Some people recently expressed interest in a more precise implementation of --limit-rate and I had a stab at it the other day. Please see:
https://github.com/curl/curl/pull/19384 - increased precision on all HTTP protocols - improved stream window handling on HTTP/2 transfers - implementation of rate limiting in HTTP/3 transfers with ngtcp2 The increased precision should not come at extra cpu costs. If you would take that out for a spin in your environment, I'd be happy to get some feedback on this change. Kind Regards, Stefan -- Unsubscribe: https://lists.haxx.se/mailman/listinfo/curl-library Etiquette: https://curl.se/mail/etiquette.html
