from Hacker News

Ask HN: How fast can you hit all 34M items in Hacker News Firebase API?

by marsupialtail_2 on 2/20/23, 3:50 AM with 5 comments

This could be a nice benchmark for a http client. I wonder if anybody with experience with reqwest in Rust could chime in. Currently using aiohttp I can get around 1k requests/s from the Hacker News API, which means it will take around 9 hours to get all the items....

  • by krapp on 2/20/23, 10:38 AM

    That depends entirely on your network latency and connection speed, doesn't it? I think a better benchmark for a client would be generating pages/threads, because HN's data structures are so suboptimal for that even HN itself sometimes has trouble with it. So take a super long thread like https://news.ycombinator.com/item?id=12907201 and see how long it takes to render it fully.

  • by aynyc on 2/20/23, 12:37 PM

    I’ve done it maybe 2 years ago learning pythons asyncio. I remember the time is around 3.5 hours. I only perform the scrape maybe 5-6 times over a week to avoid overloading the server.

  • by avinassh on 2/20/23, 6:50 AM

    If it makes your experiment any easier, I have written a hacker news API wrapper in Python - https://github.com/avinassh/haxor

  • by sc90 on 2/20/23, 8:10 AM

    Isn't it rate limited?