RELEASE: GStreamer FFmpeg Plug-ins 0.10.12 "A year in hell"

[Edward Hervey]

This mail announces the release of GStreamer FFmpeg Plug-ins 0.10.12 "A
year in hell"

GStreamer FFmpeg plug-in contains  elements using the FFmpeg library
code. It contains most popular decoders and encoders.

Highlights of this release:
      * Bump to LibAV 0.7
      * Timestamp tracking fixes
      * New codecs
      * Multi-threaded decoding


For more information, see
http://gstreamer.freedesktop.org/modules/gst-ffmpeg.html
To file bugs, go to
http://bugzilla.gnome.org/enter_bug.cgi?product=GStreamer&component=gst-ffmpeg

Direct links:
http://gstreamer.freedesktop.org/data/src/gst-ffmpeg/gst-ffmpeg-0.10.12.tar.bz2
http://gstreamer.freedesktop.org/data/src/gst-ffmpeg/gst-ffmpeg-0.10.12.tar.gz

_______________________________________________
gstreamer-devel mailing list
[hidden email]
http://lists.freedesktop.org/mailman/listinfo/gstreamer-devel

[Kocsis Tibor]

Hi,

I can't found the mpeg4 demuxer under the name
ffdemux_mov_mp4_m4a_3gp_3g2_mj2 in this release. Is there any new
plugin for that?

Regards,
Tibor


On Wed, Jul 20, 2011 at 1:02 PM, Edward Hervey <[hidden email]> wrote:

> This mail announces the release of GStreamer FFmpeg Plug-ins 0.10.12 "A
> year in hell"
>
> GStreamer FFmpeg plug-in contains  elements using the FFmpeg library
> code. It contains most popular decoders and encoders.
>
> Highlights of this release:
>      * Bump to LibAV 0.7
>      * Timestamp tracking fixes
>      * New codecs
>      * Multi-threaded decoding
>
>
> For more information, see
> http://gstreamer.freedesktop.org/modules/gst-ffmpeg.html
> To file bugs, go to
> http://bugzilla.gnome.org/enter_bug.cgi?product=GStreamer&component=gst-ffmpeg
>
> Direct links:
> http://gstreamer.freedesktop.org/data/src/gst-ffmpeg/gst-ffmpeg-0.10.12.tar.bz2
> http://gstreamer.freedesktop.org/data/src/gst-ffmpeg/gst-ffmpeg-0.10.12.tar.gz
>
> _______________________________________________
> gstreamer-devel mailing list
> [hidden email]
> http://lists.freedesktop.org/mailman/listinfo/gstreamer-devel
>
>

_______________________________________________
gstreamer-devel mailing list
[hidden email]
http://lists.freedesktop.org/mailman/listinfo/gstreamer-devel

[Sebastian Dröge-7]

On Wed, 2011-07-27 at 09:05 +0200, Kocsis Tibor wrote:
> Hi,
>
> I can't found the mpeg4 demuxer under the name
> ffdemux_mov_mp4_m4a_3gp_3g2_mj2 in this release. Is there any new
> plugin for that?

qtdemux in gst-plugins-good, which is the recommended
Quicktime/MP4/3GPP/etc demuxer since years.

_______________________________________________
gstreamer-devel mailing list
[hidden email]
http://lists.freedesktop.org/mailman/listinfo/gstreamer-devel

[Tvrtko Ursulin]

On Wednesday 20 Jul 2011 12:02:00 Edward Hervey wrote:

> This mail announces the release of GStreamer FFmpeg Plug-ins 0.10.12 "A
> year in hell"
>
> GStreamer FFmpeg plug-in contains  elements using the FFmpeg library
> code. It contains most popular decoders and encoders.
>
> Highlights of this release:
>       * Bump to LibAV 0.7
>       * Timestamp tracking fixes
>       * New codecs
>       * Multi-threaded decoding

I see that multi-threaded decoding is disabled by default in gstffmpegdec. Is this
because of known issues and how serious are they?

