Hacker News Reader

Comment by TZubiri | Hacker News Reader

[flagged]

Maybe you should’ve spent 2 minutes reading the article instead of arrogantly dismissing it with layman knowledge.

The article explains optimizations to spend less cycles parsing URLs than other libraries. Very interesting work, there's no reason not to do things efficiently when it's possible.

Also, good luck using regex to write a RFC or WHATWG conformant URL parser.

2 minutes reading an rfc about uris and I find a regex literally used in the specs:

https://www.rfc-editor.org/rfc/rfc3986#page-7

>> The following line is the regular expression for breaking-down a well-formed URI reference into its components.

>> ^(([^:/?#]+):)?(//([^/?#]))?([^?#])(\?([^#]))?(#(.))?

Did Chomsky die for nothing?

I guess there are reasons to do things efficiently when possible, but a million URLs is not it, the adage about the root of all evil comes to mind. A billion URLs per second and it's almost interesting, but not really.

RFC 3986 is a lot simpler than WHATWG spec. You can literally write a zero copy 3986 parser whereas you can’t with WHATWG. (And Ada is still faster than 3986 parsers)

Last I heard Noam Chomsky was still alive, and a quick Google doesn’t contradict that. Or is this some kind of high brow joke that went over my head?

https://www.cbc.ca/news/world/noam-chomsky-not-dead-1.723937...

Damn. I got fake newsd. It was probably a close call.

My commiseration email must have looked so silly.

I think they're joking about Chomsky's work with formal languages (https://en.wikipedia.org/wiki/Chomsky_hierarchy).

That doesn't normalize the URL nor does it handle relative URL joining logic. It also doesn't handle URLs like: `file:///foo.txt`

The spec provides an expression for "well-formed" URIs. Good luck with real-world input.

That will get you only one million per second.

And depending on the length of your URL, 4000 will not be trivial, depending on the output format.

Correct, URLs can be something like 4,000 characters long in 15 year old Firefox. I wonder what the current maximum length is?

Today, Chrome supports 32,768 characters.. good luck processing that in 4,000 cycles! It'd require SIMD or some other fanciness.

> That will get you only one million per second.

But per core, right?

Reply on news.ycombinator.com