[Bluez-devel] Patch: inheritance problem in gsta2dpsink

[Bluez-devel] Patch: inheritance problem in gsta2dpsink

[thiagoss]

[-- Attachment #1.1: Type: text/plain, Size: 405 bytes --]

GstA2dpSink was extending GstAudioSink, but there is a problem in that,
GstAudioSink expects only
"audio/x-raw-int", "audio/x-raw-float", "audio/x-mulaw" or "audio/x-alaw" as
media types. GstA2dpSink should support mp3 and sbc, that's causing a
runtime error. This patch changes the base class from GstAudioSink to
GstBaseSink, changing the stub functions to be implemented also.


Best regards,


Thiago

[-- Attachment #1.2: Type: text/html, Size: 476 bytes --]

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #2: gsta2dpsink.patch --]
[-- Type: text/x-patch; name="gsta2dpsink.patch", Size: 4259 bytes --]

Index: audio/gsta2dpsink.c
===================================================================
RCS file: /cvsroot/bluez/utils/audio/gsta2dpsink.c,v
retrieving revision 1.3
diff -u -r1.3 gsta2dpsink.c
--- audio/gsta2dpsink.c	26 Aug 2007 14:14:34 -0000	1.3
+++ audio/gsta2dpsink.c	1 Oct 2007 21:07:19 -0000
@@ -43,7 +43,7 @@
 	PROP_DEVICE,
 };
 
-GST_BOILERPLATE(GstA2dpSink, gst_a2dp_sink, GstAudioSink, GST_TYPE_AUDIO_SINK);
+GST_BOILERPLATE(GstA2dpSink, gst_a2dp_sink, GstBaseSink, GST_TYPE_BASE_SINK);
 
 static const GstElementDetails a2dp_sink_details =
 	GST_ELEMENT_DETAILS("Bluetooth A2DP sink",
@@ -85,6 +85,8 @@
 
 	g_free(sink->device);
 
+	g_io_channel_unref(sink->server);
+
 	G_OBJECT_CLASS(parent_class)->finalize(object);
 }
 
@@ -124,56 +126,26 @@
 	}
 }
 
-static gboolean gst_a2dp_sink_open(GstAudioSink *self)
-{
-	GstA2dpSink *sink = GST_A2DP_SINK(self);
-
-	printf("device %s\n", sink->device);
-	printf("open\n");
-
-	return TRUE;
-}
-
-static gboolean gst_a2dp_sink_prepare(GstAudioSink *self,
-						GstRingBufferSpec *spec)
+static gboolean gst_a2dp_sink_start(GstBaseSink *basesink)
 {
-	printf("perpare\n");
-	printf("rate %d\n", spec->rate);
-	printf("channels %d\n", spec->channels);
+	g_print("start\n");
 
 	return TRUE;
 }
 
-static gboolean gst_a2dp_sink_unprepare(GstAudioSink *self)
+static gboolean gst_a2dp_sink_stop(GstBaseSink *basesink)
 {
-	printf("unprepare\n");
+	g_print("stop\n");
 
 	return TRUE;
 }
 
-static gboolean gst_a2dp_sink_close(GstAudioSink *self)
+static GstFlowReturn gst_a2dp_sink_render(GstBaseSink *basesink, 
+					GstBuffer *buffer)
 {
-	printf("close\n");
+	g_print("render\n");
 
-	return TRUE;
-}
-
-static guint gst_a2dp_sink_write(GstAudioSink *self,
-					gpointer data, guint length)
-{
-	return 0;
-}
-
-static guint gst_a2dp_sink_delay(GstAudioSink *audiosink)
-{
-	printf("delay\n");
-
-	return 0;
-}
-
-static void gst_a2dp_sink_reset(GstAudioSink *audiosink)
-{
-	printf("reset\n");
+	return GST_FLOW_OK;
 }
 
 static gboolean server_callback(GIOChannel *chan,
@@ -187,7 +159,7 @@
 static void gst_a2dp_sink_class_init(GstA2dpSinkClass *klass)
 {
 	GObjectClass *object_class = G_OBJECT_CLASS(klass);
-	GstAudioSinkClass *audiosink_class = GST_AUDIO_SINK_CLASS(klass);
+	GstBaseSinkClass *basesink_class = GST_BASE_SINK_CLASS(klass);
 
 	parent_class = g_type_class_peek_parent(klass);
 
@@ -195,13 +167,9 @@
 	object_class->set_property = GST_DEBUG_FUNCPTR(gst_a2dp_sink_set_property);
 	object_class->get_property = GST_DEBUG_FUNCPTR(gst_a2dp_sink_get_property);
 
-	audiosink_class->open = GST_DEBUG_FUNCPTR(gst_a2dp_sink_open);
-	audiosink_class->prepare = GST_DEBUG_FUNCPTR(gst_a2dp_sink_prepare);
-	audiosink_class->unprepare = GST_DEBUG_FUNCPTR(gst_a2dp_sink_unprepare);
-	audiosink_class->close = GST_DEBUG_FUNCPTR(gst_a2dp_sink_close);
-	audiosink_class->write = GST_DEBUG_FUNCPTR(gst_a2dp_sink_write);
-	audiosink_class->delay = GST_DEBUG_FUNCPTR(gst_a2dp_sink_delay);
-	audiosink_class->reset = GST_DEBUG_FUNCPTR(gst_a2dp_sink_reset);
+	basesink_class->start = GST_DEBUG_FUNCPTR(gst_a2dp_sink_start);
+	basesink_class->stop = GST_DEBUG_FUNCPTR(gst_a2dp_sink_stop);
+	basesink_class->render = GST_DEBUG_FUNCPTR(gst_a2dp_sink_render);
 
 	g_object_class_install_property(object_class, PROP_DEVICE,
 				g_param_spec_string("device", "Device",
@@ -232,6 +200,4 @@
 
 	g_io_add_watch(self->server, G_IO_IN | G_IO_HUP | G_IO_ERR | G_IO_NVAL,
 							server_callback, self);
-
-	g_io_channel_unref(self->server);
 }
Index: audio/gsta2dpsink.h
===================================================================
RCS file: /cvsroot/bluez/utils/audio/gsta2dpsink.h,v
retrieving revision 1.3
diff -u -r1.3 gsta2dpsink.h
--- audio/gsta2dpsink.h	26 Aug 2007 14:14:34 -0000	1.3
+++ audio/gsta2dpsink.h	1 Oct 2007 21:07:19 -0000
@@ -22,7 +22,7 @@
  */
 
 #include <gst/gst.h>
-#include <gst/audio/gstaudiosink.h>
+#include <gst/base/gstbasesink.h>
 
 G_BEGIN_DECLS
 
@@ -41,7 +41,7 @@
 typedef struct _GstA2dpSinkClass GstA2dpSinkClass;
 
 struct _GstA2dpSink {
-	GstAudioSink sink;
+	GstBaseSink sink;
 
 	gchar *device;
 
@@ -49,7 +49,7 @@
 };
 
 struct _GstA2dpSinkClass {
-	GstAudioSinkClass parent_class;
+	GstBaseSinkClass parent_class;
 };
 
 GType gst_a2dp_sink_get_type(void);

[-- Attachment #3: Type: text/plain, Size: 228 bytes --]

-------------------------------------------------------------------------
This SF.net email is sponsored by: Microsoft
Defy all challenges. Microsoft(R) Visual Studio 2005.
http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/

[-- Attachment #4: Type: text/plain, Size: 164 bytes --]

_______________________________________________
Bluez-devel mailing list
Bluez-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/bluez-devel

Re: [Bluez-devel] Patch: inheritance problem in gsta2dpsink

[thiagoss]

[-- Attachment #1.1: Type: text/plain, Size: 828 bytes --]

This patch adds the prerolling to the gsta2dpsink gstreamer plugin, it sends
the configuration packet to the device and prepares the connection.

