Michael Färber

Hi @jakobhellermann, thanks for your improvements. Still, I do not see any tests for your code yet. Tests are important to me, otherwise I cannot refactor fearlessly. One function or...

May I ask if there is any interest in merging this? Or is there some conflict? I'm currently depending on my `regex_lite` fork in [jaq](https://github.com/01mf02/jaq) via `[patch.crates-io]`, but this way,...

Hi @BurntSushi, thanks for answering so quickly! :) I understand about your review bandwidth. I started using `regex-lite` in jaq after noticing that it reduces runtime drastically, especially for small...

For examples of regexes that appear within jqjq, see https://github.com/wader/jqjq/blob/736b06d8a6a2093d8a41527e67d7439247756803/jqjq.jq#L122 and the following 100 lines.

Thanks a lot for your so detailed advice, @BurntSushi! First of all, I do not really know on what kinds of inputs jaq users run regexes. The jqjq example was...

I have now made a prototype that (re-)integrates `regex` into jaq, using a PikeVM as underlying engine. Running jqjq with this prototype is about twice as slow as with `regex-lite`....

> The thing I was mostly trying to emphasize was whether there were use cases with regexes on large haystacks. If those use cases don't exist or are vanishingly rare,...

So I think that I would like to keep `regex-lite` in `jaq` for the time being, given their nice match of design philosophy. If it helps you with the reviewing...

> Thanks! But there are some [subtle differences](https://docs.rs/regex/latest/regex/bytes/index.html#syntax) that I need to think/review carefully. And there is also how iteration is handled in the presence of invalid UTF-8. And making...

@nicowilliams, I understand! Given that you have invested so much work in this, I would be very curious about what you think about my proposed semantics. Does what I describe...