On Sat, Nov 15, 2014 at 6:23 AM, Michael Niedermayer <michae...@gmx.at> wrote:
> On Fri, Nov 14, 2014 at 06:05:44PM -0500, Brandon Lees wrote: > > In http_open_cnx, the patch restores the AVDictionary if connection > needs to be re-tried > > because of a authentication/redirect status code. > > > > Previously, if a 401/407/30x status code was encountered, http_open_cnx > would restart at the redo label, but any options > > used by the underlying protocol would be missing because they were > removed by the first attempt. > > > > Signed-off-by: Brandon Lees <bran...@n-hega.com> > > --- > > libavformat/http.c | 3 +++ > > 1 file changed, 3 insertions(+) > > > > diff --git a/libavformat/http.c b/libavformat/http.c > > index dfc01ee..f17b2f8 100644 > > --- a/libavformat/http.c > > +++ b/libavformat/http.c > > @@ -207,6 +207,9 @@ static int http_open_cnx(URLContext *h, AVDictionary > **options) > > HTTPContext *s = h->priv_data; > > int location_changed, attempts = 0, redirects = 0; > > redo: > > + if (attempts > 0) > > + av_dict_copy(options, s->chained_options, 0); > > i might be missing something, but why is this conditional on > attempts > 0 ? > i think if its unconditional some calling code could be simplified > > [...] > -- > Michael GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB > > In fact, the RIAA has been known to suggest that students drop out > of college or go to community college in order to be able to afford > settlements. -- The RIAA > the condition is there because for the first attempt (attempt == 0) options are already set to the correct value so the copy is unnecessary. I guess it just depends on if you prefer an extra copy or an extra line of code. Brandon _______________________________________________ ffmpeg-devel mailing list ffmpeg-devel@ffmpeg.org http://ffmpeg.org/mailman/listinfo/ffmpeg-devel