-
Notifications
You must be signed in to change notification settings - Fork 195
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
fix(DataStore): created and deleted model on one device appears as created on the other #3554
fix(DataStore): created and deleted model on one device appears as created on the other #3554
Conversation
…eated on the other (aws-amplify#3553)
} | ||
|
||
guard !localMetadatas.isEmpty else { | ||
return remoteModels.compactMap { getDisposition($0, localMetadata: nil) } |
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 line was causing the bug:
here we were geting an empty localMetadatas and remoteModels with two elements for the same model (the same ID):
eg. [
ModelA - id: aaa, version: 1, deleted: false,
ModelA - id: aaa, version: 2, deleted: true
]
and then "deleted: true disposition" was dropped so only "create disposition" was returned
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 certainly appears to be a bug. Previously, the reconciliation for remoteModels was executed individually. We've transitioned to batching, but overlooked this particular scenario. I agree with your proposed approach; we should groupBy and optimize the remoteModels accordingly.
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.
Let me try to summarize this, the reconciliation path for DataStore takes the sync query remote model results and decides to apply the remote model to the local database or drop it. Previously the logic processed one model at a time. We changed this to a batch operation by putting the reconciliation logic under a transaction block, operating on the entire list of remote models. The necessary logic updates to perform the SQL operations more efficiently included constructing a predicate of all the model identifiers as a filter criteria for querying local data that is used for the reconciliation logic. There exists a bug with this optimization when having multiple entries of remote model belong to the same identifier and the changes to the remote model ends up as a deleted model on the latest version.
Bug example
remoteModels = [id=111 version=1 deleted=false, id=111 version=2 deleted=true]
When the disposition of the model is calculated for the create model, it is to create the model with the local database, because there is no local model, nor is there a local model with version less than the remote model.
When the disposition of the model is calculated for the delete model, it is to drop it because no local model yet to exist, so it thinks there is nothing to delete.
The end result is the model is created but not deleted. The fix we are applying in the PR is to optimize the reconciliation by only taking the latest versions of the remote model. In the example, this becomes [id=111 version=2, deleted=true]
. The latest version represents the final state of the model that we want applied to the local database. In this case, the remote model will be dropped. There is no create disposition so it will not be created either.
...es/AWSDataStorePlugin/Sync/SubscriptionSync/ReconcileAndLocalSave/RemoteSyncReconciler.swift
Outdated
Show resolved
Hide resolved
…eated on the other aws-amplify#3553 Co-authored-by: Di Wu <[email protected]>
Bug: #3553
General Checklist
By submitting this pull request, I confirm that my contribution is made under the terms of the Apache 2.0 license.