TheNeuralBit commented on pull request #11609: URL: https://github.com/apache/beam/pull/11609#issuecomment-624149550
Hm so there are several ways of manually building a Row instance that provide different levels of runtime type-checking. `Row#addValues` explicitly validates everything, and `Row#attachValues` explicitly does not, for performance reasons. In SQL we have an option to switch between the two: https://github.com/apache/beam/blob/34c58c42f14d2534a2f72f9194ddf9cc69138eea/sdks/java/extensions/sql/src/main/java/org/apache/beam/sdk/extensions/sql/impl/rel/BeamAggregationRel.java#L354-L357 So we can have runtime type-checking for debugging, but then turn it off for performance. I'm not sure how `withFieldValue` is intended to work. I'm not sure if the missing `toInputType(toBaseType(value))` for that code path is intentional or an oversight. Can you clarify @reuvenlax? ---------------------------------------------------------------- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: [email protected]
