[
https://issues.apache.org/jira/browse/FLINK-39520?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
ASF GitHub Bot updated FLINK-39520:
-----------------------------------
Labels: pull-request-available (was: )
> Preparatory refactoring for channel-state recovery: split the recovered-state
> handler, add InputGate#getChannel(InputChannelInfo)
> ---------------------------------------------------------------------------------------------------------------------------------
>
> Key: FLINK-39520
> URL: https://issues.apache.org/jira/browse/FLINK-39520
> Project: Flink
> Issue Type: Sub-task
> Reporter: Rui Fan
> Assignee: Rui Fan
> Priority: Major
> Labels: pull-request-available
>
> Pure refactoring — no behavior or semantic change; all existing tests pass
> unmodified except mechanical adaptations (a renamed trace label, constructor
> plumbing in tests).
> Scope:
> * Decouple LocalInputChannel recovery wiring from toBeConsumedBuffers.
> * Split the monolithic InputChannelRecoveredStateHandler into an abstract
> base AbstractInputChannelRecoveredStateHandler (channel-mapping state and
> logic, extracted verbatim) with two concrete handlers — NoSpillingHandler
> (the old non-filtering branch) and FilteringHandler (the old filtering
> branch) — selected by a static create(...) factory equivalent to the old
> internal branch. ResultSubpartitionRecoveredStateHandler is untouched.
> * Add InputGate#getChannel(InputChannelInfo) (UnionInputGate resolves by
> gate index, fixing global-vs-member index semantics) and switch
> AbstractStreamTaskNetworkInput to it.
> * Additive NetworkActionsLogger#tracePersist overload; widen
> RecoveredInputChannel#releaseAllResources to public.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