With regards to ffdeinterlace - I did a pretty mechanical (not understading GStreamer
internals and code base that well) patch  porting the mode property from good plugins
deinterlace because having this control and automatic mode is quite convenient for us.

Would something like this be acceptable for merging upstream? At least in principle, I realize patch might have issues. :)

Thanks,

Tvrtko


diff -uprN a/gst-ffmpeg-0.10.12/ext/ffmpeg/gstffmpegdeinterlace.c b/gst-ffmpeg-0.10.12/ext/ffmpeg/gstffmpegdeinterlace.c
--- a/gst-ffmpeg-0.10.12/ext/ffmpeg/gstffmpegdeinterlace.c 2011-07-13 10:07:28.000000000 +0100
+++ b/gst-ffmpeg-0.10.12/ext/ffmpeg/gstffmpegdeinterlace.c 2011-08-04 11:11:30.542756021 +0100
@@ -37,6 +37,44 @@
 #include "gstffmpegcodecmap.h"
 #include "gstffmpegutils.h"
 
+
+/* Properties */
+
+#define DEFAULT_MODE            GST_FFMPEGDEINTERLACE_MODE_AUTO
+
+enum
+{
+  PROP_0,
+  PROP_MODE,
+  PROP_LAST
+};
+
+typedef enum {
+  GST_FFMPEGDEINTERLACE_MODE_AUTO,
+  GST_FFMPEGDEINTERLACE_MODE_INTERLACED,
+  GST_FFMPEGDEINTERLACE_MODE_DISABLED
+} GstFFMpegDeinterlaceMode;
+
+#define GST_TYPE_FFMPEGDEINTERLACE_MODES (gst_ffmpegdeinterlace_modes_get_type ())
+static GType
+gst_ffmpegdeinterlace_modes_get_type (void)
+{
+  static GType deinterlace_modes_type = 0;
+
+  static const GEnumValue modes_types[] = {
+    {GST_FFMPEGDEINTERLACE_MODE_AUTO, "Auto detection", "auto"},
+    {GST_FFMPEGDEINTERLACE_MODE_INTERLACED, "Force deinterlacing", "interlaced"},
+    {GST_FFMPEGDEINTERLACE_MODE_DISABLED, "Run in passthrough mode", "disabled"},
+    {0, NULL, NULL},
+  };
+
+  if (!deinterlace_modes_type) {
+    deinterlace_modes_type =
+        g_enum_register_static ("GstFFMpegDeinterlaceModes", modes_types);
+  }
+  return deinterlace_modes_type;
+}
+
 typedef struct _GstFFMpegDeinterlace
 {
   GstElement element;
@@ -46,6 +84,14 @@ typedef struct _GstFFMpegDeinterlace
   gint width, height;
   gint to_size;
 
+  GstFFMpegDeinterlaceMode mode;
+  
+  gboolean interlaced; /* is input interlaced? */
+  gboolean passthrough;
+  
+  gboolean reconfigure;
+  GstFFMpegDeinterlaceMode new_mode;
+  
   enum PixelFormat pixfmt;
   AVPicture from_frame, to_frame;
 } GstFFMpegDeinterlace;
@@ -68,6 +114,11 @@ typedef struct _GstFFMpegDeinterlaceClas
 
 GType gst_ffmpegdeinterlace_get_type (void);
 
