audioresample: On the first buffer we need discont handling

Otherwise we won't get upstream timestamps and everything and all
output buffers would have -1 timestamps.

diff --git a/gst/audioresample/gstaudioresample.c b/gst/audioresample/gstaudioresample.c
index 13db12a..2421208 100644 (file)
--- a/gst/audioresample/gstaudioresample.c
+++ b/gst/audioresample/gstaudioresample.c
@@ -223,8 +223,6 @@ gst_audio_resample_init (GstAudioResample * resample,
 
   resample->quality = SPEEX_RESAMPLER_QUALITY_DEFAULT;
 
-  resample->need_discont = FALSE;
-
   gst_pad_set_query_function (trans->srcpad, gst_audio_resample_query);
   gst_pad_set_query_type_function (trans->srcpad,
       gst_audio_resample_query_type);
@@ -236,6 +234,8 @@ gst_audio_resample_start (GstBaseTransform * base)
 {
   GstAudioResample *resample = GST_AUDIO_RESAMPLE (base);
 
+  resample->need_discont = TRUE;
+
   resample->t0 = GST_CLOCK_TIME_NONE;
   resample->in_offset0 = GST_BUFFER_OFFSET_NONE;
   resample->out_offset0 = GST_BUFFER_OFFSET_NONE;