def on_send_join_request( | |||
state_event_ids: Collection[str] | |||
servers_in_room: Optional[Collection[str]] | |||
if caller_supports_partial_state: | |||
summary = await self.store.get_room_summary(room_id) | |||
summary = await self.store.get_room_summary(room_id) # type: ignore[synapse-@cached-mutable] |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Exciting to see!
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Terrifying IMO of how we use this.
Testing with:
On this branch:
On
So ~3s slower, say 10% overhead. I think I'd consider that good enough as-is. |
Yeah that sounds reasonable to me. |
I think this probably needs a review from not @DMRobertson and not myself since we collaborated on this. |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Otherwise looks good!
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
This is great! 🎉
scripts-dev/mypy_synapse_plugin.py
Outdated
""" | ||
Get the "final" return type of a callable which might return returned an Awaitable/Deferred. | ||
""" |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Aside: If we have function returning e.g. Deferred[Deferred[T]]
, it looks like this function doesn't loop and extract the inner T. But we shouldn't be doing that in the first place.
(Would such a type even be legitimate?)
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
(Would such a type even be legitimate?)
I think it wouldn't make sense? Please don't do that.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I'm not sure that's even really doable since a Deferred
returning a Deferred
will wait on the inner Deferred
? https://docs.twistedmatrix.com/en/stable/core/howto/defer.html#chaining-deferreds
Anyway, that interaction always confuses me and I'm willing to not support it.
# its not the end of the world. | ||
# it's not the end of the world. |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
<3
"""Asserts that the signature of a method returns a value which can be cached. | ||
|
||
Makes no changes to the provided method signature. |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Can we add a sentence explaining why there is a wrapper? It looks like the point is mainly to check that we have a proper CallableType
(?)
OTOH if it's just "to have smaller functions" then let's leave it as is.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
OTOH if it's just "to have smaller functions" then let's leave it as is.
I almost inlined it, but it seemed nice to keep them separate.
Co-authored-by: David Robertson <[email protected]>
Thanks so much for the help @erikjohnston and @DMRobertson! Happy to see this merged! 🎉 |
No significant changes since 1.94.0rc1. - Render plain, CSS, CSV, JSON and common image formats in the browser (inline) when requested through the /download endpoint. ([\matrix-org#15988](matrix-org#15988)) - Add experimental support for [MSC4028](matrix-org/matrix-spec-proposals#4028) to push all encrypted events to clients. ([\matrix-org#16361](matrix-org#16361)) - Minor performance improvement when sending presence to federated servers. ([\matrix-org#16385](matrix-org#16385)) - Minor performance improvement by caching server ACL checking. ([\matrix-org#16360](matrix-org#16360)) - Add developer documentation concerning gradual schema migrations with column alterations. ([\matrix-org#15691](matrix-org#15691)) - Improve documentation of the user directory search algorithm. ([\matrix-org#16320](matrix-org#16320)) - Fix rendering of user admin API documentation around deactivation. This was broken in Synapse 1.91.0. ([\matrix-org#16355](matrix-org#16355)) - Update documentation around message retention policies. ([\matrix-org#16382](matrix-org#16382)) - Add note to `federation_domain_whitelist` config option to clarify its usage. ([\matrix-org#16416](matrix-org#16416)) - Improve legacy release notes. ([\matrix-org#16418](matrix-org#16418)) - Remove Python version from `/_synapse/admin/v1/server_version`. ([\matrix-org#16380](matrix-org#16380)) - Avoid running CI steps when