Any plan to support schema evolution in Scan? #4477
-
DescriptionHi,
We notice in HiveConnector.cpp#L472-L474, the @mbasmanova Would you like to take a look on this issue? Thanks! |
Beta Was this translation helpful? Give feedback.
Replies: 5 comments 1 reply
-
@rui-mo Rui, thank you for reporting this gap. We usually refer to this functionality as schema evolution. Table schema may evolve: new columns are added, existing column types are modified. TableScan needs to support this by being able to read the files written before the schema chance. Looks like we are missing support for column type change. It would be nice to add that. CC: @Yuhta |
Beta Was this translation helpful? Give feedback.
-
Yes I notice this while we are reviewing #4040. Both names and types (including nested field names) should be able to overwrite from the planner instead of just reading from file. This is definitely something we should add in the future, probably in 2023 H2. |
Beta Was this translation helpful? Give feedback.
-
@mbasmanova @Yuhta Thanks for your answering! Glad to hear Velox plans to add it. Closing this question. |
Beta Was this translation helpful? Give feedback.
-
Hello @mbasmanova and @Yuhta , |
Beta Was this translation helpful? Give feedback.
@rui-mo Rui, thank you for reporting this gap. We usually refer to this functionality as schema evolution. Table schema may evolve: new columns are added, existing column types are modified. TableScan needs to support this by being able to read the files written before the schema chance. Looks like we are missing support for column type change. It would be nice to add that.
CC: @Yuhta