[
https://issues.apache.org/jira/browse/S2GRAPH-53?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15175523#comment-15175523
]
DOYUNG YOON commented on S2GRAPH-53:
------------------------------------
here is my wip PR.
https://github.com/SteamShon/incubator-s2graph/tree/S2GRAPH-53
added package name serde and seperate tall row schema and wide row schema into
seperate classes and make storage to decide which class to use for given schema
version.
any feedback would be appreciated.
> Refactor Storage to decide which serializer/deserializer for
> IndexEdge/SnapshotEdge/Vertex.
> -------------------------------------------------------------------------------------------
>
> Key: S2GRAPH-53
> URL: https://issues.apache.org/jira/browse/S2GRAPH-53
> Project: S2Graph
> Issue Type: Improvement
> Reporter: DOYUNG YOON
> Assignee: DOYUNG YOON
> Priority: Minor
> Labels: refactor, serde
> Original Estimate: 48h
> Remaining Estimate: 48h
>
> SerDe for IndexEdge/SnapshotEdge/Vertex works differently based on label's
> schemaVersion and Storage only use one SerDe pair for each ones now.
> Decision for which SerDe is used for each label should be controlled by
> Storage since some of storage implementation only use certain SerDe pair for
> every label schema version.
> The only reason we can't truncate schema v1, v2 is only for backward
> compatibility.
> I am suggesting moving logic to decide which SerDe when given label's
> schemaVersion from IndexEdge/SnapshotEdge SerDe to Storage so newly added
> storage implementation override this logic to avoid manually change test
> label`s schema version, since some of newly added storage implementation only
> support some of schema version.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