It is currently only supporting sbc, needs to be extended to mp3.

This patch includes the full set of changes the previous one did.

I'll be waiting comments, reviews or sugestions.

BR,

Thiago

On 10/1/07, thiagoss <thiagossantos@gmail.com> wrote:
>
> GstA2dpSink was extending GstAudioSink, but there is a problem in that,
> GstAudioSink expects only
> "audio/x-raw-int", "audio/x-raw-float", "audio/x-mulaw" or "audio/x-alaw"
> as media types. GstA2dpSink should support mp3 and sbc, that's causing a
> runtime error. This patch changes the base class from GstAudioSink to
> GstBaseSink, changing the stub functions to be implemented also.
>
>
> Best regards,
>
>
> Thiago
>
>

[-- Attachment #1.2: Type: text/html, Size: 1231 bytes --]

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #2: gsta2dpsink_preroll.patch --]
[-- Type: text/x-patch; name="gsta2dpsink_preroll.patch", Size: 21202 bytes --]

Index: audio/gsta2dpsink.c
===================================================================
RCS file: /cvsroot/bluez/utils/audio/gsta2dpsink.c,v
retrieving revision 1.3
diff -u -r1.3 gsta2dpsink.c
--- audio/gsta2dpsink.c	26 Aug 2007 14:14:34 -0000	1.3
+++ audio/gsta2dpsink.c	5 Oct 2007 18:49:39 -0000
@@ -28,8 +28,11 @@
 #include <unistd.h>
 #include <sys/un.h>
 #include <sys/socket.h>
+#include <fcntl.h>
+#include <pthread.h>
 
 #include "ipc.h"
+#include "sbc.h"
 
 #include "gsta2dpsink.h"
 
@@ -37,13 +40,117 @@
 #define GST_CAT_DEFAULT a2dp_sink_debug
 
 #define DEFAULT_DEVICE "default"
+#define BUFFER_SIZE 2048
+
+#define GST_A2DP_SINK_MUTEX_LOCK(s) G_STMT_START {	\
+		g_mutex_lock (s->sink_lock);		\
+	} G_STMT_END
+
+#define GST_A2DP_SINK_MUTEX_UNLOCK(s) G_STMT_START {	\
+		g_mutex_unlock (s->sink_lock);		\
+	} G_STMT_END
+
+#define GST_A2DP_SINK_WAIT_CON_END(s) G_STMT_START {			\
+		s->waiting_con_conf = TRUE;				\
+		g_cond_wait (s->con_conf_end, s->sink_lock);		\
+		s->waiting_con_conf = FALSE;				\
+	} G_STMT_END
+
+#define GST_A2DP_SINK_CONFIGURATION_FAIL(s) G_STMT_START {		\
+		s->con_state = NOT_CONFIGURED;				\
+		g_cond_signal (s->con_conf_end);			\
+	} G_STMT_END
+
+#define GST_A2DP_SINK_CONFIGURATION_SUCCESS(s) G_STMT_START {		\
+		s->con_state = CONFIGURED;				\
+		g_cond_signal (s->con_conf_end);			\
+	} G_STMT_END
+
+struct bluetooth_a2dp {
+	sbc_t sbc;			/* Codec data */
+	int codesize;			/* SBC codesize */
+	int samples;			/* Number of encoded samples */
+	uint8_t buffer[BUFFER_SIZE];	/* Codec transfer buffer */
+	int count;			/* Codec transfer buffer counter */
+
+	int nsamples;			/* Cumulative number of codec samples */
+	uint16_t seq_num;		/* Cumulative packet sequence */
+	int frame_count;		/* Current frames in buffer*/
+};
+struct bluetooth_data {
+	struct ipc_data_cfg cfg;	/* Bluetooth device config */
+	uint8_t buffer[BUFFER_SIZE];	/* Encoded transfer buffer */
+	int count;			/* Transfer buffer counter */
+	struct bluetooth_a2dp a2dp;	/* A2DP data */
+
+	pthread_t hw_thread;		/* Makes virtual hw pointer move */
+	int pipefd[2];			/* Inter thread communication */
+	int stopped;
+};
+
+#if __BYTE_ORDER == __LITTLE_ENDIAN
+
+struct rtp_header {
+	uint8_t cc:4;
+	uint8_t x:1;
+	uint8_t p:1;
+	uint8_t v:2;
+
+	uint8_t pt:7;
+	uint8_t m:1;
+
+	uint16_t sequence_number;
+	uint32_t timestamp;
+	uint32_t ssrc;
+	uint32_t csrc[0];
+} __attribute__ ((packed));
+
+struct rtp_payload {
+	uint8_t frame_count:4;
+	uint8_t rfa0:1;
+	uint8_t is_last_fragment:1;
+	uint8_t is_first_fragment:1;
+	uint8_t is_fragmented:1;
+} __attribute__ ((packed));
+
+#elif __BYTE_ORDER == __BIG_ENDIAN
+
+struct rtp_header {
+	uint8_t v:2;
+	uint8_t p:1;
+	uint8_t x:1;
+	uint8_t cc:4;
+
+	uint8_t m:1;
+	uint8_t pt:7;
+
+	uint16_t sequence_number;
+	uint32_t timestamp;
+	uint32_t ssrc;
+	uint32_t csrc[0];
+} __attribute__ ((packed));
+
+struct rtp_payload {
+	uint8_t is_fragmented:1;
+	uint8_t is_first_fragment:1;
+	uint8_t is_last_fragment:1;
+	uint8_t rfa0:1;
+	uint8_t frame_count:4;
+} __attribute__ ((packed));
+
+#else
+#error "Unknown byte order"
+#endif
+
+#define IS_SBC(n) (strcmp(n, "audio/x-sbc") == 0)
+#define IS_MPEG(n) (strcmp(n, "audio/mpeg") == 0)
 
 enum {
 	PROP_0,
 	PROP_DEVICE,
 };
 
-GST_BOILERPLATE(GstA2dpSink, gst_a2dp_sink, GstAudioSink, GST_TYPE_AUDIO_SINK);
+GST_BOILERPLATE(GstA2dpSink, gst_a2dp_sink, GstBaseSink, GST_TYPE_BASE_SINK);
 
 static const GstElementDetails a2dp_sink_details =
 	GST_ELEMENT_DETAILS("Bluetooth A2DP sink",
@@ -77,14 +184,45 @@
 	gst_element_class_set_details(element_class, &a2dp_sink_details);
 }
 
