mosche commented on PR #24380: URL: https://github.com/apache/beam/pull/24380#issuecomment-1339054515
@aromanenko-dev Please have a look at the `NamedAggregators` class. Spark requires `Accumulators` to be registered ahead of time. `NamedAggregators` is meant to be a more dynamic container that allows dynamic accumulators. However, the whole thing is totally useless. It's not possible to add any such accumulator into the internal state map of `NamedAggregators`, it's always empty! Here's two things you can check to understand this better: - Have a look at `NamedAggregators#mNamedAggregators` to check where entries are added (`put`). There's no such place where this is possible. - Check usage of the `NamedAggregators.State` interface, which describes a named aggregator. This is never used. I have no glue for what reasons that code was ever added, but it's broken and can't be used. -- 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. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
