Interesting. I would think that either cherry-picking the subsystems of Hyperscan that provide the most performance benefit and integrating them into ripgrep, or cherry-picking the features they want from ripgrep and integrating them into HS or its tools, would be the way.
-
-
-
Both of those sound like a tremendous amount of work compared to this approach honestly. ripgrep already supports two different regex engines. I built it such that _most_ pieces are agnostic to the underlying engine.
End of conversation
New conversation -
-
-
Wait... "it starts to be useful when having at least 1000 regexps to parse more than 10GB of data"? That's... Quite extreme? I mean I know hyperscan is about multiple/concurrent regexps, but no less than 1000...?
-
I would be very surprised if this were actually true. We crossed over vs RE2::Set in tens/hundreds of regexes, and 10GB of data is quite a lot. Maybe they are factoring in compile times... which is legit if you aren't reusing your bytecodes.
- 6 more replies
New conversation -
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.