+static void gst_a2dp_sink_bluetooth_exit(GstA2dpSink *sink)
+{
+	struct bluetooth_a2dp *a2dp = &sink->data->a2dp;
+
+	if (sink->stream_fd >= 0) {
+		close(sink->stream_fd);
+		sink->stream_fd = -1;
+	}
+
+	if (sink->data->hw_thread) {
+		pthread_cancel(sink->data->hw_thread);
+		pthread_join(sink->data->hw_thread, 0);
+	}
+
+	if (sink->data->cfg.codec == CFG_CODEC_SBC)
+		sbc_finish(&a2dp->sbc);
+
+	if (sink->data->pipefd[0] > 0)
+		close(sink->data->pipefd[0]);
+
+	if (sink->data->pipefd[1] > 0)
+		close(sink->data->pipefd[1]);
+
+	g_free(sink->data);
+}
+
 static void gst_a2dp_sink_finalize(GObject *object)
 {
 	GstA2dpSink *sink = GST_A2DP_SINK(object);
 
 	g_io_channel_close(sink->server);
+	g_io_channel_unref(sink->server);
+	gst_a2dp_sink_bluetooth_exit(sink);
 
 	g_free(sink->device);
 
+	g_cond_free(sink->con_conf_end);
+	g_mutex_free(sink->sink_lock);
+
 	G_OBJECT_CLASS(parent_class)->finalize(object);
 }
 
@@ -124,62 +262,447 @@
 	}
 }
 
-static gboolean gst_a2dp_sink_open(GstAudioSink *self)
+static gint gst_a2dp_sink_bluetooth_recvmsg_fd(GstA2dpSink *sink)
+{
+	char cmsg_b[CMSG_SPACE(sizeof(int))], m;
+	int err, ret;
+	struct iovec iov = { &m, sizeof(m) };
+	struct msghdr msgh;
+	struct cmsghdr *cmsg;
+
+	memset(&msgh, 0, sizeof(msgh));
+	msgh.msg_iov = &iov;
+	msgh.msg_iovlen = 1;
+	msgh.msg_control = &cmsg_b;
+	msgh.msg_controllen = CMSG_LEN(sizeof(int));
+
+	ret = recvmsg(g_io_channel_unix_get_fd(sink->server), &msgh, 0);
+	if (ret < 0) {
+		err = errno;
+		GST_ERROR_OBJECT(sink, "Unable to receive fd: %s (%d)", 
+				strerror(err), err);
+		return -err;
+	}
+
+	/* Receive auxiliary data in msgh */
+	for (cmsg = CMSG_FIRSTHDR(&msgh); cmsg != NULL;
+			cmsg = CMSG_NXTHDR(&msgh, cmsg)) {
+		if (cmsg->cmsg_level == SOL_SOCKET
+				&& cmsg->cmsg_type == SCM_RIGHTS) {
+			sink->stream_fd = (*(int *) CMSG_DATA(cmsg));
+			GST_DEBUG_OBJECT(sink, "stream_fd=%d", 
+				sink->stream_fd);
+			return 0;
+		}
+	}
+
+	return -EINVAL;
+}
+
+static int gst_a2dp_sink_bluetooth_a2dp_init(GstA2dpSink *sink, 
+			struct ipc_codec_sbc *sbc)
 {
-	GstA2dpSink *sink = GST_A2DP_SINK(self);
+	struct bluetooth_a2dp *a2dp = &sink->data->a2dp;
+	struct ipc_data_cfg *cfg = &sink->data->cfg;
+
+	if (cfg == NULL) {
+		GST_ERROR_OBJECT(sink, "Error getting codec parameters");
+		return -1;
+	}
+
+	if (cfg->codec != CFG_CODEC_SBC)
+		return -1;
 
-	printf("device %s\n", sink->device);
-	printf("open\n");
+	/* FIXME: init using flags? */
+	sbc_init(&a2dp->sbc, 0);
+	a2dp->sbc.rate = cfg->rate;
+	a2dp->sbc.channels = cfg->mode == CFG_MODE_MONO ? 1 : 2;
+	if (cfg->mode == CFG_MODE_MONO || cfg->mode == CFG_MODE_JOINT_STEREO)
+		a2dp->sbc.joint = 1;
+	a2dp->sbc.allocation = sbc->allocation;
+	a2dp->sbc.subbands = sbc->subbands;
+	a2dp->sbc.blocks = sbc->blocks;
+	a2dp->sbc.bitpool = sbc->bitpool;
+	a2dp->codesize = a2dp->sbc.subbands * a2dp->sbc.blocks *
+						a2dp->sbc.channels * 2;
+	a2dp->count = sizeof(struct rtp_header) + sizeof(struct rtp_payload);
+
+	GST_DEBUG_OBJECT(sink, "Codec parameters: \
+				\tallocation=%u\n\tsubbands=%u\n \
+				\tblocks=%u\n\tbitpool=%u\n",
+		a2dp->sbc.allocation, a2dp->sbc.subbands, a2dp->sbc.blocks,
+		a2dp->sbc.bitpool);
+
+	return 0;
+}
+
+static gboolean gst_a2dp_sink_init_pkt_conf(GstA2dpSink *sink, 
+		GstCaps *caps, struct ipc_packet *pkt)
+{
+
+	struct ipc_data_cfg *cfg = (void *) pkt->data;
+	struct ipc_codec_sbc *sbc = (void *) cfg->data;
+	const GValue *value = NULL;
+	const char *pref, *name;
+	GstStructure *structure = gst_caps_get_structure(caps,0);
+
+	name = gst_structure_get_name(structure);
+	/* FIXME only sbc supported here, should suport mp3 */
+	if (!(IS_SBC(name))) {
+		GST_ERROR_OBJECT(sink, "Unsupported format %s", name);
+		return FALSE;
+	}
+
+	strncpy(pkt->device, sink->device, 18);
+	pkt->role = PKT_ROLE_HIFI;
+
+	value = gst_structure_get_value(structure, "rate");
+	cfg->rate = g_value_get_int(value);
+
+	value = gst_structure_get_value(structure, "mode");
+	pref = g_value_get_string(value);
+	if (strcmp(pref, "auto") == 0)
+		cfg->mode = CFG_MODE_AUTO;
+	else if (strcmp(pref, "mono") == 0)
+		cfg->mode = CFG_MODE_MONO;
+	else if (strcmp(pref, "dual") == 0)
+		cfg->mode = CFG_MODE_DUAL_CHANNEL;
+	else if (strcmp(pref, "stereo") == 0)
+		cfg->mode = CFG_MODE_STEREO;
+	else if (strcmp(pref, "joint") == 0)
+		cfg->mode = CFG_MODE_JOINT_STEREO;
+	else {
+		GST_ERROR_OBJECT(sink, "Invalid mode %s", pref);
+		return FALSE;
+	}
+
+	value = gst_structure_get_value(structure, "allocation");
+	pref = g_value_get_string(value);
+	if (strcmp(pref, "auto") == 0)
+		sbc->allocation = CFG_ALLOCATION_AUTO;
+	else if (strcmp(pref, "loudness") == 0)
+		sbc->allocation = CFG_ALLOCATION_LOUDNESS;
+	else if (strcmp(pref, "snr") == 0)
+		sbc->allocation = CFG_ALLOCATION_SNR;
+	else {
+		GST_ERROR_OBJECT(sink, "Invalid allocation: %s", pref);
+		return FALSE;
+	}
+
+	value = gst_structure_get_value(structure, "subbands");
+	sbc->subbands = g_value_get_int(value);
+
+	value = gst_structure_get_value(structure, "blocks");
+	sbc->blocks = g_value_get_int(value);
+/* FIXME how can I obtain the bitpool ?
+		if (strcmp(id, "bitpool") == 0) {
+			if (snd_config_get_string(n, &bitpool) < 0) {
+				SNDERR("Invalid type for %s", id);
+				return -EINVAL;
+			}
+
+			sbc->bitpool = atoi(bitpool);
+			continue;
+		}
+
+		SNDERR("Unknown field %s", id);
+		return -EINVAL;
+	}
+*/
+	sbc->bitpool = 32;
+
+	pkt->length = sizeof(*cfg) + sizeof(*sbc);
+	pkt->type = PKT_TYPE_CFG_REQ;
+	pkt->error = PKT_ERROR_NONE;
 
