Re: Infinite loop when calling curl_multi_cleanup

2016-10-20 Thread Valentin David
there is a request sent, but no answer received yet. We remove the easy from the multi. Then we close the multi. But there might not have been any curl_multi_socket_action called between. -- Valentin David --- List admin: https://cool.ha

Infinite loop when calling curl_multi_cleanup

2016-10-17 Thread Valentin David
isconnect" would not close properly the connection and "close_all_connections" would call it in loop forever. Adding "connclose" before call the "Curl_disconnect" fixes the problem. -- Valentin David diff -ur curl-7.50.3.old/lib/multi.c curl-7.50.3/lib/multi.c --- curl-7.50

Using curl_multi some requests get ignored after connection succeeded

2017-09-01 Thread Valentin David
we notify curl. But curl just asks to remove the socket from polling, and it never sends the request. I have made an example for Linux (epoll) that fails to complete on cURL 7.55.0 and 7.55.1, but succeeds in 7.54.1, though sometimes it can get a bit slow. -- Valentin David This email refle

Re: Using curl_multi some requests get ignored after connection succeeded

2017-09-05 Thread Valentin David
On Tue, 2017-09-05 at 12:59 +0200, Daniel Stenberg wrote: > On Tue, 5 Sep 2017, Valentin David wrote: > > > > Thanks. Can you elborate on how exactly this test application > > > reacts when  > > > the  problem occurs? I mean, how do I det

Re: Using curl_multi some requests get ignored after connection succeeded

2017-09-05 Thread Valentin David
On Mon, 2017-09-04 at 20:23 +0200, Daniel Stenberg wrote: > On Fri, 1 Sep 2017, Valentin David wrote: > > > I have made an example for Linux (epoll) that fails to complete on > > cURL  > > 7.55.0 and 7.55.1, but succeeds in 7.54.1, though sometimes it can > > get a b

Re: Using curl_multi some requests get ignored after connection succeeded

2017-09-05 Thread Valentin David
On Tue, 2017-09-05 at 14:12 +0200, Daniel Stenberg wrote: > On Tue, 5 Sep 2017, Valentin David wrote: > > > Master *is not* affected, 7.55.1 *is* affected however. So > > something broke  > > but got fixed. > > Ah, cool. I suspect 74dac344b2f is what fixed it,