+1 On Wed, Apr 13, 2016 at 5:16 AM, Wes McKinney <w...@cloudera.com> wrote:
> +1 from me. Columnar all the way. > > - Wes > > On Sun, Apr 10, 2016 at 2:58 AM, Zheng, Kai <kai.zh...@intel.com> wrote: > > It's a good idea. There is no physical type corresponding to int48, > using int48 will need extra decoding step to access the type and offset > values. As it would change to use int8 type for the union type, this > alternative proposal would also be invalidated and removed. > > > > Regards, > > Kai > > > > -----Original Message----- > > From: Micah Kornfield [mailto:emkornfi...@gmail.com] > > Sent: Sunday, April 10, 2016 6:23 AM > > To: dev@arrow.apache.org > > Subject: Removng alternate proposal from layout of types and offsets for > unions in layout.md > > > > The current layout.md lists an alternate proposal for the layout of of > these values: > > > > "Alternate proposal (TBD): the types and offset values may be packed > into an > > int48 with 2 bytes for the type and 4 bytes for the offset." > > > > Any objections to removing this proposal and moving forward with keeping > them as two separate arrays? > > > > Thanks, > > Micah >
