-
Notifications
You must be signed in to change notification settings - Fork 321
feat: Implement EventCacheStoreLock::lock()
with poison error, and ::lock_unchecked
#4285
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Open
Hywan
wants to merge
8
commits into
matrix-org:main
Choose a base branch
from
Hywan:feat-event-cache-lock-poisoned
base: main
Could not load branches
Branch not found: {{ refName }}
Loading
Could not load tags
Nothing to show
Loading
Are you sure you want to change the base?
Some commits from the old base branch may be removed from the timeline,
and old review comments may become outdated.
+428
−69
Open
Changes from 1 commit
Commits
Show all changes
8 commits
Select commit
Hold shift + click to select a range
b06dac3
feat: Event cache lock has a _generation_ value.
Hywan dc31f93
refactor: Rename `EventCacheStoreLock::lock` to `lock_unchecked`.
Hywan d4e69a3
feat(common): Implement `BackingStore` for all `Arc<T>` where `T: Bac…
Hywan eff9827
feat(base): Implement `LockableEventCacheStore::is_poisoned`.
Hywan fa18ed4
feat(base): Implement `EventCacheStoreLock::lock`.
Hywan 4ace281
test(base): Add `logged_in_base_client_with_store_config`.
Hywan 5073cd1
test(base): Add tests for `EventCacheStoreLock::lock_*`.
Hywan cd4c12e
test(base): Improve test coverage.
Hywan File filter
Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
There are no files selected for viewing
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
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.
main comment here
Now that I see the full picture, I doubt that this is the best way to go. Why? Because the callers of
lock()
are now the ones who have to decide what to do based on getting a poisoned lock; and my guess is that they'll all do the same thing, that is, resetting the in-memory content of the event cache store. Since we are in the event cache store, could we instead reset the inner content ourselves, and then return theEventCacheStoreLockGuard
? (This would be aligned with the end target of the crypto store cross-process lock, fwiw)And considering the little code that was in
lock_unchecked()
, I'm sure we can keep the previous code oflock_unchecked()
as it was, and not havelock_unchecked()
call intolock()
. That would also avoid theEventCacheStoreLockPoisonError
type, which is super weird.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.
The
EventCacheStoreLock
isn't responsible to invalidate/refresh the in-memory data we have. However,EventCache
and other users of this lock are responsible of that. Example: every timeEventCache
is callingstore.lcok()
, and is responsible of refreshing its own data. I don't see howEventCacheStoreLock
can track all data used everywhere related to the event cache store. It's basically impossible.Remember that the
EventCache
uses theEventCacheStoreLock
to read load data from the store into memory, or to save data from memory into the store. TheEventCacheStoreLock
has no idea where the data that it has loaded or saved are used, and by whom.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.
If it owns it, as I suggested, it doesn't need to track it.
A thought experiment: if we keep the current API (modulo making the return type of the lock look better, to not have to deal with a result of a result), then the next thing I would implement is a thin wrapper on top of the lock, to invalidate the data manually. So it seems like this is just kicking the can down the road, and pushing the problem elsewhere in our architecture.
Moreover, using a generic parameter for the data stored along the lock and that needs a reset would nicely solve the layer separation. A thin
trait InvalidatableData { fn invalidate(); }
(lol @ naming) would be sufficient to represent data that needs invalidation, and mocking it in testing situations should be straightforward (use a bool, toggle it wheninvalidate()
is called).I am really not sure the splitting of responsibility gives us anything valuable, so I still need to be convinced this is the best approach.
Would you have WIP code making use of this, so I can get a better opinion about how it's going to be used?