Kevin Menard

I'm looking at `String#blank?` from ActiveSupport independently of this issue as part of an in-depth look at optimizing regexps in TruffleRuby. I had planned on looking at _fast_blank_, too. I'm...

I've begun looking at this a bit. As Benoit noted, the `new_slow_blank?` method does not match what ActiveSupport does. I dug into the history and it looks like the definition...

As a follow-up, I did extract the code out into simpler benchmarks without the embedded `while` loops and used the _benchmark-ips_ `hold!` method to ensure different benchmarks weren't adversely affecting...

> IMHO benchmark-ips is most reliable in the regular block form (no `while` inside), so if it doesn't make much difference for CRuby I think it's best to use that....

I'm just circling back to this. We made some big improvements to regex processing that should ship in TruffleRuby 22.0.0. Running benchmarks on a native Ryzen 3700X system with Ubuntu...

I see what you're saying for the general case, but by far the most common case that I'm seeing is a `SUBJECT` without any newline characters at all. I strongly...

FYI, I've updated the original benchmarks and results to include cases where only one of the anchors is being used. That way, we can consider cases other than exact string...

For what it's worth, in my extremely non-scientific experience, I've rarely ever encountered anyone using `\A` and `\z`. I'd hazard to say most Rubyists don't even know those exist, especially...

The note about MRI is mostly to avoid entries like you see in the [fast-ruby](https://github.com/JuanitoFatas/fast-ruby) project, where developers end up changing the way they write code in order to better...

@aardvark179 I have a ZFS on Linux system running if you need me to test/verify anything.