Daniel Lindner

(just want to test if the changes fix #2632) diff: https://github.com/hyrise/hyrise/compare/martin/test/ubuntu2310...dey4ss/odr-fix

Now that Hyrise supports FKs (not in the LQP yet, but as table constraints), this PR adds FKs defined by benchmark specifications in the table generators of each benchmark executable.

Do not schedule tasks for correlated subqueries

2

Comparing ST and MT ordered benchmarks hinted that executing correlated subqueries got drastically slower when using multiple cores / the `NodeQueueScheduler`, e.g., in TPC-DS Q 41 (benchmarks to follow). The...

This PR adds order dependencies (ODs) on an LQP level and forwards them for all LQP nodes. Do not be afraid when looking at the diff: In most cases, the...

During recent work, we considerably often found the following pattern in optimizer rule tests: ```c++ const auto input_lqp = ...; const auto result_lqp = apply_rule(rule, input_lqp); const auto expected_lqp =...

Using clang-tidy 16, the build fails with the following (not very speaking issue, macOS): ``` ➜ clang16-debug git:(dey4ss/clang-16) ✗ ninja hyrise_impl hyriseBenchmarkFileBased hyriseBenchmarkTPCH hyriseBenchmarkTPCDS hyriseBenchmarkJoinOrder hyriseConsole hyriseServer hyriseMvccDeletePlugin [3/13] Building...

While working on #2590, we noticed that the optimizer runtimes for TPC-C are quite high and often, predicate reordering takes long (especially for TPC-C and JOB, see below). After some...

Many optimizer rules assert that (i) the input of `_apply_to_plan_without_subqueries(...)` is an `LQPRootNode` and (ii) the rule's `cost_estimator` is set. This assertion should be made either in `AbstractRule::apply_to_plan(...)` (can check...

Reduce cardinality estimation times

3

#2626 showed that we spend long times in the optimizer, especially during cardinality estimation. We do not really see this issue in our benchmarks because we do not have changing...

Cover two additional straight-forward cases in the `ExpressionReductionRule`: 1. Rewrite a `LIKE` predicate without wildcard to equals predicate. Not seen in benchmarks, but super easy to do and should help...