mirrors/gstreamer
mirrors/gstreamer
1
1
Fork 0
mirror of https://gitlab.freedesktop.org/gstreamer/gstreamer.git synced 2024-12-02 06:31:08 +00:00
Code Issues 1 Projects Releases Wiki Activity
gstreamer/subprojects/gst-editing-services/tests/check/nle/tempochange.c

672 lines
24 KiB
C
Raw Normal View History

Handle changing playback rate Before this patch, NLE and GES did not support NleOperations (respectively GESEffects) that changed the speed/tempo/rate at which the source plays. For example, the 'pitch' element can make audio play faster or slower. In GES 1.5.90 and before, an NleOperation containing the pitch element to change the rate (or tempo) would cause a pipeline state change to PAUSED after that stack; that has been fixed in 1.5.91 (see #755012 [0]). But even then, in 1.5.91 and later, NleComposition would send segment events to its NleSources assuming that one source second is equal to one pipeline second. The resulting early EOS event (in the case of a source rate higher than 1.0) would cause it to switch stacks too early, causing confusion in the timeline and spectacularly messed up output. This patch fixes that by searching for rate-changing elements in GESTrackElements such as GESEffects. If such rate-changing elements are found, their final effect on the playing rate is stored in the corresponding NleObject as the 'media duration factor', named like this because the 'media duration', or source duration, of an NleObject can be computed by multiplying the duration with the media duration factor of that object and its parents (this is called the 'recursive media duration factor'). For example, a 4-second NleSource with an NleOperation with a media duration factor of 2.0 will have an 8-second media duration, which means that for playing 4 seconds in the pipeline, the seek event sent to it must span 8 seconds of media. (So, the 'duration' of an NleObject or GES object always refers to its duration in the timeline, not the media duration.) To summarize: * Rate-changing elements are registered in the GESEffectClass (pitch::tempo and pitch::rate are registered by default); * GESTimelineElement is responsible for detecting rate-changing elements and computing the media_duration_factor; * GESTrackElement is responsible for storing the media_duration_factor in NleObject; * NleComposition is responsible for the recursive_media_duration_factor; * The latter property finally fixes media time computations in NleObject. NLE and GES tests are included. [0] https://bugzilla.gnome.org/show_bug.cgi?id=755012 Differential Revision: https://phabricator.freedesktop.org/D276
2015-12-20 13:03:57 +00:00
/* GStreamer Editing Services
* Copyright (C) 2016 Sjors Gielen <mixml-ges@sjorsgielen.nl>
*
* This library is free software; you can redistribute it and/or
* modify it under the terms of the GNU Library General Public
* License as published by the Free Software Foundation; either
* version 2 of the License, or (at your option) any later version.
*
* This library is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
* Library General Public License for more details.
*
* You should have received a copy of the GNU Library General Public
* License along with this library; if not, write to the
* Free Software Foundation, Inc., 51 Franklin St, Fifth Floor,
* Boston, MA 02110-1301, USA.
*/
#include "common.h"
#include "plugins/nle/nleobject.h"
tests: nle: Call ges_init() only inside test cases Some NLE tests were calling ges_init() from the test suite initialization function. This was causing them to deadlock when running with glib 2.76.3. ges_init() causes a GThreadPool to be initialized. Even if GES at this point doesn't request any thread to be created, since glib 2.63.4+ (see https://gitlab.gnome.org/GNOME/glib/-/commit/8aeca4fa647bfd0f35c4a86b1e6ca6e955519ca5) the first time a GThreadPool is initialized a "pool-spawner" thread is created, which is later used by g_thread_pool_push(). The default behavior of the GStreamer check tests is to fork for every test case. This is not safe if any thread has been created at this point. In this particular case, GThreadPool preserves the state that says a "pool-spawner" thread has been created, and will have access to its mutex and condition variable, but their queues will have different contents as the memory has been forked. In consequence, calls to g_thread_pool_push() will deadlock. The deadlock will not occur if running the tests with CK_FORK=no. This patch modifies the affected tests to only call ges_init() from inside the test cases, fixing the deadlock. Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/4915>
2023-06-21 14:30:52 +00:00
static void late_ges_init (void);
nleobject: stop using media-duration-factor The property had been deprecated and is unused. This property is not needed. Any internal time effect that an nleoperation wraps is itself responsible for converting seek/segment timestamps. Previously, the ghostpads were performing a rate conversion after the rate element had already done so, essentially doubling their effect on seeks and segment times. This was always unnecessary, but went unnoticed by the tempochange test because it was using an identity element rather than an actual rate-changing element. Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-editing-services/-/merge_requests/160>
2020-04-08 16:11:14 +00:00
typedef struct _PadEventData