 	return TRUE;
+
 }
 
-static gboolean gst_a2dp_sink_prepare(GstAudioSink *self,
-						GstRingBufferSpec *spec)
+static gboolean gst_a2dp_sink_start(GstBaseSink *basesink)
 {
-	printf("perpare\n");
-	printf("rate %d\n", spec->rate);
-	printf("channels %d\n", spec->channels);
+	g_print("start\n");
 
 	return TRUE;
 }
 
-static gboolean gst_a2dp_sink_unprepare(GstAudioSink *self)
+static gboolean gst_a2dp_sink_stop(GstBaseSink *basesink)
 {
-	printf("unprepare\n");
+	g_print("stop\n");
 
 	return TRUE;
 }
 
-static gboolean gst_a2dp_sink_close(GstAudioSink *self)
+static gboolean gst_a2dp_sink_send_conf_pkt(GstA2dpSink *sink, GstCaps *caps)
 {
-	printf("close\n");
+	gchar buf[IPC_MTU];
+	struct ipc_packet *pkt = (void *) buf;
+	gboolean ret;
+	GIOError io_error;
+
+	g_assert(sink->con_state == NOT_CONFIGURED);
+
+	pkt = (void*) buf;
+	ret = gst_a2dp_sink_init_pkt_conf(sink, caps, pkt);
+	if (!ret) {
+		GST_ERROR_OBJECT(sink, "Couldn't initialize parse caps \
+				to packet configuration");
+		return FALSE;
+	}
+
+	sink->con_state = CONFIGURING_INIT;
+
+	io_error = g_io_channel_write(sink->server, (gchar*)pkt, 
+			sizeof(*pkt) + pkt->length, &ret);
+	if (io_error != G_IO_ERROR_NONE) {
+		GST_ERROR_OBJECT(sink, "Error ocurred while sending \
+					configurarion packet");
+		sink->con_state = NOT_CONFIGURED;
+		return FALSE;
+	}
+
+	GST_DEBUG_OBJECT(sink, "%d bytes sent", ret);
+	sink->con_state = CONFIGURING_SENT_CONF;
 
 	return TRUE;
 }
 
-static guint gst_a2dp_sink_write(GstAudioSink *self,
-					gpointer data, guint length)
+static gboolean gst_a2dp_sink_start_dev_conf(GstA2dpSink *sink, GstCaps *caps)
 {
-	return 0;
+	gboolean ret;
+
+	g_assert(sink->con_state == NOT_CONFIGURED);
+
+	GST_DEBUG_OBJECT(sink, "starting device configuration");
+
+	ret = gst_a2dp_sink_send_conf_pkt(sink, caps);
+
+	return ret;
 }
 
-static guint gst_a2dp_sink_delay(GstAudioSink *audiosink)
+static GstFlowReturn gst_a2dp_sink_preroll(GstBaseSink *basesink, 
+					GstBuffer *buffer)
 {
-	printf("delay\n");
+	GstA2dpSink *sink;
+	sink = GST_A2DP_SINK(basesink);
 
-	return 0;
+	GST_A2DP_SINK_MUTEX_LOCK(sink);
+
+	if (sink->con_state == NOT_CONFIGURED)
+		gst_a2dp_sink_start_dev_conf(sink, GST_BUFFER_CAPS(buffer));
+
+	/* wait for the connection process to finish */
+	if (sink->con_state != CONFIGURED)
+		GST_A2DP_SINK_WAIT_CON_END(sink);
+
+	GST_A2DP_SINK_MUTEX_UNLOCK(sink);
+
+	if (sink->con_state != CONFIGURED)
+		return GST_FLOW_ERROR;
+
+	return GST_FLOW_OK;
 }
 
