On Mon, 9 Sep 2013, Diego Biurrun wrote:
On Tue, Aug 06, 2013 at 09:41:13AM +0200, Martin Storsjö wrote:
--- a/libavformat/network.h
+++ b/libavformat/network.h
@@ -238,11 +238,14 @@ int ff_listen_bind(int fd, const struct sockaddr *addr,
* @param timeout Polling timeout in milliseconds.
* @param h URLContext providing interrupt check
* callback and logging context.
+ * @param will_try_next Whether the caller will try to connect to another
+ * address for the same host name, affecting the form of
+ * logged errors.
* @return 0 on success, AVERROR on failure.
*/
int ff_listen_connect(int fd, const struct sockaddr *addr,
socklen_t addrlen, int timeout,
- URLContext *h);
+ URLContext *h, int will_try_next);
--- a/libavformat/tcp.c
+++ b/libavformat/tcp.c
@@ -100,7 +100,7 @@ static int tcp_open(URLContext *h, const char *uri, int
flags)
}
} else {
if ((ret = ff_listen_connect(fd, cur_ai->ai_addr, cur_ai->ai_addrlen,
- timeout * 100, h)) < 0) {
+ timeout * 100, h, cur_ai->ai_next)) < 0) {
libavformat/tcp.c: In function ‘tcp_open’:
libavformat/tcp.c:103:38: warning: passing argument 6 of ‘ff_listen_connect’
makes integer from pointer without a cast [enabled by default]
This new warning looks scary and I'm surprised this works ...
What's surprising about it? If I want to check whether a pointer is null
or not, I do "if (foo)", I don't normally do "if (foo ? 1 : 0)". This is
pretty much the same, but due to the different context it indeed isn't
technically totally ok to do it like this.
However, if I had written it the right way originally I'm pretty sure
someone would have asked me to simplify it to what we have right now.
Nevertheless, fix patch coming up.
// Martin
_______________________________________________
libav-devel mailing list
[email protected]
https://lists.libav.org/mailman/listinfo/libav-devel