Handle changing playback rate Before this patch, NLE and GES did not support NleOperations (respectively GESEffects) that changed the speed/tempo/rate at which the source plays. For example, the 'pitch' element can make audio play faster or slower. In GES 1.5.90 and before, an NleOperation containing the pitch element to change the rate (or tempo) would cause a pipeline state change to PAUSED after that stack; that has been fixed in 1.5.91 (see #755012 [0]). But even then, in 1.5.91 and later, NleComposition would send segment events to its NleSources assuming that one source second is equal to one pipeline second. The resulting early EOS event (in the case of a source rate higher than 1.0) would cause it to switch stacks too early, causing confusion in the timeline and spectacularly messed up output. This patch fixes that by searching for rate-changing elements in GESTrackElements such as GESEffects. If such rate-changing elements are found, their final effect on the playing rate is stored in the corresponding NleObject as the 'media duration factor', named like this because the 'media duration', or source duration, of an NleObject can be computed by multiplying the duration with the media duration factor of that object and its parents (this is called the 'recursive media duration factor'). For example, a 4-second NleSource with an NleOperation with a media duration factor of 2.0 will have an 8-second media duration, which means that for playing 4 seconds in the pipeline, the seek event sent to it must span 8 seconds of media. (So, the 'duration' of an NleObject or GES object always refers to its duration in the timeline, not the media duration.) To summarize: * Rate-changing elements are registered in the GESEffectClass (pitch::tempo and pitch::rate are registered by default); * GESTimelineElement is responsible for detecting rate-changing elements and computing the media_duration_factor; * GESTrackElement is responsible for storing the media_duration_factor in NleObject; * NleComposition is responsible for the recursive_media_duration_factor; * The latter property finally fixes media time computations in NleObject. NLE and GES tests are included. [0] https://bugzilla.gnome.org/show_bug.cgi?id=755012 Differential Revision: https://phabricator.freedesktop.org/D276
2015-12-20 13:03:57 +00:00
{
nleobject: stop using media-duration-factor The property had been deprecated and is unused. This property is not needed. Any internal time effect that an nleoperation wraps is itself responsible for converting seek/segment timestamps. Previously, the ghostpads were performing a rate conversion after the rate element had already done so, essentially doubling their effect on seeks and segment times. This was always unnecessary, but went unnoticed by the tempochange test because it was using an identity element rather than an actual rate-changing element. Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-editing-services/-/merge_requests/160>
2020-04-08 16:11:14 +00:00
gchar *name;
guint expect_num_segments;
guint num_segments;
GArray *expect_segment_time;
GArray *expect_segment_num_seeks;
guint expect_num_seeks;
guint num_seeks;
GArray *expect_seek_start;
GArray *expect_seek_stop;
GArray *expect_seek_num_segments;
guint num_eos;
guint expect_num_eos;
} PadEventData;
#define _SEGMENT_FORMAT "flags: %i, rate: %g, applied_rate: %g, format: %i" \
", base: %" G_GUINT64_FORMAT ", offset: %" G_GUINT64_FORMAT ", start: %" \
G_GUINT64_FORMAT ", stop: %" G_GUINT64_FORMAT ", time: %" G_GUINT64_FORMAT \
", position: %" G_GUINT64_FORMAT ", duration: %" G_GUINT64_FORMAT
#define _SEGMENT_ARGS(seg) (seg).flags, (seg).rate, (seg).applied_rate, \
(seg).format, (seg).base, (seg).offset, (seg).start, (seg).stop, \
(seg).time, (seg).position, (seg).duration
static GstPadProbeReturn
_test_pad_events (GstPad * pad, GstPadProbeInfo * info, PadEventData * data)
{
guint num;
GstEvent *event = info->data;
if (GST_EVENT_TYPE (event) == GST_EVENT_SEGMENT) {
guint expect_num_seeks;
const GstSegment *segment;
/* copy the segment start, stop, position and duration since these are
* not yet translated by nleghostpad. Also, don't care about flags. */
GstSegment expect_segment;
gst_event_parse_segment (event, &segment);
GST_DEBUG ("%s segment: " _SEGMENT_FORMAT, data->name,
_SEGMENT_ARGS (*segment));
if (!GST_CLOCK_TIME_IS_VALID (segment->stop)) {
GST_DEBUG ("%s: ignoring pre-roll segment", data->name);
return GST_PAD_PROBE_OK;
}
data->num_segments++;
num = data->num_segments;
fail_unless (num <= data->expect_num_segments, "%s received %u "
"segments, more than the expected %u segments", data->name, num,
data->expect_num_segments);
expect_num_seeks =
g_array_index (data->expect_segment_num_seeks, gint, num - 1);
fail_unless (data->num_seeks == expect_num_seeks, "%s has received %u "
"segments, compared to %u seeks, but expected %u seeks",
data->name, num, data->num_seeks, expect_num_seeks);
/* copy the segment start, stop, position, duration, offset, base
* since these are not yet translated by nleghostpad. */
gst_segment_copy_into (segment, &expect_segment);
expect_segment.rate = 1.0;
expect_segment.applied_rate = 1.0;
expect_segment.format = GST_FORMAT_TIME;
expect_segment.time = g_array_index (data->expect_segment_time,
GstClockTime, num - 1);
fail_unless (gst_segment_is_equal (segment, &expect_segment),
"%s %uth segment is not equal to the expected. Received:\n"
