PSeitz

When I look at assembly, I want to understand the commands the CPU is receiving to analyze performance. That is never in dev. I understand the rationale though that an...

Awesome that support for `llvm-mca` has been added Could uica also be supported? (via their cli tool https://github.com/andreas-abel/uiCA)

I'm on Linux, which is I guess a common platform to do that kind of performance analysis. Wouldn't rust's cross compilation help for other platforms? https://github.com/pacak/cargo-show-asm/issues/137 > 2. preparing an...

Currently only u64 fast fields are supported. If we can map the str ranges to term ids as a preprocessing, reusing the existing u64 range queries on fast fields should...

Not by default, but the information is there from which the Snippet is constructed https://github.com/quickwit-oss/tantivy/blob/88a3275dbb230e5c7a708f53c8a2027858652a43/src/snippet/mod.rs#L233

bool should be simple for ip addresses we need to change the fast field API to get a u64 representation for the u128 values (which may limit possible codecs), to...

There is a hidden contract currently on the tokenizer API which expects positions to be sorted incrementally. This happens in the serialization part in [recorder.rs](https://github.com/quickwit-oss/tantivy/blob/main/src/postings/recorder.rs#L291), when positions are delta encoded....

The change caused some weird side-effects (I think it was recursion) and the tests to fail. I didn't investigate yet

Can you provide a stack trace or ideally something to reproduce?

Can you share your schema, tokenizer and hardware info?