Markus Dreseler

At least I can't see anything wrong with it right now.

@julianmenzler will also look into index questions, but from the perspective of transactional workloads. In the TPC-C, the same data is updated over and over again, so we end up...

Notes from our meeting: - Indexes are currently not used by default - New index-based join is currently only used when PQPs are built by hand; index-based scan is theoretically...

Our constraints (PRIMARY, UNIQUE) are currently only modeled as "soft constraints". This means that we simply promise to the optimizer that these constraints are true. If you were to insert...

Having a chunk id list in the LQP is hacky anyway, as it refers to physical properties at a time where only logical properties should be considered. Chunk pruning being...

@julianmenzler is working on an architecture proposal

Which part of this do you consider dangerous? The whole idea is "Never spill to disk, rather crash (sigsegv)"?

Highly unlikely.

Maybe I am just misreading this. Two follow-up questions: * Why is this part of the table instead of something that the client tracks? * Should we use system_clock as...

> you are right, the client could track the time between requests as well, but this might not be as exact as if we track it due to scheduling, (network)...