_SEGMENT_FORMAT "\nExpected\n" _SEGMENT_FORMAT, data->name,
num - 1, _SEGMENT_ARGS (*segment), _SEGMENT_ARGS (expect_segment));
} else if (GST_EVENT_TYPE (event) == GST_EVENT_SEEK) {
gdouble rate;
GstFormat format;
GstClockTime expect;
gint64 start, stop;
GstSeekType start_type, stop_type;
guint expect_num_segments;
gst_event_parse_seek (event, &rate, &format, NULL, &start_type, &start,
&stop_type, &stop);
GST_DEBUG ("%s seek: rate: %g, start: %" G_GINT64_FORMAT ", stop: %"
G_GINT64_FORMAT, data->name, rate, start, stop);
data->num_seeks++;
num = data->num_seeks;
fail_unless (num <= data->expect_num_seeks, "%s received %u "
"seeks, more than the expected %u seeks", data->name, num,
data->expect_num_seeks);
expect_num_segments =
g_array_index (data->expect_seek_num_segments, gint, num - 1);
fail_unless (data->num_segments == expect_num_segments, "%s has "
"received %u seeks, compared to %u segments, but expected %u "
"segments", data->name, num, data->num_segments, expect_num_segments);
fail_unless (rate == 1.0, "%s %uth seek has a rate of %g rather than 1.0",
data->name, num - 1, rate);
fail_unless (format == GST_FORMAT_TIME, "%s %uth seek has a format of %i "
" than a time format", data->name, num - 1, format);
/* expect seek-set or seek-none */
fail_if (start_type == GST_SEEK_TYPE_END, "%s %uth seek-start is "
"seek-end", data->name, num - 1);
fail_if (stop_type == GST_SEEK_TYPE_END, "%s %uth seek-stop is "
"seek-end", data->name, num - 1);
expect = g_array_index (data->expect_seek_start, GstClockTime, num - 1);
fail_unless (start == expect, "%s %uth seek start is %" GST_TIME_FORMAT
", rather than the expected %" GST_TIME_FORMAT, data->name, num - 1,
GST_TIME_ARGS (start), GST_TIME_ARGS (expect));
expect = g_array_index (data->expect_seek_stop, GstClockTime, num - 1);
fail_unless (stop == expect, "%s %uth seek stop is %" GST_TIME_FORMAT
", rather than the expected %" GST_TIME_FORMAT, data->name, num - 1,
GST_TIME_ARGS (stop), GST_TIME_ARGS (expect));
} else if (GST_EVENT_TYPE (event) == GST_EVENT_EOS) {
data->num_eos++;
fail_unless (data->num_eos <= data->expect_num_eos, "%s received %u "
"EOS, more than the expected %u EOS", data->name, data->num_eos,
data->expect_num_seeks);
}
return GST_PAD_PROBE_OK;
}
static void
_pad_event_data_check_received (PadEventData * data)
{
fail_unless (data->num_eos == data->expect_num_eos, "%s received %u "
"EOS, rather than %u", data->num_eos, data->expect_num_eos);
fail_unless (data->num_segments == data->expect_num_segments,
"%s received %u segments, rather than %u", data->name,
data->num_segments, data->expect_num_segments);
fail_unless (data->num_seeks == data->expect_num_seeks,
"%s received %u seeks, rather than %u", data->name, data->num_seeks,
data->expect_num_seeks);
}
static void
_pad_event_data_free (PadEventData * data)
{
g_free (data->name);
g_array_unref (data->expect_segment_time);
g_array_unref (data->expect_seek_start);
g_array_unref (data->expect_seek_stop);
g_array_unref (data->expect_segment_num_seeks);
g_array_unref (data->expect_seek_num_segments);
g_free (data);
}
Handle changing playback rate Before this patch, NLE and GES did not support NleOperations (respectively GESEffects) that changed the speed/tempo/rate at which the source plays. For example, the 'pitch' element can make audio play faster or slower. In GES 1.5.90 and before, an NleOperation containing the pitch element to change the rate (or tempo) would cause a pipeline state change to PAUSED after that stack; that has been fixed in 1.5.91 (see #755012 [0]). But even then, in 1.5.91 and later, NleComposition would send segment events to its NleSources assuming that one source second is equal to one pipeline second. The resulting early EOS event (in the case of a source rate higher than 1.0) would cause it to switch stacks too early, causing confusion in the timeline and spectacularly messed up output. This patch fixes that by searching for rate-changing elements in GESTrackElements such as GESEffects. If such rate-changing elements are found, their final effect on the playing rate is stored in the corresponding NleObject as the 'media duration factor', named like this because the 'media duration', or source duration, of an NleObject can be computed by multiplying the duration with the media duration factor of that object and its parents (this is called the 'recursive media duration factor'). For example, a 4-second NleSource with an NleOperation with a media duration factor of 2.0 will have an 8-second media duration, which means that for playing 4 seconds in the pipeline, the seek event sent to it must span 8 seconds of media. (So, the 'duration' of an NleObject or GES object always refers to its duration in the timeline, not the media duration.) To summarize: * Rate-changing elements are registered in the GESEffectClass (pitch::tempo and pitch::rate are registered by default); * GESTimelineElement is responsible for detecting rate-changing elements and computing the media_duration_factor; * GESTrackElement is responsible for storing the media_duration_factor in NleObject; * NleComposition is responsible for the recursive_media_duration_factor; * The latter property finally fixes media time computations in NleObject. NLE and GES tests are included. [0] https://bugzilla.gnome.org/show_bug.cgi?id=755012 Differential Revision: https://phabricator.freedesktop.org/D276
2015-12-20 13:03:57 +00:00
