On Wed, 25 Aug 2010, Dirk Manske wrote:

> My patch fixes that. But indeed it's a bad patch. The operation timeout
> should be set only one time. But I'm not sure where the correct place is. I
> think it must be added to many different states which would be ugly c&p, so
> I've put it into Curl_pgrsTime's TIMER_CONNECT case block:

Thanks a lot for your work and for posting your results here. I plan to
proceed within this area and work on:

A) add the ability to remove only a single specific timeout

B) make sure the expire call from the threaded resolver is done thread-safe

C) Write up a test app and server-side similar to what it sounds like you've
    done and then experiment with the timeouts to make sure that they
    truly trigger as planned when using the multi and multi_socket interfaces.

-- 
  / daniel.haxx.se
-------------------------------------------------------------------
List admin: http://cool.haxx.se/list/listinfo/curl-library
Etiquette:  http://curl.haxx.se/mail/etiquette.html