+static void gst_ffmpegdeinterlace_set_property (GObject * self, guint prop_id,
+    const GValue * value, GParamSpec * pspec);
+static void gst_ffmpegdeinterlace_get_property (GObject * self, guint prop_id,
+    GValue * value, GParamSpec * pspec);
+
 static GstStaticPadTemplate src_factory = GST_STATIC_PAD_TEMPLATE ("src",
     GST_PAD_SRC,
     GST_PAD_ALWAYS,
@@ -103,6 +154,34 @@ gst_ffmpegdeinterlace_base_init (gpointe
 static void
 gst_ffmpegdeinterlace_class_init (GstFFMpegDeinterlaceClass * klass)
 {
+  GObjectClass *gobject_class = (GObjectClass *) klass;
+
+  gobject_class->set_property = gst_ffmpegdeinterlace_set_property;
+  gobject_class->get_property = gst_ffmpegdeinterlace_get_property;
+
+  /**
+   * GstDeinterlace:mode
+   *
+   * This selects whether the deinterlacing methods should
+   * always be applied or if they should only be applied
+   * on content that has the "interlaced" flag on the caps.
+   *
+   */
+  g_object_class_install_property (gobject_class, PROP_MODE,
+      g_param_spec_enum ("mode",
+          "Mode",
+          "Deinterlace Mode",
+          GST_TYPE_FFMPEGDEINTERLACE_MODES,
+          DEFAULT_MODE, G_PARAM_READWRITE | G_PARAM_STATIC_STRINGS)
+      );
+}
+
+static void
+gst_ffmpegdeinterlace_update_passthrough (GstFFMpegDeinterlace * deinterlace)
+{
+  deinterlace->passthrough = (deinterlace->mode == GST_FFMPEGDEINTERLACE_MODE_DISABLED
+      || (!deinterlace->interlaced && deinterlace->mode != GST_FFMPEGDEINTERLACE_MODE_INTERLACED));
+  GST_DEBUG_OBJECT (deinterlace, "Passthrough: %d", deinterlace->passthrough);
 }
 
 static gboolean
@@ -120,6 +199,10 @@ gst_ffmpegdeinterlace_sink_setcaps (GstP
   if (!gst_structure_get_int (structure, "height", &deinterlace->height))
     return FALSE;
 
+  deinterlace->interlaced = FALSE;
+  gst_structure_get_boolean (structure, "interlaced", &deinterlace->interlaced);
+  gst_ffmpegdeinterlace_update_passthrough (deinterlace);
+
   ctx = avcodec_alloc_context ();
   ctx->width = deinterlace->width;
   ctx->height = deinterlace->height;
@@ -138,10 +221,10 @@ gst_ffmpegdeinterlace_sink_setcaps (GstP
       avpicture_get_size (deinterlace->pixfmt, deinterlace->width,
       deinterlace->height);
 
-  src_caps = gst_caps_copy (caps);
-  gst_caps_set_simple (src_caps, "interlaced", G_TYPE_BOOLEAN, FALSE, NULL);
-  ret = gst_pad_set_caps (deinterlace->srcpad, src_caps);
-  gst_caps_unref (src_caps);
+  src_caps = gst_caps_copy (caps);
+  gst_caps_set_simple (src_caps, "interlaced", G_TYPE_BOOLEAN, deinterlace->interlaced, NULL);
+  ret = gst_pad_set_caps (deinterlace->srcpad, src_caps);
+  gst_caps_unref (src_caps);
 
   return ret;
 }
@@ -162,6 +245,12 @@ gst_ffmpegdeinterlace_init (GstFFMpegDei
   gst_element_add_pad (GST_ELEMENT (deinterlace), deinterlace->srcpad);
 
   deinterlace->pixfmt = PIX_FMT_NB;
+
+  deinterlace->interlaced = FALSE;
+  deinterlace->passthrough = FALSE;
+  deinterlace->reconfigure = FALSE;
+  deinterlace->mode = DEFAULT_MODE;
+  deinterlace->new_mode = -1;
 }
 
 static GstFlowReturn
@@ -172,6 +261,23 @@ gst_ffmpegdeinterlace_chain (GstPad * pa
   GstBuffer *outbuf = NULL;
   GstFlowReturn result;
 