-static void gst_a2dp_sink_reset(GstAudioSink *audiosink)
+static GstFlowReturn gst_a2dp_sink_render(GstBaseSink *basesink,
+					GstBuffer *buffer)
 {
-	printf("reset\n");
+	g_print("render\n");
+
+	return GST_FLOW_OK;
+}
+
+static gboolean gst_a2dp_sink_conf_resp(GstA2dpSink *sink)
+{
+	gchar buf[IPC_MTU];
+	GIOError io_error;
+	guint ret;
+	struct ipc_packet *pkt = (void *) buf;
+	struct ipc_data_cfg *cfg = (void *) pkt->data;
+
+	memset(buf, 0, sizeof(buf));
+
+	io_error = g_io_channel_read(sink->server, (gchar*)buf, 
+			sizeof(*pkt) + sizeof(*cfg), &ret);
+	if (io_error != G_IO_ERROR_NONE && ret > 0) {
+		GST_ERROR_OBJECT(sink, "Error ocurred while receiving \
+					configurarion packet answer");
+		return FALSE;
+	}
+
+	sink->total = ret;
+	if (pkt->type != PKT_TYPE_CFG_RSP) {
+		GST_ERROR_OBJECT(sink, "Unexpected packet type %d \
+					received", pkt->type);
+		return FALSE;
+	}
+
+	if (pkt->error != PKT_ERROR_NONE) {
+		GST_ERROR_OBJECT(sink, "Error %d while configuring \
+					device", pkt->error);
+		return FALSE;
+	}
+
+	if (cfg->codec != CFG_CODEC_SBC) {
+		GST_ERROR_OBJECT(sink, "Unsupported format");
+		return FALSE;
+	}
+
+	return TRUE;
+}
+
+static gboolean gst_a2dp_sink_conf_recv_dev_conf(GstA2dpSink *sink)
+{
+	gchar buf[IPC_MTU];
+	GIOError io_error;
+	guint ret=0;
+	struct ipc_packet *pkt = (void *) buf;
+	struct ipc_data_cfg *cfg = (void *) pkt->data;
+	struct ipc_codec_sbc *sbc = (void *) cfg->data;
+
+	io_error = g_io_channel_read(sink->server, (gchar*)sbc, 
+			sizeof(*sbc), &ret);
+	if (io_error != G_IO_ERROR_NONE) {
+		GST_ERROR_OBJECT(sink, "Error while reading data from socket \
+				%s (%d)", strerror(errno), errno); 
+		return FALSE;
+	} else if (ret == 0) {
+		GST_ERROR_OBJECT(sink, "Read 0 bytes from socket"); 
+		return FALSE;
+	}
+
+	sink->total += ret;
+	GST_DEBUG_OBJECT(sink, "OK - %d bytes received", sink->total);
+
+	if (pkt->length != (sink->total - sizeof(struct ipc_packet))) {
+		GST_ERROR_OBJECT(sink, "Error while configuring device: \
+				packet size doesn't match");
+		return FALSE;
+	}
+
+	memcpy(&sink->data->cfg, cfg, sizeof(*cfg));
+
+	GST_DEBUG_OBJECT(sink, "Device configuration:\n\tfd=%d\n\tfd_opt=%u\n\
+			\tpkt_len=%u\n\tsample_size=%u\n\trate=%u",
+			sink->stream_fd, sink->data->cfg.fd_opt, \
+			sink->data->cfg.pkt_len, sink->data->cfg.sample_size, \
+			sink->data->cfg.rate);
+
+	if (sink->data->cfg.codec == CFG_CODEC_SBC) {
+		ret = gst_a2dp_sink_bluetooth_a2dp_init(sink, sbc);
+		if (ret < 0)
+			return FALSE;
+
+	}
+	return TRUE;
+}
+
+static gboolean gst_a2dp_sink_conf_recv_stream_fd(GstA2dpSink *sink)
+{
+	guint ret=0;
+	ret = gst_a2dp_sink_bluetooth_recvmsg_fd(sink);
+	if (ret < 0)
+		return FALSE;
+
+	if (sink->stream_fd == -1) {
+		GST_ERROR_OBJECT(sink, "Error while configuring device: \
+				could not acquire audio socket");
+		return FALSE;
+	}
+
+	/* It is possible there is some outstanding
+	data in the pipe - we have to empty it */
+	while (recv(sink->stream_fd, sink->data->buffer, 
+				sink->data->cfg.pkt_len, MSG_DONTWAIT) > 0);
+
+	memset(sink->data->buffer, 0, sizeof(sink->data->buffer));
+
+	return TRUE;
+}
+
+static void gst_a2dp_sink_conf_recv_data(GstA2dpSink *sink)
+{
+	/* 
+	 * We hold the lock, since we can send a signal.
+	 * It is a good practice, according to the glib api.
+	 */
+	GST_A2DP_SINK_MUTEX_LOCK(sink);
+
+	switch (sink->con_state) {
+		case CONFIGURING_SENT_CONF:
+			if (gst_a2dp_sink_conf_resp(sink)) {
+				sink->con_state = CONFIGURING_RCVD_CONF_RSP;
+			} else {
+				GST_A2DP_SINK_CONFIGURATION_FAIL(sink);
+			}
+		break;
+		case CONFIGURING_RCVD_CONF_RSP:
+			if (gst_a2dp_sink_conf_recv_dev_conf(sink)) {
+				sink->con_state = CONFIGURING_RCVD_DEV_CONF;
+			} else {
+				GST_A2DP_SINK_CONFIGURATION_FAIL(sink);
+			}
+		case CONFIGURING_RCVD_DEV_CONF:
+			if (gst_a2dp_sink_conf_recv_stream_fd(sink)) {
+				GST_A2DP_SINK_CONFIGURATION_SUCCESS(sink);
+			} else {
+				GST_A2DP_SINK_CONFIGURATION_FAIL(sink);
+			}
+		break;
+	}
+
+	GST_A2DP_SINK_MUTEX_UNLOCK(sink);
+
+}
+
+static gboolean gst_a2dp_sink_set_caps(GstBaseSink *self, GstCaps *caps)
+{
+	GstA2dpSink *sink;
+
+	sink = GST_A2DP_SINK(self);
+
+	GST_A2DP_SINK_MUTEX_LOCK(sink);
+	if (sink->con_state == NOT_CONFIGURED) {
+		gst_a2dp_sink_start_dev_conf(sink, caps);
+	}
+	GST_A2DP_SINK_MUTEX_UNLOCK(sink);
+
+	return TRUE;
+}
+
+static gboolean gst_a2dp_sink_unlock(GstBaseSink *self)
+{
+	g_print("unlock\n");
+
+	return FALSE;
 }
 
 static gboolean server_callback(GIOChannel *chan,
 					GIOCondition cond, gpointer data)
 {
-	printf("callback\n");
+	GstA2dpSink *sink = GST_A2DP_SINK(data);
+
+	switch (cond) {
+		case G_IO_IN:
+			if (sink->con_state != NOT_CONFIGURED && 
+					sink->con_state != CONFIGURED)
+				gst_a2dp_sink_conf_recv_data(sink);
+			else
+				GST_WARNING_OBJECT(sink, "Unexpected data received");
+			break;
+		case G_IO_HUP:
+			g_print("callback G_IO_HUP\n");
+			break;
+		case G_IO_ERR:
+			g_print("callback G_IO_ERR\n");
+			break;
+		case G_IO_NVAL:
+			g_print("callback G_IO_NVAL\n");
+			break;
+		default:
+			g_print("unexpected callback\n");
+			break;
+
+	}
 
 	return TRUE;
 }
@@ -187,21 +710,22 @@
 static void gst_a2dp_sink_class_init(GstA2dpSinkClass *klass)
 {
 	GObjectClass *object_class = G_OBJECT_CLASS(klass);
-	GstAudioSinkClass *audiosink_class = GST_AUDIO_SINK_CLASS(klass);
+	GstBaseSinkClass *basesink_class = GST_BASE_SINK_CLASS(klass);
 
 	parent_class = g_type_class_peek_parent(klass);
 
 	object_class->finalize = GST_DEBUG_FUNCPTR(gst_a2dp_sink_finalize);
-	object_class->set_property = GST_DEBUG_FUNCPTR(gst_a2dp_sink_set_property);
-	object_class->get_property = GST_DEBUG_FUNCPTR(gst_a2dp_sink_get_property);
-
-	audiosink_class->open = GST_DEBUG_FUNCPTR(gst_a2dp_sink_open);
-	audiosink_class->prepare = GST_DEBUG_FUNCPTR(gst_a2dp_sink_prepare);
-	audiosink_class->unprepare = GST_DEBUG_FUNCPTR(gst_a2dp_sink_unprepare);
-	audiosink_class->close = GST_DEBUG_FUNCPTR(gst_a2dp_sink_close);
-	audiosink_class->write = GST_DEBUG_FUNCPTR(gst_a2dp_sink_write);
-	audiosink_class->delay = GST_DEBUG_FUNCPTR(gst_a2dp_sink_delay);
-	audiosink_class->reset = GST_DEBUG_FUNCPTR(gst_a2dp_sink_reset);
+	object_class->set_property = GST_DEBUG_FUNCPTR(
+					gst_a2dp_sink_set_property);
+	object_class->get_property = GST_DEBUG_FUNCPTR(
+					gst_a2dp_sink_get_property);
+
+	basesink_class->start = GST_DEBUG_FUNCPTR(gst_a2dp_sink_start);
+	basesink_class->stop = GST_DEBUG_FUNCPTR(gst_a2dp_sink_stop);
+	basesink_class->render = GST_DEBUG_FUNCPTR(gst_a2dp_sink_render);
+	basesink_class->preroll = GST_DEBUG_FUNCPTR(gst_a2dp_sink_preroll);
+	basesink_class->set_caps = GST_DEBUG_FUNCPTR(gst_a2dp_sink_set_caps);
+	basesink_class->unlock = GST_DEBUG_FUNCPTR(gst_a2dp_sink_unlock);
 
