API rate limit

bodleianwmss

Hi,
We (Bodleian Library Oxford) are looking to use Zotero to host an electronic version of our card index bibliography of Western medieval manuscripts. Users of the online medieval manuscript catalogue will click through from the catalogue record to access the bibliography. Behind the scenes there would be a call to the Zotero API. My question relates to the rate limits for the API, which I can't find any specific information on in the documentation - can anyone help out?

Thanks!

adamsmith

By specific you're referring to specific numbers, right? You've seen https://www.zotero.org/support/dev/web_api/v3/basics#rate_limiting I assume?
I think Dan is the only one with the answer to that.

bodleianwmss

Thanks; yes, we're looking for specific numbers (or at least relatively specific, likely maximum of typical requests allowed per minute for example).

dstillman

There aren't any specific numbers. Basically, as noted in the linked section, your code should be prepared to handle any Backoff or Retry-After headers automatically. Otherwise you can make a reasonable number of requests and expect not to run into trouble.

What constitutes a reasonable number of requests depends on how you implement the integration. If your server will be performing the API requests, you'll want to do some caching, since there'd be no need to hit the API for every visitor and you'd be more likely to have your server's IP address rate-limited. If you do this entirely in client-side JS, with API requests coming directly from your visitors, you could consider doing some client-side caching so that browsing around the library was quicker, but visitors likely wouldn't be rate-limited, since they'd only be making a few requests each. (We reserve the right to limit the number of API requests made to a given library, but that's never been necessary, and we'd try to get in touch first if it were to discuss alternative approaches.)