+  GST_OBJECT_LOCK (deinterlace);
+  if (deinterlace->reconfigure) {
+    if (deinterlace->new_mode != -1)
+      deinterlace->mode = deinterlace->new_mode;
+    deinterlace->new_mode = -1;
+
+    deinterlace->reconfigure = FALSE;
+    GST_OBJECT_UNLOCK (deinterlace);
+    if (GST_PAD_CAPS (deinterlace->srcpad))
+      gst_ffmpegdeinterlace_sink_setcaps (deinterlace->sinkpad, GST_PAD_CAPS (deinterlace->sinkpad));
+  } else {
+    GST_OBJECT_UNLOCK (deinterlace);
+  }
+  
+  if (deinterlace->passthrough)
+      return gst_pad_push (deinterlace->srcpad, inbuf);
+
   result =
       gst_pad_alloc_buffer (deinterlace->srcpad, GST_BUFFER_OFFSET_NONE,
       deinterlace->to_size, GST_PAD_CAPS (deinterlace->srcpad), &outbuf);
@@ -202,3 +308,53 @@ gst_ffmpegdeinterlace_register (GstPlugi
   return gst_element_register (plugin, "ffdeinterlace",
       GST_RANK_NONE, GST_TYPE_FFMPEGDEINTERLACE);
 }
+
+static void
+gst_ffmpegdeinterlace_set_property (GObject * object, guint prop_id,
+    const GValue * value, GParamSpec * pspec)
+{
+  GstFFMpegDeinterlace *self;
+
+  g_return_if_fail (GST_IS_FFMPEGDEINTERLACE (object));
+  self = GST_FFMPEGDEINTERLACE (object);
+
+  switch (prop_id) {
+    case PROP_MODE:{
+      gint new_mode;
+
+      GST_OBJECT_LOCK (self);
+      new_mode = g_value_get_enum (value);
+      if (self->mode != new_mode && GST_PAD_CAPS (self->srcpad)) {
+        self->reconfigure = TRUE;
+        self->new_mode = new_mode;
+      } else {
+        self->mode = new_mode;
+        gst_ffmpegdeinterlace_update_passthrough (self);
+      }
+      GST_OBJECT_UNLOCK (self);
+      break;
+    }
+    default:
+      G_OBJECT_WARN_INVALID_PROPERTY_ID (self, prop_id, pspec);
+  }
+
+}
+
+static void
+gst_ffmpegdeinterlace_get_property (GObject * object, guint prop_id,
+    GValue * value, GParamSpec * pspec)
+{
+  GstFFMpegDeinterlace *self;
+
+  g_return_if_fail (GST_IS_FFMPEGDEINTERLACE (object));
+  self = GST_FFMPEGDEINTERLACE (object);
+
+  switch (prop_id) {
+    case PROP_MODE:
+      g_value_set_enum (value, self->mode);
+      break;
+    default:
+      G_OBJECT_WARN_INVALID_PROPERTY_ID (self, prop_id, pspec);
+  }
+}
+
_______________________________________________
gstreamer-devel mailing list
[hidden email]
http://lists.freedesktop.org/mailman/listinfo/gstreamer-devel

[Edward Hervey]

Hi,

On Thu, 2011-08-11 at 10:02 +0100, Tvrtko Ursulin wrote:
> On Wednesday 20 Jul 2011 12:02:00 Edward Hervey wrote:
[...]
> I see that multi-threaded decoding is disabled by default in gstffmpegdec. Is this
> because of known issues and how serious are they?

  The support for it was only dropped a month before release. The reason
why I set the default back to only using one thread was:
  * Lots of warnings from the libav side on some formats
  * Reports of segfaulting in some situations
  * Critical basic MT-safety patches landing the day I was going to do
the release.

  That being said, the multi-threading support in gst-ffmpeg hasn't been
removed/disabled. It's just that I've re-setted the default number of
threads to 1 (instead of 0 which is automatic numbers).
  So you can still force the number of threads you want manually.

  Hopefully everything will be more stable soon and we'll be able to do
a new gst-ffmpeg release with the default multi-threading enabled within
a couple of months.

>
> With regards to ffdeinterlace - I did a pretty mechanical (not understading GStreamer
> internals and code base that well) patch  porting the mode property from good plugins
> deinterlace because having this control and automatic mode is quite convenient for us.
>
> Would something like this be acceptable for merging upstream? At least in principle, I realize patch might have issues. :)
>

  Put that patch in a new bug report on bugzilla.gnome.org . It'll be
handled there like all patches. I have no objections per-se.

