On 02/01/2011 01:07 AM, Daniel Stenberg wrote:
On Sun, 30 Jan 2011, Nicholas Maniscalco wrote:
It appears that multi_runsingle may be leaving sockets in
CURLM_STATE_TOOFAST for too long because it is working off of
outdated progress data. Attached is a patch that calls
Curl_pgrsUpdate just inside the TOOFAST case
(multi_toofast_update_progress.patch). With this patch, I observe a
nice smooth rate-limited transfer (see hiperfifo_output_good.txt).
Is there a better approach to fixing this?
Thanks a lot. I think this is a good and cheap fix!
Thanks for looking. I've tested the patch in the application I'm
working on and it behaves as I'd expect. Will it (or an equivalent) be
included in the next release?
-------------------------------------------------------------------
List admin: http://cool.haxx.se/list/listinfo/curl-library
Etiquette: http://curl.haxx.se/mail/etiquette.html