nleobject: stop using media-duration-factor The property had been deprecated and is unused. This property is not needed. Any internal time effect that an nleoperation wraps is itself responsible for converting seek/segment timestamps. Previously, the ghostpads were performing a rate conversion after the rate element had already done so, essentially doubling their effect on seeks and segment times. This was always unnecessary, but went unnoticed by the tempochange test because it was using an identity element rather than an actual rate-changing element. Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-editing-services/-/merge_requests/160>
2020-04-08 16:11:14 +00:00
static PadEventData *
_pad_event_data_new (GstElement * element, const gchar * pad_name,
const gchar * suffix)
{
GstPad *pad;
PadEventData *data = g_new0 (PadEventData, 1);
data->expect_segment_time = g_array_new (FALSE, FALSE, sizeof (GstClockTime));
data->expect_seek_start = g_array_new (FALSE, FALSE, sizeof (GstClockTime));
data->expect_seek_stop = g_array_new (FALSE, FALSE, sizeof (GstClockTime));
data->expect_seek_num_segments = g_array_new (FALSE, FALSE, sizeof (guint));
data->expect_segment_num_seeks = g_array_new (FALSE, FALSE, sizeof (guint));
data->name = g_strdup_printf ("%s:%s(%s):%s", G_OBJECT_TYPE_NAME (element),
GST_ELEMENT_NAME (element), pad_name, suffix);
pad = gst_element_get_static_pad (element, pad_name);
fail_unless (pad, "%s not found", data->name);
gst_pad_add_probe (pad, GST_PAD_PROBE_TYPE_EVENT_DOWNSTREAM |
GST_PAD_PROBE_TYPE_EVENT_UPSTREAM, (GstPadProbeCallback) _test_pad_events,
data, (GDestroyNotify) _pad_event_data_free);
gst_object_unref (pad);
return data;
}
static void
_pad_event_data_add_expect_segment (PadEventData * data, GstClockTime time)
{
data->expect_num_segments++;
g_array_append_val (data->expect_segment_time, time);
g_array_append_val (data->expect_segment_num_seeks, data->expect_num_seeks);
}
static void
_pad_event_data_add_expect_seek (PadEventData * data, GstClockTime start,
GstClockTime stop)
{
data->expect_num_seeks++;
g_array_append_val (data->expect_seek_start, start);
g_array_append_val (data->expect_seek_stop, stop);
g_array_append_val (data->expect_seek_num_segments,
data->expect_num_segments);
}
static void
_pad_event_data_add_expect_seek_then_segment (PadEventData * data,
GstClockTime start, GstClockTime stop)
{
_pad_event_data_add_expect_seek (data, start, stop);
_pad_event_data_add_expect_segment (data, start);
}
#define _EXPECT_SEEK_SEGMENT(data, start, stop) \
_pad_event_data_add_expect_seek_then_segment (data, start, stop)
static GstElement *
_get_source (GstElement * nle_source)
{
GList *tmp;
GstElement *bin, *src = NULL;
fail_unless (g_list_length (GST_BIN_CHILDREN (nle_source)), 1);
bin = GST_BIN_CHILDREN (nle_source)->data;
fail_unless (GST_IS_BIN (bin));
for (tmp = GST_BIN_CHILDREN (bin); src == NULL && tmp; tmp = tmp->next) {
if (g_strrstr (GST_ELEMENT_NAME (tmp->data), "audiotestsrc"))
src = tmp->data;
}
fail_unless (src);
return src;
}
enum
{
NLE_PREV_SRC,
NLE_POST_SRC,
NLE_SOURCE_SRC,
NLE_OPER_SRC,
NLE_OPER_SINK,
NLE_IDENTITY_SRC,
PREV_SRC,
POST_SRC,
SOURCE_SRC,
PITCH_SRC,
PITCH_SINK,
IDENTITY_SRC,
SINK_SINK,
NUM_DATA
};
static PadEventData **
_setup_test (GstElement * pipeline, gdouble rate)
{
GstElement *sink, *pitch, *src, *prev, *post, *identity;
GstElement *comp, *nle_source, *nle_prev, *nle_post, *nle_oper, *nle_identity;
gboolean ret;
gchar *suffix;
PadEventData **data = g_new0 (PadEventData *, NUM_DATA);
/* composition */
Handle changing playback rate Before this patch, NLE and GES did not support NleOperations (respectively GESEffects) that changed the speed/tempo/rate at which the source plays. For example, the 'pitch' element can make audio play faster or slower. In GES 1.5.90 and before, an NleOperation containing the pitch element to change the rate (or tempo) would cause a pipeline state change to PAUSED after that stack; that has been fixed in 1.5.91 (see #755012 [0]). But even then, in 1.5.91 and later, NleComposition would send segment events to its NleSources assuming that one source second is equal to one pipeline second. The resulting early EOS event (in the case of a source rate higher than 1.0) would cause it to switch stacks too early, causing confusion in the timeline and spectacularly messed up output. This patch fixes that by searching for rate-changing elements in GESTrackElements such as GESEffects. If such rate-changing elements are found, their final effect on the playing rate is stored in the corresponding NleObject as the 'media duration factor', named like this because the 'media duration', or source duration, of an NleObject can be computed by multiplying the duration with the media duration factor of that object and its parents (this is called the 'recursive media duration factor'). For example, a 4-second NleSource with an NleOperation with a media duration factor of 2.0 will have an 8-second media duration, which means that for playing 4 seconds in the pipeline, the seek event sent to it must span 8 seconds of media. (So, the 'duration' of an NleObject or GES object always refers to its duration in the timeline, not the media duration.) To summarize: * Rate-changing elements are registered in the GESEffectClass (pitch::tempo and pitch::rate are registered by default); * GESTimelineElement is responsible for detecting rate-changing elements and computing the media_duration_factor; * GESTrackElement is responsible for storing the media_duration_factor in NleObject; * NleComposition is responsible for the recursive_media_duration_factor; * The latter property finally fixes media time computations in NleObject. NLE and GES tests are included. [0] https://bugzilla.gnome.org/show_bug.cgi?id=755012 Differential Revision: https://phabricator.freedesktop.org/D276