the files they check have not been changed. ([\matrix-org#14745](matrix-org#14745), [\matrix-org#16387](matrix-org#16387)) - Improve type hints. ([\matrix-org#14911](matrix-org#14911), [\matrix-org#16350](matrix-org#16350), [\matrix-org#16356](matrix-org#16356), [\matrix-org#16395](matrix-org#16395)) - Added support for pydantic v2 in addition to pydantic v1. Contributed by Maxwell G (@gotmax23). ([\matrix-org#16332](matrix-org#16332)) - Get CI to check PRs have been signed-off. ([\matrix-org#16348](matrix-org#16348)) - Add missing licence header. ([\matrix-org#16359](matrix-org#16359)) - Improve type hints, and bump types-psycopg2 from 2.9.21.11 to 2.9.21.14. ([\matrix-org#16381](matrix-org#16381)) - Improve comments in `StateGroupBackgroundUpdateStore`. ([\matrix-org#16383](matrix-org#16383)) - Update maturin configuration. ([\matrix-org#16394](matrix-org#16394)) - Downgrade replication stream time out error log lines to warning. ([\matrix-org#16401](matrix-org#16401)) * Bump actions/checkout from 3 to 4. ([\matrix-org#16250](matrix-org#16250)) * Bump cryptography from 41.0.3 to 41.0.4. ([\matrix-org#16362](matrix-org#16362)) * Bump dawidd6/action-download-artifact from 2.27.0 to 2.28.0. ([\matrix-org#16374](matrix-org#16374)) * Bump docker/setup-buildx-action from 2 to 3. ([\matrix-org#16375](matrix-org#16375)) * Bump gitpython from 3.1.35 to 3.1.37. ([\matrix-org#16376](matrix-org#16376)) * Bump msgpack from 1.0.5 to 1.0.6. ([\matrix-org#16377](matrix-org#16377)) * Bump msgpack from 1.0.6 to 1.0.7. ([\matrix-org#16412](matrix-org#16412)) * Bump phonenumbers from 8.13.19 to 8.13.22. ([\matrix-org#16413](matrix-org#16413)) * Bump psycopg2 from 2.9.7 to 2.9.8. ([\matrix-org#16409](matrix-org#16409)) * Bump pydantic from 2.3.0 to 2.4.2. ([\matrix-org#16410](matrix-org#16410)) * Bump regex from 1.9.5 to 1.9.6. ([\matrix-org#16408](matrix-org#16408)) * Bump sentry-sdk from 1.30.0 to 1.31.0. ([\matrix-org#16378](matrix-org#16378)) * Bump types-netaddr from 0.8.0.9 to 0.9.0.1. ([\matrix-org#16411](matrix-org#16411)) * Bump types-psycopg2 from 2.9.21.11 to 2.9.21.14. ([\matrix-org#16381](matrix-org#16381)) * Bump urllib3 from 1.26.15 to 1.26.17. ([\matrix-org#16422](matrix-org#16422)) # -----BEGIN PGP SIGNATURE----- # # iQIzBAABCAAdFiEE8SRSDO7gYkSP4chELS76LzL74EcFAmUlINwACgkQLS76LzL7 # 4EdvExAAgjk6+/Fu45MRG7u5kFmFzoZWLOPD10XROANaSeqW1l/pBhFh+XvwR4TZ # l/FdkSfS9YpHnw3aof13TclLu6IVWDM+vqYFuY2HSY/yzbcGvJFHqr26kOccpTTd # 2r9m/AkguyHEBECDW8qJLXb8M7dqNa2SydTBu1+IrKfj6nq+fRxVyQhyAJXrI1Ta # Dnz0XJ4TcwTrMPVk4MYrAcYjID6IV89dtp7ttH4DwXKDeSjMtxM/46EIg4u+VXDz # fzK25JHVFYJA5+/rOn/RslmxjJHQfEIEB6NYxQwLeMeZuGSZooTebKn1odwogvhI # Srtfsytum+twgSHD1s+7KldM+EjTiu7ouKi8VcfOlFuLnuBiROEc5WUljcL5K63F # kVx2bXGU/eNkPp6ntNhYfgswx+yk2rXFqkTjz+xZQIZcOBqehHBDy8VhtwlRkTUw # bzocdKkLMA4nfSlq5fFOAErMqJKsPS8aN9yYPShqEUiSUOKle8eHfA1cTXJuK0MS # K2/YcDDZmJBrwVADyNDk5GKaDx39rR752OSuJb57Sp/edwUg6+H1I6lIN6YTeoJw # FzJwGMzuMCktOQRW2enxQiA6RZjXFCwvD1LoWMjyO4YTXQwXxNCXsb0kLKUqfwsy # qMGphWEl3rdzVSuFapNAgOLF0RfFNYZdhQnk+3fNEwxumxoqgho= # =hx8G # -----END PGP SIGNATURE----- # gpg: Signature made Tue Oct 10 11:01:00 2023 BST # gpg: using RSA key F124520CEEE062448FE1C8442D2EFA2F32FBE047 # gpg: Can't check signature: No public key # Conflicts: # .github/workflows/docker.yml # .github/workflows/docs-pr-netlify.yaml # .github/workflows/docs-pr.yaml # .github/workflows/docs.yaml # .github/workflows/latest_deps.yml # .github/workflows/poetry_lockfile.yaml # .github/workflows/push_complement_image.yml # .github/workflows/release-artifacts.yml # .github/workflows/tests.yml # .github/workflows/twisted_trunk.yml # poetry.lock # rust/src/push/base_rules.rs
The things that are cached should be immutable to avoid callers modifying cached values (which has lead to bugs in the past). This updates our mypy plug-in to enforce this via the type system.
Complements #13755.
Errors which have been fixed
The main obstruction seems to be that we have lots of Dict[str, Any] or Dict[str, str] in caches.
Sample:
Some outstanding TODOs: