Closed realgenekim closed 6 months ago
Oops. Forgot to modify documentation in README.md. (I hope I'm doing this correctly — I haven't done this very often!)
Thank you @realgenekim ! I didn't realize we can have such little rate :). Could you please add a test case in tests/diehar/rate_limiter_test.clj
by replacing the current test with a float point rate so that we can verify this behaviour
@realgenekim Are you going to finish this PR? This is a worthwhile addition to the lib
Hi, @kevinmershon and @sunng87 — sorry for the delay on this. I updated the PR that includes a test. (My apologies in advance if I did something wrong. Just let me know, and I will fix.)
LGTM. Thank you for following up.
Thank you for such a wonderful library — I am using this library so often when I hit external API endpoints.
Recently, I started using it to query the OpenAI API, which has token limits that often push the rate limit below one request per second. (I use thread pools via claypoole library to fire off parallel requests)
I found that it was super-easy to specify rate limits of less then 1 req/sec, merely by specifying floats less then 1.0. All I had to do was relax the spec to take
number?
instead ofint?
.PS: I haven't done any formalized or methodical testing, beyond just observation, but it works well enough for my purposes. (Before opening up this PR, I did some more measurements. A rate limit of 0.5 will fire 10 calls in 25 seconds. 0.33 will take 36 seconds. Good enough for me!)