Marcel Weisgut

Thank you for reporting the issue. The benchmark target is not up-to-date anymore and we should clean it up respectively. Related to #187

Let's discuss this again soon, @Bouncner @tjjordan @dey4ss @benrobby @mweisgut.

possibly related to #2185?

> I have a working branch that gets the data to

Please introduce UCC (i.e., `Unique column combination (UCC)` in this PR's description for people unfamiliar with the acronym.

> This is also the case for TPC-DS, were the overall gain is negligible. But we can see a throughput degradation for `hyriseBenchmarkTPCDS - multi-threaded, shuffled, 15 clients, 15 cores`...

> Also, now, that the plug-in is merged: Do we want to add CLI option to enable UCC discovery in the benchmarks (and do sth. like in `Reproducibility` here [#2486...

>- build/include_what_you_use might be correct. When I spent tens of hours reducing the compile time, I also looked at where that time was going. The lion's share goes to template...

Good point. Since `cpplint` also reports false positive `build/include_what_you_use` errors in some situations, having it as a hard CI barrier/constraint for a PR wouldn't be a good solution.

Bifortnightly discussion result: we should investigate which error types of cpplint/cppcheck/other tools are justified and create hard CI constraints for those.