Yes, an API that is isomorphic to the website itself would be much easier to use. Our plan is for the next version of HN's API to simply serve a JSON version of every page.
In the meantime, if you scrape the HTML, please respect https://news.ycombinator.com/robots.txt and don't do it more than once per 30 seconds. HN's app server runs on a single core, so we don't have a lot of performance to spare! This is in the works to change as well.
> HN's app server runs on a single core, so we don't have a lot of performance to spare!
That's interesting. Is there any more information about HN's tech stack? I know it's built on top of the Lisp dialect Arc, but I wonder about the server specs and would be happy to be pointed at some resource.
Although I experienced HN being down, I guess the overall uptime and availability is pretty decent. Keep up the good work!
This still holds: https://news.ycombinator.com/item?id=16076041
Number of requests is closer to 5M per day now.
Nice, thanks! :)
HN uses Cloudflare, doesn’t it? I would’ve assumed reads would be heavily cached.
Not any more.
What made you switch / leave CloudFlare?
The desire to have fewer moving parts.
I stand corrected.