Tyler Kennedy

> > I'd like if there was one spot, but that repo definitely seemed to be abandoned. PRs and issues have been ignored for years. > > We made this...

Can you try it with this branch: https://github.com/androguard/androguard/pull/825

> I am not sure pysimdjson does selective parsing. I think it parses the whole thing. Correct, it's still just using an unmodified simdjson under the hood. It does selective...

> > > Does that mean `yyjson` may become an optional build dependency of simdjson? I'll have to package that for debian as well. It would almost certainly be bundled...

Is the idea to just generate a tape so the same APIs can be used? Ex: ```csv 1,2,3 4,5,6 ``` just turns into a tape of `r[["""]["""]]r` which can be...

Additionally (and why I asked for this feature in the first place), it's pretty important for most of the binding libraries. Both for performance (reusing the JVM/CPython/Ruby heap), and for...

As an example, pybind11 has `py::make_key_iterator()`, which produces an efficient iterator over a mapping's keys. Normally this would "just work" since the iterators are mostly std::iterator-compliant, but they don't return...

This is definitely just a nice-to-have, I wouldn't give it any specific target milestone. It's just a nitpick, since the naive fix adds about [~70LoC](https://github.com/TkTech/pysimdjson/blob/master/simdjson/binding.cpp#L179) to pysimdjson because we can't...

I'm very excited for this. When folks talk about using pysimdjson, ujson, rapidjson, or the alternatives I try to collect their pros/cons. It's almost always the lack of streaming that...

The version is in the header as SIMDJSON_VERSION.