Bouncner

Not that I am aware of. But definitely something we'll need. So can we agree on that minimalistic chunk-index interface?

``` TYPED_TEST(OperatorsIndexScanTest, EmptyIncludeList) { const auto right_values = std::vector{AllTypeVariant{4}}; auto scan = std::make_shared(this->_int_int, this->_index_type, std::vector{ColumnID{1}}, PredicateCondition::LessThan, right_values); scan->included_chunk_ids = {}; scan->execute(); } ``` I expected this test to run without...

The problem over which we stumbled was the following: not every chunk is indexed (which will always be the case as do not yet have indexes on mutable chunks). So...

Yea, but I don’t see a good way of implementing it before we have merged the index and table scan. And I don’t know when this is going to happen.

> * RadixClusterSort does not do RadixClusterSort for non-equi predicates. Confusing name. Jep, I can tackle that in a upcoming PR for the SMJ. > * _join_runs resolves the predicate...

> * RadixClusterSort does not do RadixClusterSort for non-equi predicates. Confusing name. I am slowing trying to finally get my sort-merge PRs in the master. What about "ClusterSortMaterializer"?

It will sort the clusters with one of the upcoming PRs.

What do we do about multiple predicates which can be answered with a single index scan using the composite group key index? The whole idea of this issue make sense...

If I remember correctly, this paper made us aware of boost sorting (at least for me): http://openproceedings.org/2019/conf/edbt/EDBT19_paper_71.pdf This paper was presented with a Hyrise flag in the auditorium, so it...

Did you measure rather manually or it is a series of commits which I could run on a more recent server? We have seen huge differences on hash-based operators between...