    Edward

> Thanks,
>
> Tvrtko
>
>
> diff -uprN a/gst-ffmpeg-0.10.12/ext/ffmpeg/gstffmpegdeinterlace.c b/gst-ffmpeg-0.10.12/ext/ffmpeg/gstffmpegdeinterlace.c
> --- a/gst-ffmpeg-0.10.12/ext/ffmpeg/gstffmpegdeinterlace.c 2011-07-13 10:07:28.000000000 +0100
> +++ b/gst-ffmpeg-0.10.12/ext/ffmpeg/gstffmpegdeinterlace.c 2011-08-04 11:11:30.542756021 +0100
> @@ -37,6 +37,44 @@
>  #include "gstffmpegcodecmap.h"
>  #include "gstffmpegutils.h"
>  
> +
> +/* Properties */
> +
> +#define DEFAULT_MODE            GST_FFMPEGDEINTERLACE_MODE_AUTO
> +
> +enum
> +{
> +  PROP_0,
> +  PROP_MODE,
> +  PROP_LAST
> +};
> +
> +typedef enum {
> +  GST_FFMPEGDEINTERLACE_MODE_AUTO,
> +  GST_FFMPEGDEINTERLACE_MODE_INTERLACED,
> +  GST_FFMPEGDEINTERLACE_MODE_DISABLED
> +} GstFFMpegDeinterlaceMode;
> +
> +#define GST_TYPE_FFMPEGDEINTERLACE_MODES (gst_ffmpegdeinterlace_modes_get_type ())
> +static GType
> +gst_ffmpegdeinterlace_modes_get_type (void)
> +{
> +  static GType deinterlace_modes_type = 0;
> +
> +  static const GEnumValue modes_types[] = {
> +    {GST_FFMPEGDEINTERLACE_MODE_AUTO, "Auto detection", "auto"},
> +    {GST_FFMPEGDEINTERLACE_MODE_INTERLACED, "Force deinterlacing", "interlaced"},
> +    {GST_FFMPEGDEINTERLACE_MODE_DISABLED, "Run in passthrough mode", "disabled"},
> +    {0, NULL, NULL},
> +  };
> +
> +  if (!deinterlace_modes_type) {
> +    deinterlace_modes_type =
> +        g_enum_register_static ("GstFFMpegDeinterlaceModes", modes_types);
> +  }
> +  return deinterlace_modes_type;
> +}
> +
>  typedef struct _GstFFMpegDeinterlace
>  {
>    GstElement element;
> @@ -46,6 +84,14 @@ typedef struct _GstFFMpegDeinterlace
>    gint width, height;
>    gint to_size;
>  
> +  GstFFMpegDeinterlaceMode mode;
> +  
> +  gboolean interlaced; /* is input interlaced? */
> +  gboolean passthrough;
> +  
> +  gboolean reconfigure;
> +  GstFFMpegDeinterlaceMode new_mode;
> +  
>    enum PixelFormat pixfmt;
>    AVPicture from_frame, to_frame;
>  } GstFFMpegDeinterlace;
> @@ -68,6 +114,11 @@ typedef struct _GstFFMpegDeinterlaceClas
>  
>  GType gst_ffmpegdeinterlace_get_type (void);
>  
> +static void gst_ffmpegdeinterlace_set_property (GObject * self, guint prop_id,
> +    const GValue * value, GParamSpec * pspec);
> +static void gst_ffmpegdeinterlace_get_property (GObject * self, guint prop_id,
> +    GValue * value, GParamSpec * pspec);
> +
>  static GstStaticPadTemplate src_factory = GST_STATIC_PAD_TEMPLATE ("src",
>      GST_PAD_SRC,
>      GST_PAD_ALWAYS,
> @@ -103,6 +154,34 @@ gst_ffmpegdeinterlace_base_init (gpointe
>  static void
>  gst_ffmpegdeinterlace_class_init (GstFFMpegDeinterlaceClass * klass)
>  {
> +  GObjectClass *gobject_class = (GObjectClass *) klass;
> +
> +  gobject_class->set_property = gst_ffmpegdeinterlace_set_property;
> +  gobject_class->get_property = gst_ffmpegdeinterlace_get_property;
> +
> +  /**
> +   * GstDeinterlace:mode
> +   *
> +   * This selects whether the deinterlacing methods should
> +   * always be applied or if they should only be applied
> +   * on content that has the "interlaced" flag on the caps.
> +   *
> +   */
> +  g_object_class_install_property (gobject_class, PROP_MODE,
> +      g_param_spec_enum ("mode",
> +          "Mode",
> +          "Deinterlace Mode",
> +          GST_TYPE_FFMPEGDEINTERLACE_MODES,
> +          DEFAULT_MODE, G_PARAM_READWRITE | G_PARAM_STATIC_STRINGS)
> +      );
> +}
> +
> +static void
> +gst_ffmpegdeinterlace_update_passthrough (GstFFMpegDeinterlace * deinterlace)
> +{
> +  deinterlace->passthrough = (deinterlace->mode == GST_FFMPEGDEINTERLACE_MODE_DISABLED
> +      || (!deinterlace->interlaced && deinterlace->mode != GST_FFMPEGDEINTERLACE_MODE_INTERLACED));
> +  GST_DEBUG_OBJECT (deinterlace, "Passthrough: %d", deinterlace->passthrough);
>  }
>  
>  static gboolean
> @@ -120,6 +199,10 @@ gst_ffmpegdeinterlace_sink_setcaps (GstP