2015-12-20 13:03:57 +00:00
comp =
gst_element_factory_make_or_warn ("nlecomposition", "test_composition");
gst_element_set_state (comp, GST_STATE_READY);
nleobject: stop using media-duration-factor The property had been deprecated and is unused. This property is not needed. Any internal time effect that an nleoperation wraps is itself responsible for converting seek/segment timestamps. Previously, the ghostpads were performing a rate conversion after the rate element had already done so, essentially doubling their effect on seeks and segment times. This was always unnecessary, but went unnoticed by the tempochange test because it was using an identity element rather than an actual rate-changing element. Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-editing-services/-/merge_requests/160>
2020-04-08 16:11:14 +00:00
/* sink */
tests/check: Use fake{audio,video}sink The tests already depend on -bad, so this should be OK. Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-editing-services/-/merge_requests/254>
2021-05-20 15:47:41 +00:00
sink = gst_element_factory_make_or_warn ("fakeaudiosink", "sink");
tests: nle: Set sync=FALSE on sinks A significant portion of the NLE test suite was often timing out due to the tests taking way longer than necessary because the sinks were synchronizing to the clock, which is the default behavior for fakevideosink and fakeaudiosink. Notable was the case of nleoperation.c:test_pyramid_operations, that ran through a 10 second stream twice. As the default timeout is 20 seconds, this made the test flaky. Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/4914>
2023-06-21 14:24:12 +00:00
g_object_set (sink, "sync", FALSE, NULL);
Handle changing playback rate Before this patch, NLE and GES did not support NleOperations (respectively GESEffects) that changed the speed/tempo/rate at which the source plays. For example, the 'pitch' element can make audio play faster or slower. In GES 1.5.90 and before, an NleOperation containing the pitch element to change the rate (or tempo) would cause a pipeline state change to PAUSED after that stack; that has been fixed in 1.5.91 (see #755012 [0]). But even then, in 1.5.91 and later, NleComposition would send segment events to its NleSources assuming that one source second is equal to one pipeline second. The resulting early EOS event (in the case of a source rate higher than 1.0) would cause it to switch stacks too early, causing confusion in the timeline and spectacularly messed up output. This patch fixes that by searching for rate-changing elements in GESTrackElements such as GESEffects. If such rate-changing elements are found, their final effect on the playing rate is stored in the corresponding NleObject as the 'media duration factor', named like this because the 'media duration', or source duration, of an NleObject can be computed by multiplying the duration with the media duration factor of that object and its parents (this is called the 'recursive media duration factor'). For example, a 4-second NleSource with an NleOperation with a media duration factor of 2.0 will have an 8-second media duration, which means that for playing 4 seconds in the pipeline, the seek event sent to it must span 8 seconds of media. (So, the 'duration' of an NleObject or GES object always refers to its duration in the timeline, not the media duration.) To summarize: * Rate-changing elements are registered in the GESEffectClass (pitch::tempo and pitch::rate are registered by default); * GESTimelineElement is responsible for detecting rate-changing elements and computing the media_duration_factor; * GESTrackElement is responsible for storing the media_duration_factor in NleObject; * NleComposition is responsible for the recursive_media_duration_factor; * The latter property finally fixes media time computations in NleObject. NLE and GES tests are included. [0] https://bugzilla.gnome.org/show_bug.cgi?id=755012 Differential Revision: https://phabricator.freedesktop.org/D276
2015-12-20 13:03:57 +00:00
gst_bin_add_many (GST_BIN (pipeline), comp, sink, NULL);
gst_element_link (comp, sink);
nleobject: stop using media-duration-factor The property had been deprecated and is unused. This property is not needed. Any internal time effect that an nleoperation wraps is itself responsible for converting seek/segment timestamps. Previously, the ghostpads were performing a rate conversion after the rate element had already done so, essentially doubling their effect on seeks and segment times. This was always unnecessary, but went unnoticed by the tempochange test because it was using an identity element rather than an actual rate-changing element. Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-editing-services/-/merge_requests/160>
2020-04-08 16:11:14 +00:00
/* sources */
nle_source =
audiotest_bin_src ("nle_source", 3 * GST_SECOND, 4 * GST_SECOND, 3,
FALSE);
ges: tests: Use correct variable types when setting properties From https://gitlab.freedesktop.org/gstreamer/gst-editing-services/-/merge_requests/247 Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/4995>