 	g_object_class_install_property(object_class, PROP_DEVICE,
 				g_param_spec_string("device", "Device",
@@ -215,15 +739,23 @@
 static void gst_a2dp_sink_init(GstA2dpSink *self, GstA2dpSinkClass *klass)
 {
 	struct sockaddr_un addr = { AF_UNIX, IPC_SOCKET_NAME };
-	int sk;
+	gint sk;
+	gint err;
 
 	self->device = g_strdup(DEFAULT_DEVICE);
 
 	sk = socket(PF_LOCAL, SOCK_STREAM, 0);
-	if (sk < 0)
+	if (sk < 0) {
+		err = errno;
+		GST_ERROR_OBJECT(self, "Cannot open socket: %s (%d)", 
+			strerror(err), err);
 		return;
+	}
 
 	if (connect(sk, (struct sockaddr *) &addr, sizeof(addr)) < 0) {
+		err = errno;
+		GST_ERROR_OBJECT(self, "Connection fail %s (%d)", 
+			strerror(err), err);
 		close(sk);
 		return;
 	}
@@ -233,5 +765,24 @@
 	g_io_add_watch(self->server, G_IO_IN | G_IO_HUP | G_IO_ERR | G_IO_NVAL,
 							server_callback, self);
 
-	g_io_channel_unref(self->server);
+	self->data = g_new0(struct bluetooth_data, 1);
+	memset(self->data, 0, sizeof(struct bluetooth_data));
+
+	self->data->pipefd[0] = -1;
+	self->data->pipefd[1] = -1;
+
+	if (pipe(self->data->pipefd) < 0)
+		GST_ERROR_OBJECT(self, "%s (%d)", strerror(errno), errno);
+	if (fcntl(self->data->pipefd[0], F_SETFL, O_NONBLOCK) < 0)
+		GST_ERROR_OBJECT(self, "%s (%d)", strerror(errno), errno);
+	if (fcntl(self->data->pipefd[1], F_SETFL, O_NONBLOCK) < 0)
+		GST_ERROR_OBJECT(self, "%s (%d)", strerror(errno), errno);
+
+	self->stream_fd = -1;
+	self->con_state = NOT_CONFIGURED;
+
+	self->con_conf_end = g_cond_new();
+	self->waiting_con_conf = FALSE;
+	self->sink_lock = g_mutex_new();
+
 }
Index: audio/gsta2dpsink.h
===================================================================
RCS file: /cvsroot/bluez/utils/audio/gsta2dpsink.h,v
retrieving revision 1.3
diff -u -r1.3 gsta2dpsink.h
--- audio/gsta2dpsink.h	26 Aug 2007 14:14:34 -0000	1.3
+++ audio/gsta2dpsink.h	5 Oct 2007 18:49:40 -0000
@@ -22,7 +22,7 @@
  */
 
 #include <gst/gst.h>
-#include <gst/audio/gstaudiosink.h>
+#include <gst/base/gstbasesink.h>
 
 G_BEGIN_DECLS
 
@@ -37,19 +37,41 @@
 #define GST_IS_A2DP_SINK_CLASS(obj) \
 	(G_TYPE_CHECK_CLASS_TYPE((klass),GST_TYPE_A2DP_SINK))
 
+enum {
+	NOT_CONFIGURED,
+	CONFIGURING_INIT,
+	CONFIGURING_SENT_CONF,
+	CONFIGURING_RCVD_CONF_RSP,
+	CONFIGURING_RCVD_DEV_CONF,
+	CONFIGURED
+};
+
 typedef struct _GstA2dpSink GstA2dpSink;
 typedef struct _GstA2dpSinkClass GstA2dpSinkClass;
 
+struct bluetooth_data;
+
 struct _GstA2dpSink {
-	GstAudioSink sink;
+	GstBaseSink sink;
 
 	gchar *device;
+	gint stream_fd;
 
+	struct bluetooth_data *data;
 	GIOChannel *server;
+
+	gint con_state;
+
+	GCond *con_conf_end;
+	gboolean waiting_con_conf;
+	GMutex *sink_lock;
+
+	gint total;
+
 };
 
 struct _GstA2dpSinkClass {
-	GstAudioSinkClass parent_class;
+	GstBaseSinkClass parent_class;
 };
 
 GType gst_a2dp_sink_get_type(void);

[-- Attachment #3: Type: text/plain, Size: 314 bytes --]

-------------------------------------------------------------------------
This SF.net email is sponsored by: Splunk Inc.
Still grepping through log files to find problems?  Stop.
Now Search log events and configuration files using AJAX and a browser.
Download your FREE copy of Splunk now >> http://get.splunk.com/

[-- Attachment #4: Type: text/plain, Size: 164 bytes --]

_______________________________________________
Bluez-devel mailing list
Bluez-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/bluez-devel

Re: [Bluez-devel] Patch: inheritance problem in gsta2dpsink

[Luiz Augusto von Dentz]

Hi thiago,

I guess we really need a tiny library for ipc, we dont want to
duplicate code from
pcm_bluetooth.c again in a2dpsink, it might happen that someone need this
in another plugin or so.

-- =

Luiz Augusto von Dentz
Engenheiro de Computa=E7=E3o

-------------------------------------------------------------------------
This SF.net email is sponsored by: Splunk Inc.
Still grepping through log files to find problems?  Stop.
Now Search log events and configuration files using AJAX and a browser.
Download your FREE copy of Splunk now >> http://get.splunk.com/
_______________________________________________
Bluez-devel mailing list
Bluez-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/bluez-devel

Re: [Bluez-devel] Patch: inheritance problem in gsta2dpsink

[Marcel Holtmann]

Hi Luiz,

> I guess we really need a tiny library for ipc, we dont want to
> duplicate code from
> pcm_bluetooth.c again in a2dpsink, it might happen that someone need this
> in another plugin or so.

no no and no. I don't care what other plugins (outside bluez-utils) need
and want at the moment. It is perfectly fine to link an object file more
than once. That must be enough and a library in the end is nothing
different.

Regards

Marcel



-------------------------------------------------------------------------
This SF.net email is sponsored by: Splunk Inc.
Still grepping through log files to find problems?  Stop.
Now Search log events and configuration files using AJAX and a browser.
Download your FREE copy of Splunk now >> http://get.splunk.com/
_______________________________________________
Bluez-devel mailing list
Bluez-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/bluez-devel

Re: [Bluez-devel] Patch: inheritance problem in gsta2dpsink

[Brad Midgley]

Hey

I asked Marcel the same thing a while back.

It would make sense to factor the common stuff into unix.c or create
an ipc.c for that. We can link these to each plugin without messing
around with libraries.

