On Sun, 1 Mar 2015, Raja Gopal M wrote:
Never? So what happens if you call *cleanup on all easy handles and the multi handle? Is that memory truly lost?
Calling *cleanup functions does release the memory. Being said that, my current application which detected the memory leaks only calls *cleanup functions when CURLMSG_DONE for that easy handle is received from libcURL.
Ok, but then I would claim that this isn't a leak but they are simply requests you start that never complete and therefore keep occupying that memory until you clean them up.
I would say that the connect timeout should be used for that. Isn't it being acknowledged?
That is the interesting question I have. CURLOPT_CONNECTTIMEOUT is set (60seconds) for each of the easy handle being added to the multi handle. I read the following line in libcURL documentation, "This only limits the connection phase, it has no impact once it has connected"
Ah yes - my gut reaction was wrong here as you do connect fine. It then just sits waiting to issue its request but it can't do that until the previous request completes.
You can still set CURLOPT_TIMEOUT of course, or one of the suggestions at http://curl.haxx.se/docs/faq.html#I_want_a_different_time_out
-- / daniel.haxx.se ------------------------------------------------------------------- List admin: http://cool.haxx.se/list/listinfo/curl-library Etiquette: http://curl.haxx.se/mail/etiquette.html
