Hi All, In researching and designing a way to implement the auto group by feature of openCypher, I believe that it is necessary to split it into two distinct cases -
1) Explicit group by keys. Fx: RETURN key, aggregate(x); 2) Implicit group by keys: Fx: RETURN aggregate(x)/key; For now, I believe that we should only implement #1. Per the documentation, tests, and other sources (see below), #2 is too ambiguous to implement and will only cause confusion. Additionally, not implementing it, would not affect feature usability - you can always explicitly state the key. https://www.opencypher.org/articles/2017/07/27/ocig1-aggregations-article/ So, for now, I am proposing to only implement #1 (explicit group by keys). Implicit group by keys should be tabled until a later date. John
