> On Oct 6, 2025, at 16:06, baigao via ffmpeg-devel <[email protected]> 
> wrote:
> 
> ---
> libavformat/whip.c | 4 +++-
> 1 file changed, 3 insertions(+), 1 deletion(-)
> 
> diff --git a/libavformat/whip.c b/libavformat/whip.c
> index 472c6dbf16..6ace9240f9 100644
> --- a/libavformat/whip.c
> +++ b/libavformat/whip.c
> @@ -617,12 +617,14 @@ static int generate_sdp_offer(AVFormatContext *s)
>         "o=FFmpeg %s 2 IN IP4 %s\r\n"
>         "s=FFmpegPublishSession\r\n"
>         "t=0 0\r\n"
> -        "a=group:BUNDLE 0 1\r\n"
>         "a=extmap-allow-mixed\r\n"
>         "a=msid-semantic: WMS\r\n",
>         WHIP_SDP_SESSION_ID,
>         WHIP_SDP_CREATOR_IP);
> 
> +    if (whip->audio_par && whip->video_par)
> +        av_bprintf(&bp, "a=group:BUNDLE 0 1\r\n");
> +
After testing this patch, I found something need to be improved

The bug cause by the a=group:BUNDLE use non-exist mid when there is single 
stream

Some SFUs reject non-bundle SDP(like SRS), and Chrome generate SDP with BUNDLE
even though there is single stream.

So, to improve compatibility, please consider making such change

diff --git a/libavformat/whip.c b/libavformat/whip.c
index 6ace9240f9..5b9520041b 100644
--- a/libavformat/whip.c
+++ b/libavformat/whip.c
@@ -622,8 +622,12 @@ static int generate_sdp_offer(AVFormatContext *s)
         WHIP_SDP_SESSION_ID,
         WHIP_SDP_CREATOR_IP);
 
-    if (whip->audio_par && whip->video_par)
-        av_bprintf(&bp, "a=group:BUNDLE 0 1\r\n");
+    av_bprintf(&bp, "a=group:BUNDLE");
+    if (whip->audio_par)
+        av_bprintf(&bp, " 0");
+    if(whip->video_par)
+        av_bprintf(&bp, " 1");
+    av_bprintf(&bp, "\r\n");
 
     if (whip->audio_par) {
         if (whip->audio_par->codec_id == AV_CODEC_ID_OPUS)
>     if (whip->audio_par) {
>         if (whip->audio_par->codec_id == AV_CODEC_ID_OPUS)
>             acodec_name = "opus";
> -- 
> 2.51.0
> 
> _______________________________________________
> ffmpeg-devel mailing list -- [email protected]
> To unsubscribe send an email to [email protected]

Thanks
Jack
_______________________________________________
ffmpeg-devel mailing list -- [email protected]
To unsubscribe send an email to [email protected]

Reply via email to