[FFmpeg-devel] [PATCH] HLS fixes: better relative path handling, larger buffer for long URIs, only send range header when necessary

[Michael Niedermayer]

On Sun, Sep 23, 2012 at 10:50:05AM +0200, Reimar Döffinger wrote:
> On 23 Sep 2012, at 02:51, Michael Niedermayer <michaelni at gmx.at> wrote:
> > Hi
> > 
> > On Sat, Sep 22, 2012 at 09:17:36PM +0100, Duncan Salerno wrote:
> >> ---
> >> libavformat/http.c  |    6 +++---
> >> libavformat/utils.c |   27 +++++++++++++++++++++++----
> >> 2 files changed, 26 insertions(+), 7 deletions(-)
> >> 
> >> diff --git a/libavformat/http.c b/libavformat/http.c
> >> index 376ff9e..669fdf4 100644
> >> --- a/libavformat/http.c
> >> +++ b/libavformat/http.c
> >> @@ -33,7 +33,7 @@
> >>    only a subset of it. */
> >> 
> >> /* used for protocol handling */
> >> -#define BUFFER_SIZE 1024
> >> +#define BUFFER_SIZE 4096
> >> #define MAX_REDIRECTS 8
> >> 
> >> typedef struct {
> >> @@ -380,7 +380,7 @@ static int http_connect(URLContext *h, const char *path, const char *local_path,
> >> {
> >>     HTTPContext *s = h->priv_data;
> >>     int post, err;
> >> -    char headers[1024] = "";
> >> +    char headers[4096] = "";
> >>     char *authstr = NULL, *proxyauthstr = NULL;
> >>     int64_t off = s->off;
> >>     int len = 0;
> >> @@ -411,7 +411,7 @@ static int http_connect(URLContext *h, const char *path, const char *local_path,
> >>     if (!has_header(s->headers, "\r\nAccept: "))
> >>         len += av_strlcpy(headers + len, "Accept: */*\r\n",
> >>                           sizeof(headers) - len);
> >> -    if (!has_header(s->headers, "\r\nRange: ") && !post)
> >> +    if (!has_header(s->headers, "\r\nRange: ") && !post && s->off > 0)
> >>         len += av_strlcatf(headers + len, sizeof(headers) - len,
> >>                            "Range: bytes=%"PRId64"-\r\n", s->off);
> >> 
> > 
> > split and applied above
> 
> May I ask if there is a specific reason for that change?
> I haven't double-checked if we use it that way, but detecting if seeking works can be difficult because many servers set the wrong things in response headers.
> I believe setting a Range in the request makes those give you a few more hints whether they support it or not.

i thought the same yesterday but then after a quick look in the http
spec thought its ok, a longer look in the spec now and actual testing
though confirms that we loose some hints on seekability with this
change, thus
reverted

[...]
-- 
Michael     GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB

While the State exists there can be no freedom; when there is freedom there
will be no State. -- Vladimir Lenin
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 198 bytes
Desc: Digital signature
URL: <http://ffmpeg.org/pipermail/ffmpeg-devel/attachments/20120923/24230fd8/attachment.asc>