samredai commented on pull request #3952: URL: https://github.com/apache/iceberg/pull/3952#issuecomment-1019331217
There was some discussion in previous python syncs about decreasing the size of this part of the codebase by not having a literal class defined separately for each type. I don't think we reached a consensus so maybe we can do that here. I believe the only place literals are used is in the expressions logic which handles filtering of the metadata (partitions, manifest files, etc). Types are used in a lot of places since `StructType` is used to define schema in many places (table schema as well as the schema of metadata files). I like the intuitiveness of only having to look in one place to understand how a type is defined as well as how an instance of that type (a literal) behaves and usage feels very close to how the python standard library works. If the types and literals actually aren't as tightly coupled as the class names in the legacy and java implementation might indicate, then I can see how keeping them as separate classes could be better and they can be tied together by just an attribute on the literal classes. Curious to hear everyone else's thoughts. -- 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] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