>    if (!gst_structure_get_int (structure, "height", &deinterlace->height))
>      return FALSE;
>  
> +  deinterlace->interlaced = FALSE;
> +  gst_structure_get_boolean (structure, "interlaced", &deinterlace->interlaced);
> +  gst_ffmpegdeinterlace_update_passthrough (deinterlace);
> +
>    ctx = avcodec_alloc_context ();
>    ctx->width = deinterlace->width;
>    ctx->height = deinterlace->height;
> @@ -138,10 +221,10 @@ gst_ffmpegdeinterlace_sink_setcaps (GstP
>        avpicture_get_size (deinterlace->pixfmt, deinterlace->width,
>        deinterlace->height);
>  
> -  src_caps = gst_caps_copy (caps);
> -  gst_caps_set_simple (src_caps, "interlaced", G_TYPE_BOOLEAN, FALSE, NULL);
> -  ret = gst_pad_set_caps (deinterlace->srcpad, src_caps);
> -  gst_caps_unref (src_caps);
> +  src_caps = gst_caps_copy (caps);
> +  gst_caps_set_simple (src_caps, "interlaced", G_TYPE_BOOLEAN, deinterlace->interlaced, NULL);
> +  ret = gst_pad_set_caps (deinterlace->srcpad, src_caps);
> +  gst_caps_unref (src_caps);
>  
>    return ret;
>  }
> @@ -162,6 +245,12 @@ gst_ffmpegdeinterlace_init (GstFFMpegDei
>    gst_element_add_pad (GST_ELEMENT (deinterlace), deinterlace->srcpad);
>  
>    deinterlace->pixfmt = PIX_FMT_NB;
> +
> +  deinterlace->interlaced = FALSE;
> +  deinterlace->passthrough = FALSE;
> +  deinterlace->reconfigure = FALSE;
> +  deinterlace->mode = DEFAULT_MODE;
> +  deinterlace->new_mode = -1;
>  }
>  
>  static GstFlowReturn
> @@ -172,6 +261,23 @@ gst_ffmpegdeinterlace_chain (GstPad * pa
>    GstBuffer *outbuf = NULL;
>    GstFlowReturn result;
>  
> +  GST_OBJECT_LOCK (deinterlace);
> +  if (deinterlace->reconfigure) {
> +    if (deinterlace->new_mode != -1)
> +      deinterlace->mode = deinterlace->new_mode;
> +    deinterlace->new_mode = -1;
> +
> +    deinterlace->reconfigure = FALSE;
> +    GST_OBJECT_UNLOCK (deinterlace);
> +    if (GST_PAD_CAPS (deinterlace->srcpad))
> +      gst_ffmpegdeinterlace_sink_setcaps (deinterlace->sinkpad, GST_PAD_CAPS (deinterlace->sinkpad));
> +  } else {
> +    GST_OBJECT_UNLOCK (deinterlace);
> +  }
> +  
> +  if (deinterlace->passthrough)
> +      return gst_pad_push (deinterlace->srcpad, inbuf);
> +
>    result =
>        gst_pad_alloc_buffer (deinterlace->srcpad, GST_BUFFER_OFFSET_NONE,
>        deinterlace->to_size, GST_PAD_CAPS (deinterlace->srcpad), &outbuf);
> @@ -202,3 +308,53 @@ gst_ffmpegdeinterlace_register (GstPlugi
>    return gst_element_register (plugin, "ffdeinterlace",
>        GST_RANK_NONE, GST_TYPE_FFMPEGDEINTERLACE);
>  }
> +
> +static void
> +gst_ffmpegdeinterlace_set_property (GObject * object, guint prop_id,
> +    const GValue * value, GParamSpec * pspec)
> +{
> +  GstFFMpegDeinterlace *self;
> +
> +  g_return_if_fail (GST_IS_FFMPEGDEINTERLACE (object));
> +  self = GST_FFMPEGDEINTERLACE (object);
> +
> +  switch (prop_id) {
> +    case PROP_MODE:{
> +      gint new_mode;
> +
> +      GST_OBJECT_LOCK (self);
> +      new_mode = g_value_get_enum (value);
> +      if (self->mode != new_mode && GST_PAD_CAPS (self->srcpad)) {
> +        self->reconfigure = TRUE;
> +        self->new_mode = new_mode;
> +      } else {
> +        self->mode = new_mode;
> +        gst_ffmpegdeinterlace_update_passthrough (self);
> +      }
> +      GST_OBJECT_UNLOCK (self);
> +      break;
> +    }
> +    default:
> +      G_OBJECT_WARN_INVALID_PROPERTY_ID (self, prop_id, pspec);
> +  }
> +
> +}
> +
> +static void
> +gst_ffmpegdeinterlace_get_property (GObject * object, guint prop_id,
> +    GValue * value, GParamSpec * pspec)
> +{
> +  GstFFMpegDeinterlace *self;
> +
> +  g_return_if_fail (GST_IS_FFMPEGDEINTERLACE (object));
> +  self = GST_FFMPEGDEINTERLACE (object);
> +
> +  switch (prop_id) {
> +    case PROP_MODE:
> +      g_value_set_enum (value, self->mode);
> +      break;
> +    default:
> +      G_OBJECT_WARN_INVALID_PROPERTY_ID (self, prop_id, pspec);
> +  }
> +}
> +