2021-05-01 20:50:11 +00:00
g_object_set (nle_source, "inpoint", (guint64) 7 * GST_SECOND, NULL);
nleobject: stop using media-duration-factor The property had been deprecated and is unused. This property is not needed. Any internal time effect that an nleoperation wraps is itself responsible for converting seek/segment timestamps. Previously, the ghostpads were performing a rate conversion after the rate element had already done so, essentially doubling their effect on seeks and segment times. This was always unnecessary, but went unnoticed by the tempochange test because it was using an identity element rather than an actual rate-changing element. Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-editing-services/-/merge_requests/160>
2020-04-08 16:11:14 +00:00
src = _get_source (nle_source);
g_object_set (src, "name", "middle-source", NULL);
nle_prev =
audiotest_bin_src ("nle_previous", 0 * GST_SECOND, 3 * GST_SECOND, 2,
FALSE);
ges: tests: Use correct variable types when setting properties From https://gitlab.freedesktop.org/gstreamer/gst-editing-services/-/merge_requests/247 Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/4995>
2021-05-01 20:50:11 +00:00
g_object_set (nle_prev, "inpoint", (guint64) 99 * GST_SECOND, NULL);
nleobject: stop using media-duration-factor The property had been deprecated and is unused. This property is not needed. Any internal time effect that an nleoperation wraps is itself responsible for converting seek/segment timestamps. Previously, the ghostpads were performing a rate conversion after the rate element had already done so, essentially doubling their effect on seeks and segment times. This was always unnecessary, but went unnoticed by the tempochange test because it was using an identity element rather than an actual rate-changing element. Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-editing-services/-/merge_requests/160>
2020-04-08 16:11:14 +00:00
prev = _get_source (nle_prev);
g_object_set (src, "name", "previous-source", NULL);
nle_post =
audiotest_bin_src ("post", 7 * GST_SECOND, 5 * GST_SECOND, 2, FALSE);
ges: tests: Use correct variable types when setting properties From https://gitlab.freedesktop.org/gstreamer/gst-editing-services/-/merge_requests/247 Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/4995>
2021-05-01 20:50:11 +00:00
g_object_set (nle_post, "inpoint", (guint64) 20 * GST_SECOND, NULL);
nleobject: stop using media-duration-factor The property had been deprecated and is unused. This property is not needed. Any internal time effect that an nleoperation wraps is itself responsible for converting seek/segment timestamps. Previously, the ghostpads were performing a rate conversion after the rate element had already done so, essentially doubling their effect on seeks and segment times. This was always unnecessary, but went unnoticed by the tempochange test because it was using an identity element rather than an actual rate-changing element. Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-editing-services/-/merge_requests/160>
2020-04-08 16:11:14 +00:00
post = _get_source (nle_post);
g_object_set (src, "name", "post-source", NULL);
/* Operation, must share the same start and duration as the upstream
* source */
nle_oper =
new_operation ("nle_oper", "pitch", 3 * GST_SECOND, 4 * GST_SECOND, 2);
fail_unless (g_list_length (GST_BIN_CHILDREN (nle_oper)) == 1);
pitch = GST_ELEMENT (GST_BIN_CHILDREN (nle_oper)->data);
g_object_set (pitch, "rate", rate, NULL);
/* cover with an identity operation
* rate effect has lower priority, so we don't need the same start or
* duration */
nle_identity =
new_operation ("nle_identity", "identity", 0, 12 * GST_SECOND, 1);
ges: tests: Use correct variable types when setting properties From https://gitlab.freedesktop.org/gstreamer/gst-editing-services/-/merge_requests/247 Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/4995>
2021-05-01 20:50:11 +00:00
g_object_set (nle_identity, "inpoint", (guint64) 5 * GST_SECOND, NULL);
nleobject: stop using media-duration-factor The property had been deprecated and is unused. This property is not needed. Any internal time effect that an nleoperation wraps is itself responsible for converting seek/segment timestamps. Previously, the ghostpads were performing a rate conversion after the rate element had already done so, essentially doubling their effect on seeks and segment times. This was always unnecessary, but went unnoticed by the tempochange test because it was using an identity element rather than an actual rate-changing element. Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-editing-services/-/merge_requests/160>
2020-04-08 16:11:14 +00:00
fail_unless (g_list_length (GST_BIN_CHILDREN (nle_oper)) == 1);
identity = GST_ELEMENT (GST_BIN_CHILDREN (nle_identity)->data);
nle_composition_add (GST_BIN (comp), nle_source);
nle_composition_add (GST_BIN (comp), nle_prev);
nle_composition_add (GST_BIN (comp), nle_post);
nle_composition_add (GST_BIN (comp), nle_oper);
nle_composition_add (GST_BIN (comp), nle_identity);
ret = FALSE;
