PSeitz

Thanks @LebranceBW for the PR!

The "0" and " " at the beginning look suspicious. A stacktrace would be really helpful (Shouldn't we have them on by default?) It cannot be reproduced like this: ```rust...

> At the threshold we would have an average of 20 elements per chimera block. Maybe linear search should be considered too? > Yes, linear search would probably be better...

I'm not sure what you mean, but detecting duplicate data is not possible during the indexing phase

https://github.com/quickwit-oss/tantivy/pull/2407

Still flaky ``` --- STDOUT: tantivy core::executor::tests::test_cancel_cpu_intensive_tasks --- running 1 test --- STDERR: tantivy core::executor::tests::test_cancel_cpu_intensive_tasks --- thread 'search-test0' panicked at src/core/executor.rs:206:42: called `Result::unwrap()` on an `Err` value: RecvError note: run...

Thanks @LebranceBW https://github.com/quickwit-oss/tantivy/pull/2405

Thanks for the PR, filtering is a great addition. I think the query side should be handled via the query list, with an added tag. The commands are more like...

I think duplicating should be fine, but we could have it in code when loading the queries. This has the advantage that you can easily get an overview and compare...

> @PSeitz We'd also need to make sure the query language handles it though (filters should not impact scoring). It might be a pain. > > I'd go with @jpountz...