LouisClt commented on PR #13962: URL: https://github.com/apache/arrow/pull/13962#issuecomment-1256255201
C Glib tests do not pass because it test the old behaviour. Looking a bit more in detail at the difference between "include" and "includeTypes", it appears that "includeTypes" is based on indices of the tables types. The root of the tree (meaning the whole table) is type of index 0, and when there are complex types such as structs would give other "child" types. See https://orc.apache.org/docs/types.html for more information. "include" select the fields in the same way than the other imports : each table has a certain number of fields, (we do not take into account the "children" types), if we want to select the first field, we put "0" in the list. I think the 2 behaviours could be understandable but the one with "include" is more coherent with the other imports. Furthermore, I do not know if there is a way in Arrow to get the list of internal ORC types, which makes selecting fields with "includeTypes" much more unreliable. -- 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]
