Sebastian Hillig

Specifically: - update-heavy workloads where one row is updated by many threads while other threads always have to find a valid version

Everything around ColumnMetadata handling is basically inviting the use to lose track of pointers and henceforth, memory. The interface should be change to either return unique pointers to ColumnMetadata or...

Since all interesting current compilers support `#pragma once` is the more logical and easier to maintain choice.

Currently, task schedulers are _way_ too nice when unexpected situations arise (for example a task is ready that is not currently in the waitset). Instead of sending a log warning...

Currently, we allocate transactions contexts and modification records on transaction creation - yet in read-only transactions the only thing actually needed is the own transaction id. Ideally, we would only...

Store: Delta resize is not thread-safe

7

While only one thread may resize the delta, other threads working on delta at the time of resizing may end up working on deallocated memory, since the underlying vectors may...

Layouter: Iteration beyond size

3

https://github.com/hyrise/hyrise/blob/master/src/lib/layouter/base.cpp#L1037 stop reaches .size() in this fragment of code.

Scheduler/SchedulerTest.dont_block_test/0 fails occasionally

1

Stacktrace Value of: long_block_test(scheduler) Actual: false Expected: true @JWUST any hints on that one?