Handle changing playback rate Before this patch, NLE and GES did not support NleOperations (respectively GESEffects) that changed the speed/tempo/rate at which the source plays. For example, the 'pitch' element can make audio play faster or slower. In GES 1.5.90 and before, an NleOperation containing the pitch element to change the rate (or tempo) would cause a pipeline state change to PAUSED after that stack; that has been fixed in 1.5.91 (see #755012 [0]). But even then, in 1.5.91 and later, NleComposition would send segment events to its NleSources assuming that one source second is equal to one pipeline second. The resulting early EOS event (in the case of a source rate higher than 1.0) would cause it to switch stacks too early, causing confusion in the timeline and spectacularly messed up output. This patch fixes that by searching for rate-changing elements in GESTrackElements such as GESEffects. If such rate-changing elements are found, their final effect on the playing rate is stored in the corresponding NleObject as the 'media duration factor', named like this because the 'media duration', or source duration, of an NleObject can be computed by multiplying the duration with the media duration factor of that object and its parents (this is called the 'recursive media duration factor'). For example, a 4-second NleSource with an NleOperation with a media duration factor of 2.0 will have an 8-second media duration, which means that for playing 4 seconds in the pipeline, the seek event sent to it must span 8 seconds of media. (So, the 'duration' of an NleObject or GES object always refers to its duration in the timeline, not the media duration.) To summarize: * Rate-changing elements are registered in the GESEffectClass (pitch::tempo and pitch::rate are registered by default); * GESTimelineElement is responsible for detecting rate-changing elements and computing the media_duration_factor; * GESTrackElement is responsible for storing the media_duration_factor in NleObject; * NleComposition is responsible for the recursive_media_duration_factor; * The latter property finally fixes media time computations in NleObject. NLE and GES tests are included. [0] https://bugzilla.gnome.org/show_bug.cgi?id=755012 Differential Revision: https://phabricator.freedesktop.org/D276
2015-12-20 13:03:57 +00:00
commit_and_wait (comp, &ret);
nleobject: stop using media-duration-factor The property had been deprecated and is unused. This property is not needed. Any internal time effect that an nleoperation wraps is itself responsible for converting seek/segment timestamps. Previously, the ghostpads were performing a rate conversion after the rate element had already done so, essentially doubling their effect on seeks and segment times. This was always unnecessary, but went unnoticed by the tempochange test because it was using an identity element rather than an actual rate-changing element. Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-editing-services/-/merge_requests/160>
2020-04-08 16:11:14 +00:00
fail_unless (ret);
check_start_stop_duration (nle_source, 3 * GST_SECOND, 7 * GST_SECOND,
4 * GST_SECOND);
check_start_stop_duration (nle_oper, 3 * GST_SECOND, 7 * GST_SECOND,
4 * GST_SECOND);
check_start_stop_duration (nle_prev, 0, 3 * GST_SECOND, 3 * GST_SECOND);
check_start_stop_duration (nle_post, 7 * GST_SECOND, 12 * GST_SECOND,
5 * GST_SECOND);
check_start_stop_duration (nle_identity, 0, 12 * GST_SECOND, 12 * GST_SECOND);
check_start_stop_duration (comp, 0, 12 * GST_SECOND, 12 * GST_SECOND);
/* create data */
suffix = g_strdup_printf ("rate=%g", rate);
/* source */
data[NLE_SOURCE_SRC] = _pad_event_data_new (nle_source, "src", suffix);
data[NLE_PREV_SRC] = _pad_event_data_new (nle_prev, "src", suffix);
data[NLE_POST_SRC] = _pad_event_data_new (nle_post, "src", suffix);
data[SOURCE_SRC] = _pad_event_data_new (src, "src", suffix);
data[PREV_SRC] = _pad_event_data_new (prev, "src", suffix);
data[POST_SRC] = _pad_event_data_new (post, "src", suffix);
/* rate operation */
data[NLE_OPER_SRC] = _pad_event_data_new (nle_oper, "src", suffix);
data[NLE_OPER_SINK] = _pad_event_data_new (nle_oper, "sink", suffix);
data[PITCH_SRC] = _pad_event_data_new (pitch, "src", suffix);
data[PITCH_SINK] = _pad_event_data_new (pitch, "sink", suffix);
/* identity: only care about the source pads */
data[NLE_IDENTITY_SRC] = _pad_event_data_new (nle_identity, "src", suffix);
data[IDENTITY_SRC] = _pad_event_data_new (identity, "src", suffix);
/* sink */
data[SINK_SINK] = _pad_event_data_new (sink, "sink", suffix);
g_free (suffix);
return data;
}
Handle changing playback rate Before this patch, NLE and GES did not support NleOperations (respectively GESEffects) that changed the speed/tempo/rate at which the source plays. For example, the 'pitch' element can make audio play faster or slower. In GES 1.5.90 and before, an NleOperation containing the pitch element to change the rate (or tempo) would cause a pipeline state change to PAUSED after that stack; that has been fixed in 1.5.91 (see #755012 [0]). But even then, in 1.5.91 and later, NleComposition would send segment events to its NleSources assuming that one source second is equal to one pipeline second. The resulting early EOS event (in the case of a source rate higher than 1.0) would cause it to switch stacks too early, causing confusion in the timeline and spectacularly messed up output. This patch fixes that by searching for rate-changing elements in GESTrackElements such as GESEffects. If such rate-changing elements are found, their final effect on the playing rate is stored in the corresponding NleObject as the 'media duration factor', named like this because the 'media duration', or source duration, of an NleObject can be