Integrate with html-parsers-benchmark

[servo]

Integrate with html-parsers-benchmark

[Ygg01]

I used those benchmarks and kinda managed to make Rust go through benchmarks. Alas, the output from Rust is jumbled (I assume due to multi-threading) and the shell program doesn't correctly recognize Rust's results.

[kmcallister]

Cool, can I see your scripts?

[Ygg01]

Yeah. I'm was out of town for a bit. I'll need to push some changes once I'm back. I apologize for the inconvenience.

[kmcallister]

No trouble at all :) I certainly have enough other things to work on...

I'll be glad to take a look whenever it's ready.

[nox]

@Ygg01 Ping?

[Ygg01]

@nox Pong :P

Sorry, RL has a way of interfering with best laid plan of mice and men.

Anyway, this been a long time, since I experimented on this, I made it work, but the Rust console output was kinda hard to get to work with what benchmark expected (the lines came out in slightly wrong, non-deterministic order), confusing the hell out of the benchmark tool.

Also a big problem is that it's not Rust friendly tool. It's a collection of bash and python scripts, so running it via rust bench, is kinda hard.

To make things worse, the benchmark wasn't updated in quite a while (Aug 2013).

[ordian]

Also, take a look at myhtml and benchmarks:
results
code

Basically, this benchmark evaluates time of parsing TOP500 alexa sites in memory.

parser	myhtml	gumbo	html5ever
overall time	0.50890	6.12951	4.50536
memory	32120832	29319168	30715904

[Ygg01]

So how do you want this done?

As a shell command that calls different html and parses them? Should we attempt to make our own benchmark or attempt to extend existing benchmarks?

[nox]

Closing this for now.