lidavidm commented on PR #31: URL: https://github.com/apache/iceberg-cpp/pull/31#issuecomment-2601202575
Just to make sure @gaborkaszab @wgtmac: are we ok with the Arrow-style type representation here? (Types are represented by a class hierarchy, erased behind smart pointers; nested types store the fields in the type object, not the field object) The alternatives are: - cuDF style: type objects are minimal, just a type ID, and there is no hierarchy, just the base `DataType` class which is not type-erased and therefore there is no need for a smart pointer. As a trade-off, nested fields have to be extracted from the field and the base DataType has to have fields for all possible parameterized types - arrow-java style: there is still a hierarchy of type objects, but child fields are stored on the Field, not the DataType. Avoids a conceptual dependency cycle between DataType and Field - variant style: like arrow-java, but instead of type erasure and a type hierarchy (or conversely, like cuDF but avoids redundant fields), we have a `std::variant` with all possibilities. Avoids boxing but the object is larger (in e.g. a vector) -- 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: issues-unsubscr...@iceberg.apache.org For queries about this service, please contact Infrastructure at: us...@infra.apache.org --------------------------------------------------------------------- To unsubscribe, e-mail: issues-unsubscr...@iceberg.apache.org For additional commands, e-mail: issues-h...@iceberg.apache.org
