On Tue, May 23, 2017 at 10:09 AM, Shirley Wang <swang@pivotal.io> wrote:
It's possible to design for the range and list partitions and know we can achieve success because we understand how users would go through this workflow. Not sure about expressions.
Maybe to pile on this a bit.
When Shirley and I were discussing the workflows it was obvious when we were looking at 'normal' range or list partition use cases. Generally the only open question we had about the workflow was whether or not users would be building tables net new or whether they were more likely to have a table that was growing too large and therefore needed to create a new partitioned table.
We couldn't think of a reason why a user would want to take the average of two columns and partition by this derived value. It added to the question of why/how a user would consider this as an idea a priori or whether this would be an insight given analysis of existing data.
I assume this was supported for a specific use case. if you could share that it would be awesome. I guess the long and short of it is, we are having a difficult time imagining the workflow for this feature.
