Hey Hao, GenericIndexed V2 is used when a column is large enough that it must be spread over multiple files (each file has a 2GB limit, since that's the most that can be mapped into a single ByteBuffer). Check out LargeColumnSupportedComplexColumnSerializer for an entry point. It's used today for most of the builtin complex columns.
On Sun, Jul 14, 2019 at 12:02 AM Hao Luo <h...@twitter.com.invalid> wrote: > Hi, > I have a few questions regarding the V2. > > 1. When is it constructed from values? The fromArray() or fromIterable() > seems always create version 1 and createGenericIndexedVersionTwo only > loads > existing V2 GenericIndexed. > 2. What is the purpose of GenericIndexed V2? When is it used? > > Thanks. > Hao Luo >