Håvard M. Ottestad

This is partially done. I haven't looked through all the deprecations, only the ones marked with forRemoval.

This allows for other code to implement performance improvements. Specifically the MemoryStore. If you extend this class then your object would also be an instanceOf AbstractIRI, so it would not...

It isn't intended to share a Sail between multiple SailRepositories. I know others have solved similar issues at the endpoint level, ex. [ReadOnlySparqlApplication.java](https://github.com/eclipse/rdf4j/blob/main/spring-components/spring-boot-sparql-web/src/main/java/org/eclipse/rdf4j/http/server/readonly/ReadOnlySparqlApplication.java)

It's definitely a use case that is worth investigating. I would lean towards an authentication/role-based solution instead since that would solve what you are trying to do while at the...

We don't have any plans at the moment no. Would you be interested in trying to add this yourself?

Impressive performance improvement!

> NativeStore: > > Benchmark Mode Cnt Score Error Units > OverflowBenchmarkSynthetic.loadLotsOfDataEmptyStore avgt 10 2090.100 ± 1886.604 ms/op > OverflowBenchmarkSynthetic.loadLotsOfDataNonEmptyStore avgt 10 2473.546 ± 1197.555 ms/op > > Benchmark Mode...

I need to take a good look at the code and test it out for myself. I also need to take a look at some benchmarks to see how your...

I'm able to reproduce a scenario where a closed Changeset is still used. On my machine I can reproduce it by debugging `org.eclipse.rdf4j.sail.shacl.MultithreadedNativeStoreIT` in `testLotsOfValidationFailuresReadUncommitted()`. Here is where the closed...

I also found that the changesets that are created by shallowClone() are not always closed. I don't know if that matters or not. You can test it out yourself with...