Daniel Lemire
Daniel Lemire
> How would you recommend integrating this into iterate_many? I am not sure. I am opening this for discussion. Note that simdjson is a community-based project.
@prashu0705 Sure. Note that we want to preserve backward compatibility. Thus I would prefer if you we do not change the existing functions. (It is bad form to break existing...
@anonrig It seems to break the tests.
@XueSongTap Yep! Conceptually, this is 'easy', but practically, it could be less trivial and require a bit of experimentation. Try something and report back.
> I don't think this would be as easy as first imagined. That’s entirely possible. > We could fix all of those if there was an intermediary representation of the...
The following commit https://github.com/simdjson/simdjson/pull/2412/commits/b7684711dfef7c2b6087f28e38d988e5bc5e36cb had a complete implementation, but I am removing it as it offers poor performance. There is a high per-element value when iterating over an array. Preliminary...
@HFTrader Thanks for the PR. The point of the blog post is to explain why we need SIMD instructions. Is it your understanding that memchr DOES NOT use SIMD instructions...
> Based on the comments in these issues, it seems like you are likely to accept well-written PRs enabling non-standard "relaxed" JSON syntax. But I just wanted to double-check before...
Yeah. I have frozen this PR for now because I am no longer convinced that the library should prescribe the behavior.
@FranciscoThiesen Running tests.