Brad

On 10/8/07, Marcel Holtmann <marcel@holtmann.org> wrote:
> Hi Luiz,
>
> > I guess we really need a tiny library for ipc, we dont want to
> > duplicate code from
> > pcm_bluetooth.c again in a2dpsink, it might happen that someone need this
> > in another plugin or so.
>
> no no and no. I don't care what other plugins (outside bluez-utils) need
> and want at the moment. It is perfectly fine to link an object file more
> than once. That must be enough and a library in the end is nothing
> different.
>
> Regards
>
> Marcel
>
>
>
> -------------------------------------------------------------------------
> This SF.net email is sponsored by: Splunk Inc.
> Still grepping through log files to find problems?  Stop.
> Now Search log events and configuration files using AJAX and a browser.
> Download your FREE copy of Splunk now >> http://get.splunk.com/
> _______________________________________________
> Bluez-devel mailing list
> Bluez-devel@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/bluez-devel
>

-------------------------------------------------------------------------
This SF.net email is sponsored by: Splunk Inc.
Still grepping through log files to find problems?  Stop.
Now Search log events and configuration files using AJAX and a browser.
Download your FREE copy of Splunk now >> http://get.splunk.com/
_______________________________________________
Bluez-devel mailing list
Bluez-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/bluez-devel

Re: [Bluez-devel] Patch: inheritance problem in gsta2dpsink

[Fabien Chevalier]

[-- Attachment #1: Type: text/plain, Size: 2289 bytes --]

Hi all,

Adding more pepper on this already spicy conversation. :-)
> Hey
> 
> I asked Marcel the same thing a while back.
> 
> It would make sense to factor the common stuff into unix.c or create
> an ipc.c for that. We can link these to each plugin without messing
> around with libraries.
> 
> Brad
> 
> On 10/8/07, Marcel Holtmann <marcel@holtmann.org> wrote:
>> Hi Luiz,
>>
>>> I guess we really need a tiny library for ipc, we dont want to
>>> duplicate code from
>>> pcm_bluetooth.c again in a2dpsink, it might happen that someone need this
>>> in another plugin or so.
>> no no and no. I don't care what other plugins (outside bluez-utils) need
>> and want at the moment. 

Well, *I do* care.
As i told you back in Montpellier, i need a way to plug an external 
(proprietary) hardware accelerated audio system onto the audio service.
In fact i could start working on such a library right away, provided 
that it is LGPL, so that i can link a proprietary stuff on top of it.
Even if it is not packaged as an external library at first, provided i 
can somehow copy the code out of bluez tree and rebuild a LGPL library 
out of it would be acceptable as a first step. :-)
If that can't be done then i'm gonna go grumble in a corner and we will
have missed an opportunity to move the bluez audio system forward. :-(
And remembering what Marcel told us during the meeting, the IPC API is 
not public because "we don't really know what we're doing".
My answer to that is:
  - having worked on audio stuff for some time now, i think i know what 
i would be doing if i was to work on this API : which mean it would be 
made stable pretty quickly.
  - We'd better start stabilizing the API before too much effort has 
been put on the gstreamer plugin or any other plugin, otherwise we will 
have to fix them all afterwards.

A last thing : this is likely to be a one time offer, because i need to 
start working on this stuff as soon as today. If we miss the window i 
won't be able to spent company time on this library, which mean i will 
only be able to get involved on my spare time, which i kind of lack
these days...

As always : comments are welcomed !!


>> It is perfectly fine to link an object file more
>> than once. 
Sure it is, we just loose some disk space.

Cheers,

Fabien

[-- Attachment #2: fchevalier.vcf --]
[-- Type: text/x-vcard, Size: 242 bytes --]

begin:vcard
fn:Fabien CHEVALIER
n:CHEVALIER;Fabien
org:SILICOM
adr:;;4 rue de Jouanet; RENNES ATALANTE;;35700;FRANCE
email;internet:fchevalier@silicom.fr
title:Software & Studies Engineer
tel;work:+33 (0) 2 99 84 17 17
version:2.1
end:vcard


[-- Attachment #3: Type: text/plain, Size: 314 bytes --]

-------------------------------------------------------------------------
This SF.net email is sponsored by: Splunk Inc.
Still grepping through log files to find problems?  Stop.
Now Search log events and configuration files using AJAX and a browser.
Download your FREE copy of Splunk now >> http://get.splunk.com/

[-- Attachment #4: Type: text/plain, Size: 164 bytes --]

_______________________________________________
Bluez-devel mailing list
Bluez-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/bluez-devel

Re: [Bluez-devel] Patch: inheritance problem in gsta2dpsink

[Marcel Holtmann]

Hi Fabien,

> Well, *I do* care.
> As i told you back in Montpellier, i need a way to plug an external 
> (proprietary) hardware accelerated audio system onto the audio service.
> In fact i could start working on such a library right away, provided 
> that it is LGPL, so that i can link a proprietary stuff on top of it.
> Even if it is not packaged as an external library at first, provided i 
> can somehow copy the code out of bluez tree and rebuild a LGPL library 
> out of it would be acceptable as a first step. :-)

I am okay with that. That is not the problem. Having that code under
LGPL license is no big deal for me. I am actually fully okay with that.
However please remember that my default license will always be GPL and
you really have to tell me if you want something different.

> If that can't be done then i'm gonna go grumble in a corner and we will
> have missed an opportunity to move the bluez audio system forward. :-(
> And remembering what Marcel told us during the meeting, the IPC API is 
> not public because "we don't really know what we're doing".
> My answer to that is:
>   - having worked on audio stuff for some time now, i think i know what 
> i would be doing if i was to work on this API : which mean it would be 
> made stable pretty quickly.
>   - We'd better start stabilizing the API before too much effort has 
> been put on the gstreamer plugin or any other plugin, otherwise we will 
> have to fix them all afterwards.

This is not gonna work out. You have to deal with API and ABI breakage
if you are working outside the tree and wanna be closed source. That is
the price you pay. I don't see any other way around that.

Remember that a lot of things are highly related to the Bluetooth
specification and A2DP is changing quickly. Actually the versions 1.4
are already on their way to be getting approved at some point. Also the
meta data transfer needs some extra love.

And we need a security audit before ever thinking about making this
really public and stabilize it.

> A last thing : this is likely to be a one time offer, because i need to 
> start working on this stuff as soon as today. If we miss the window i 
> won't be able to spent company time on this library, which mean i will 
> only be able to get involved on my spare time, which i kind of lack
> these days...

The LGPL wish is easy to solve. Get Johan and Luiz to agree here on the
mailing list and then we make ipc.h and an upcoming ipc.c licensed under
LGPL and then you can copy it out from the bluez-utils source and use
it. Just a warning up front. You have to wrap it into a library by
yourself, because otherwise you might be violating the LGPL.

Regards

Marcel



-------------------------------------------------------------------------
This SF.net email is sponsored by: Splunk Inc.
Still grepping through log files to find problems?  Stop.
Now Search log events and configuration files using AJAX and a browser.
Download your FREE copy of Splunk now >> http://get.splunk.com/
_______________________________________________
Bluez-devel mailing list
Bluez-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/bluez-devel

Re: [Bluez-devel] Patch: inheritance problem in gsta2dpsink

[Fabien Chevalier]

Hi Marcel, Luiz, Johan, Brad,
> 
>> Well, *I do* care.
>> As i told you back in Montpellier, i need a way to plug an external 
>> (proprietary) hardware accelerated audio system onto the audio service.
>> In fact i could start working on such a library right away, provided 
>> that it is LGPL, so that i can link a proprietary stuff on top of it.
>> Even if it is not packaged as an external library at first, provided i 
>> can somehow copy the code out of bluez tree and rebuild a LGPL library 
>> out of it would be acceptable as a first step. :-)
> 
> I am okay with that. That is not the problem. Having that code under
> LGPL license is no big deal for me. I am actually fully okay with that.
> However please remember that my default license will always be GPL and
> you really have to tell me if you want something different.

