> > Quick question: do you actually have an issue with truncate on binary > columns ?
No issue - as a consumer of Iceberg metadata I'd just like to clarify if we should expect to see partition fields with truncated binary. I was initially coding against the spec and planned to reject a truncate transform over anything but int, long, decimal, or string, until the TruncateByteBuffer support was brought to my attention. To Amogh's point - I suppose it's possible other consumers have already made that decision and would be broken by changing this in the spec. But I agree it feels like overkill to put this in v3, it does seem like it was just a miss in the original spec. On Thu, Apr 4, 2024 at 6:19 AM Jean-Baptiste Onofré <j...@nanthrax.net> wrote: > Hi Brian, > > Welcome to this list :) > > Quick question: do you actually have an issue with truncate on binary > columns ? The Truncate transform (in Iceberg API) supports BINARY > using TruncateByteBuffer, so it would make sense to clearly state this > in the spec. > > Regards > JB > > On Thu, Apr 4, 2024 at 2:45 AM Brian Hulette <bhule...@apache.org> wrote: > > > > Hello, this is my first time writing on this list so I'll introduce > myself. I'm Brian Hulette, I've been involved with a couple of Apache > projects in the past (Arrow and Beam), and now I'm working on BigQuery's > support for Iceberg. > > > > My colleague raised an issue [1] a while ago about a discrepancy between > the specification and the implementation - truncate is not supposed to work > on binary column, but it looks like it does. It seems unlikely that we > will drop support for something that is working now, so I figured we should > document the current behavior instead. I drafted a PR for this [2], could > someone help review? > > > > Thanks! > > Brian > > > > [1] https://github.com/apache/iceberg/issues/5251 > > [2] https://github.com/apache/iceberg/pull/10079 >
