On Sun, 19 Jul 2009, Adrian Batzill wrote:

> The problem was already described here:
> http://curl.haxx.se/mail/lib-2009-01/0218.html> This is basically about how
> CURLOPT_MAX_SEND_SPEED_LARGE/CURLOPT_MAX_RECV_SPEED_LARGE should not use the
> all-time average of a specific download to calculate if libcurl needs to
> speed up/slow down the transfer, but that it should use a timeframe and base
> its calculation on the "current" speed.

[...]

> I just wanted to know what the status about this matter is, since the
> regulation diddn't work too well in my tests.

To my knowledge nobody has worked on it and I don't have it added to my
personal agenda either. Feel free to jump in and send us a patch!

-- 
  / daniel.haxx.se