Ok, I 100% agree with you : GPL per default, LGPL for "some special 
bits" :-)

> 
>> If that can't be done then i'm gonna go grumble in a corner and we will
>> have missed an opportunity to move the bluez audio system forward. :-(
>> And remembering what Marcel told us during the meeting, the IPC API is 
>> not public because "we don't really know what we're doing".
>> My answer to that is:
>>   - having worked on audio stuff for some time now, i think i know what 
>> i would be doing if i was to work on this API : which mean it would be 
>> made stable pretty quickly.
>>   - We'd better start stabilizing the API before too much effort has 
>> been put on the gstreamer plugin or any other plugin, otherwise we will 
>> have to fix them all afterwards.
> 
> This is not gonna work out. You have to deal with API and ABI breakage
> if you are working outside the tree and wanna be closed source. That is
> the price you pay. I don't see any other way around that.
> 
> Remember that a lot of things are highly related to the Bluetooth
> specification and A2DP is changing quickly. Actually the versions 1.4
> are already on their way to be getting approved at some point. Also the
> meta data transfer needs some extra love.

I 100% agree with that too. Maybe we just misundertood ourselves about 
how "public" it should bet :-)

Basically what i meant is that we should define & stabilize this API for 
our own internal use (Gst, ALSA, PulseAudio), and for other people 
involved in the Bluez project working also on some out of tree project 
(that would be me and maybe the Access guys AFAIK)
However i'm fully aware that this API will break from time to time due
to changes or new features, and won't blame anybody for that :-)

> 
> And we need a security audit before ever thinking about making this
> really public and stabilize it.

Well, i didn't mean something *that* public (at least not for now).
I just meant something not-completely-unstable-not-documented-that-
will-keep-changing-and-breaking-my-software-for-no-reason. :-)

So this looks that we agree on everything afterall. :-)

> The LGPL wish is easy to solve. Get Johan and Luiz to agree here on the
> mailing list and then we make ipc.h and an upcoming ipc.c licensed under
> LGPL and then you can copy it out from the bluez-utils source and use
> it. 

I was more thinking of as a complete rewrite of ipc.h (we should find it 
another name btw : lets call it newipc.h for now ;-) ).
I already started thinking to how i would like to see it look.
It should be relatively lightweight, expose directly a message based API 
(as you would find on most RTOS), with just a few functions to create 
and free a session to the audio daemon, and to retrieve the data socket, 
so that each user wouldn't have to know the tricks about ancilliary data 
passing :-)
I fact it should be so lightweight that i think we could static inline 
all the functions, and thus we wouldn't even need a *.c file.

I'm waiting for comments, int the meantime i'm gonna start drafting a 
newipc.h :-)

Cheers,

Fabien

-------------------------------------------------------------------------
This SF.net email is sponsored by: Splunk Inc.
Still grepping through log files to find problems?  Stop.
Now Search log events and configuration files using AJAX and a browser.
Download your FREE copy of Splunk now >> http://get.splunk.com/
_______________________________________________
Bluez-devel mailing list
Bluez-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/bluez-devel

Re: [Bluez-devel] Patch: inheritance problem in gsta2dpsink

[Luiz Augusto von Dentz]

Hi Fabien

On 10/9/07, Fabien Chevalier <fabchevalier@free.fr> wrote:
> I was more thinking of as a complete rewrite of ipc.h (we should find it
> another name btw : lets call it newipc.h for now ;-) ).
> I already started thinking to how i would like to see it look.
> It should be relatively lightweight, expose directly a message based API
> (as you would find on most RTOS), with just a few functions to create
> and free a session to the audio daemon, and to retrieve the data socket,
> so that each user wouldn't have to know the tricks about ancilliary data
> passing :-)
> I fact it should be so lightweight that i think we could static inline
> all the functions, and thus we wouldn't even need a *.c file.
>
> I'm waiting for comments, int the meantime i'm gonna start drafting a
> newipc.h :-)

Good, I need to rework ipc anyway since there is no capabilities exchange
between the daemon and the plugin and that cause some problems to make
plugin able choose codec parameters. Right now the daemon choose the best
option it find and return to plugin, we should change that so the plugin could
get the device capabilities and choose the best option by it own.

Thiago is having a good progress with gstreamer plugin, so this is a very good
opportunity to work together and make it happen.

My name suggestion is: libbtaudio

Btw I have no problem with LGPL.

-------------------------------------------------------------------------
This SF.net email is sponsored by: Splunk Inc.
Still grepping through log files to find problems?  Stop.
Now Search log events and configuration files using AJAX and a browser.
Download your FREE copy of Splunk now >> http://get.splunk.com/
_______________________________________________
Bluez-devel mailing list
Bluez-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/bluez-devel

Re: [Bluez-devel] Patch: inheritance problem in gsta2dpsink

[Marcel Holtmann]

Hi Luiz,

> My name suggestion is: libbtaudio

no extra library. Period. I am serious about it. We simply link the
plain object file twice.

Regards

Marcel



-------------------------------------------------------------------------
This SF.net email is sponsored by: Splunk Inc.
Still grepping through log files to find problems?  Stop.
Now Search log events and configuration files using AJAX and a browser.
Download your FREE copy of Splunk now >> http://get.splunk.com/
_______________________________________________
Bluez-devel mailing list
Bluez-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/bluez-devel

Re: [Bluez-devel] Patch: inheritance problem in gsta2dpsink

[Luiz Augusto von Dentz]

Hi Marcel,

On 10/9/07, Marcel Holtmann <marcel@holtmann.org> wrote:
> Hi Luiz,
>
> > My name suggestion is: libbtaudio
>
> no extra library. Period. I am serious about it. We simply link the
> plain object file twice.

I confused, do you agreed or not to create the LGPL library?

-------------------------------------------------------------------------
This SF.net email is sponsored by: Splunk Inc.
Still grepping through log files to find problems?  Stop.
Now Search log events and configuration files using AJAX and a browser.
Download your FREE copy of Splunk now >> http://get.splunk.com/
_______________________________________________
Bluez-devel mailing list
Bluez-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/bluez-devel

Re: [Bluez-devel] Patch: inheritance problem in gsta2dpsink

[Marcel Holtmann]

Hi Luiz,

> > > My name suggestion is: libbtaudio
> >
> > no extra library. Period. I am serious about it. We simply link the
> > plain object file twice.
> 
> I confused, do you agreed or not to create the LGPL library?

we create an object file (like for any other *.c files) under LGPL and
then simply link them.

Regards

Marcel



-------------------------------------------------------------------------
This SF.net email is sponsored by: Splunk Inc.
Still grepping through log files to find problems?  Stop.
Now Search log events and configuration files using AJAX and a browser.
Download your FREE copy of Splunk now >> http://get.splunk.com/
_______________________________________________
Bluez-devel mailing list
Bluez-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/bluez-devel