_______________________________________________
gstreamer-devel mailing list
[hidden email]
http://lists.freedesktop.org/mailman/listinfo/gstreamer-devel

[Tvrtko Ursulin]

Hi,

On Thursday 11 Aug 2011 10:30:57 Edward Hervey wrote:

> On Thu, 2011-08-11 at 10:02 +0100, Tvrtko Ursulin wrote:
> > On Wednesday 20 Jul 2011 12:02:00 Edward Hervey wrote:
> [...]
>
> > I see that multi-threaded decoding is disabled by default in
> > gstffmpegdec. Is this because of known issues and how serious are they?
>
>   The support for it was only dropped a month before release. The reason
> why I set the default back to only using one thread was:
>   * Lots of warnings from the libav side on some formats
>   * Reports of segfaulting in some situations
>   * Critical basic MT-safety patches landing the day I was going to do
> the release.

Any known issues with H.264 decoding?
 
>   That being said, the multi-threading support in gst-ffmpeg hasn't been
> removed/disabled. It's just that I've re-setted the default number of
> threads to 1 (instead of 0 which is automatic numbers).
>   So you can still force the number of threads you want manually.

Sure, perhaps when I wrote "disabled by default" it wasn't 100% clear - I
meant defaults to one thread.

>   Hopefully everything will be more stable soon and we'll be able to do
> a new gst-ffmpeg release with the default multi-threading enabled within
> a couple of months.

