Bernhard Scholz

I also have optimized Galan by using better schedules. With more time, better schedules could be found. On my computer better schedules halve the runtime.

I had a quick look. This rule ``` skipBlank(From,To) :- nextElem(From,Via), !hasValue(Via), skipBlank(Via,To). ``` needs to be rewritten to this one ``` skipBlank(From,To) :- skipBlank(Via,To), nextElem(From,Via), !hasValue(Via). ``` if you...

Well - as we all know, current Datalog engines are not declarative when it comes to performance. Without providing appropriate hints, performance suffers severely. Hence, performance comparisons between different engines...

I had only a quick look at which rules can be improved immediately; at the moment we are trying to find appropriate benchmarks. Your benchmark suite is excellent! We may...

I made a pull-request substantially improving the runtime. See PR #3.

Aggregates are still very brittle. We would need to refactor the code for aggregates. Would you have time to help? I would defer fixing the documentation (it is hit and...

It seems a problem with the reporting. The relation is too fast to record any usage. You may need to increase the runtime of that relation to see any usage...

I would suggest a different message, i.e., no usage data for a relation does not mean that the relation does not exist; it might be simply too fast so that...

We would need a RAM transformer for optimising record accesses.

This is an abandoned prototype and hasn’t been completed. Please check the SWIG interface. You need to install all dependencies to make it work.