I'll keep an eye for that - thanks for the explanation!

> > With regards to ffdeinterlace - I did a pretty mechanical (not
> > understading GStreamer internals and code base that well) patch  porting
> > the mode property from good plugins deinterlace because having this
> > control and automatic mode is quite convenient for us.
> >
> > Would something like this be acceptable for merging upstream? At least in
> > principle, I realize patch might have issues. :)
>
>   Put that patch in a new bug report on bugzilla.gnome.org . It'll be
> handled there like all patches. I have no objections per-se.

Done: https://bugzilla.gnome.org/show_bug.cgi?id=656328

I have noticed there seems to be no development activity on gstreamer-devel
which had me puzzled. Can I ask where is it actually taking place? Because I
find just reading discussion about adding features very informative and
valuable to learning a certain code base.

Thanks,

Tvrtko
_______________________________________________
gstreamer-devel mailing list
[hidden email]
http://lists.freedesktop.org/mailman/listinfo/gstreamer-devel

[Stefan Sauer]

On 08/11/11 11:46, Tvrtko Ursulin wrote:

> Hi,
>
> On Thursday 11 Aug 2011 10:30:57 Edward Hervey wrote:
>> On Thu, 2011-08-11 at 10:02 +0100, Tvrtko Ursulin wrote:
>>> On Wednesday 20 Jul 2011 12:02:00 Edward Hervey wrote:
>> [...]
>>
>>> I see that multi-threaded decoding is disabled by default in
>>> gstffmpegdec. Is this because of known issues and how serious are they?
>>   The support for it was only dropped a month before release. The reason
>> why I set the default back to only using one thread was:
>>   * Lots of warnings from the libav side on some formats
>>   * Reports of segfaulting in some situations
>>   * Critical basic MT-safety patches landing the day I was going to do
>> the release.
> Any known issues with H.264 decoding?
>  
>>   That being said, the multi-threading support in gst-ffmpeg hasn't been
>> removed/disabled. It's just that I've re-setted the default number of
>> threads to 1 (instead of 0 which is automatic numbers).
>>   So you can still force the number of threads you want manually.
> Sure, perhaps when I wrote "disabled by default" it wasn't 100% clear - I
> meant defaults to one thread.
>
>>   Hopefully everything will be more stable soon and we'll be able to do
>> a new gst-ffmpeg release with the default multi-threading enabled within
>> a couple of months.
> I'll keep an eye for that - thanks for the explanation!
>
>>> With regards to ffdeinterlace - I did a pretty mechanical (not
>>> understading GStreamer internals and code base that well) patch  porting
>>> the mode property from good plugins deinterlace because having this
>>> control and automatic mode is quite convenient for us.
>>>
>>> Would something like this be acceptable for merging upstream? At least in
>>> principle, I realize patch might have issues. :)
>>   Put that patch in a new bug report on bugzilla.gnome.org . It'll be
>> handled there like all patches. I have no objections per-se.
> Done: https://bugzilla.gnome.org/show_bug.cgi?id=656328
>
> I have noticed there seems to be no development activity on gstreamer-devel
> which had me puzzled. Can I ask where is it actually taking place? Because I
> find just reading discussion about adding features very informative and
> valuable to learning a certain code base.

Most discussion is on irc (irc.freenode.net/#gstreamer)

Stefan

> Thanks,
>
> Tvrtko
> _______________________________________________
> gstreamer-devel mailing list
> [hidden email]
> http://lists.freedesktop.org/mailman/listinfo/gstreamer-devel

_______________________________________________
gstreamer-devel mailing list
[hidden email]
http://lists.freedesktop.org/mailman/listinfo/gstreamer-devel

[Tim-Philipp Müller-2]

On Fri, 2011-08-12 at 15:22 +0200, Stefan Kost wrote:

> Most discussion is on irc (irc.freenode.net/#gstreamer)

And in bugzilla...

Cheers
 -Tim


_______________________________________________
gstreamer-devel mailing list
[hidden email]
http://lists.freedesktop.org/mailman/listinfo/gstreamer-devel