computed by multiplying the duration with the media duration factor of that object and its parents (this is called the 'recursive media duration factor'). For example, a 4-second NleSource with an NleOperation with a media duration factor of 2.0 will have an 8-second media duration, which means that for playing 4 seconds in the pipeline, the seek event sent to it must span 8 seconds of media. (So, the 'duration' of an NleObject or GES object always refers to its duration in the timeline, not the media duration.) To summarize: * Rate-changing elements are registered in the GESEffectClass (pitch::tempo and pitch::rate are registered by default); * GESTimelineElement is responsible for detecting rate-changing elements and computing the media_duration_factor; * GESTrackElement is responsible for storing the media_duration_factor in NleObject; * NleComposition is responsible for the recursive_media_duration_factor; * The latter property finally fixes media time computations in NleObject. NLE and GES tests are included. [0] https://bugzilla.gnome.org/show_bug.cgi?id=755012 Differential Revision: https://phabricator.freedesktop.org/D276
2015-12-20 13:03:57 +00:00
nleobject: stop using media-duration-factor The property had been deprecated and is unused. This property is not needed. Any internal time effect that an nleoperation wraps is itself responsible for converting seek/segment timestamps. Previously, the ghostpads were performing a rate conversion after the rate element had already done so, essentially doubling their effect on seeks and segment times. This was always unnecessary, but went unnoticed by the tempochange test because it was using an identity element rather than an actual rate-changing element. Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-editing-services/-/merge_requests/160>
2020-04-08 16:11:14 +00:00
GST_START_TEST (test_tempochange_play)
{
GstElement *pipeline;
GstBus *bus;
GstMessage *message;
gboolean carry_on;
PadEventData **data;
gdouble rates[3] = { 0.5, 4.0, 1.0 };
guint i, j;
tests: nle: Call ges_init() only inside test cases Some NLE tests were calling ges_init() from the test suite initialization function. This was causing them to deadlock when running with glib 2.76.3. ges_init() causes a GThreadPool to be initialized. Even if GES at this point doesn't request any thread to be created, since glib 2.63.4+ (see https://gitlab.gnome.org/GNOME/glib/-/commit/8aeca4fa647bfd0f35c4a86b1e6ca6e955519ca5) the first time a GThreadPool is initialized a "pool-spawner" thread is created, which is later used by g_thread_pool_push(). The default behavior of the GStreamer check tests is to fork for every test case. This is not safe if any thread has been created at this point. In this particular case, GThreadPool preserves the state that says a "pool-spawner" thread has been created, and will have access to its mutex and condition variable, but their queues will have different contents as the memory has been forked. In consequence, calls to g_thread_pool_push() will deadlock. The deadlock will not occur if running the tests with CK_FORK=no. This patch modifies the affected tests to only call ges_init() from inside the test cases, fixing the deadlock. Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/4915>
2023-06-21 14:30:52 +00:00
late_ges_init ();
nleobject: stop using media-duration-factor The property had been deprecated and is unused. This property is not needed. Any internal time effect that an nleoperation wraps is itself responsible for converting seek/segment timestamps. Previously, the ghostpads were performing a rate conversion after the rate element had already done so, essentially doubling their effect on seeks and segment times. This was always unnecessary, but went unnoticed by the tempochange test because it was using an identity element rather than an actual rate-changing element. Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-editing-services/-/merge_requests/160>
2020-04-08 16:11:14 +00:00
for (i = 0; i < G_N_ELEMENTS (rates); i++) {
gdouble rate = rates[i];
GST_DEBUG ("rate = %g", rate);
pipeline = gst_pipeline_new ("test_pipeline");
data = _setup_test (pipeline, rate);
/* initial seek */
_EXPECT_SEEK_SEGMENT (data[SINK_SINK], 0, 3 * GST_SECOND);
_EXPECT_SEEK_SEGMENT (data[NLE_IDENTITY_SRC], 0, 3 * GST_SECOND);
/* nleobject will convert the seek by removing start and adding inpoint */
_EXPECT_SEEK_SEGMENT (data[IDENTITY_SRC], 5 * GST_SECOND, 8 * GST_SECOND);
_EXPECT_SEEK_SEGMENT (data[NLE_PREV_SRC], 0, 3 * GST_SECOND);
_EXPECT_SEEK_SEGMENT (data[PREV_SRC], 99 * GST_SECOND, 102 * GST_SECOND);
/* rate-stack seek */
_EXPECT_SEEK_SEGMENT (data[SINK_SINK], 3 * GST_SECOND, 7 * GST_SECOND);
_EXPECT_SEEK_SEGMENT (data[NLE_IDENTITY_SRC], 3 * GST_SECOND,
7 * GST_SECOND);
_EXPECT_SEEK_SEGMENT (data[IDENTITY_SRC], 8 * GST_SECOND, 12 * GST_SECOND);
_EXPECT_SEEK_SEGMENT (data[NLE_OPER_SRC], 3 * GST_SECOND, 7 * GST_SECOND);
_EXPECT_SEEK_SEGMENT (data[PITCH_SRC], 0, 4 * GST_SECOND);
/* pitch element will change the stop time, e.g. if rate=2.0, then we
* want to use up twice as much source, so the stop time doubles */
_EXPECT_SEEK_SEGMENT (data[PITCH_SINK], 0, rate * 4 * GST_SECOND);
_EXPECT_SEEK_SEGMENT (data[NLE_OPER_SINK], 3 * GST_SECOND,
(GstClockTime) (rate * 4 * GST_SECOND) + 3 * GST_SECOND);
_EXPECT_SEEK_SEGMENT (data[NLE_SOURCE_SRC], 3 * GST_SECOND,
(GstClockTime) (rate * 4 * GST_SECOND) + 3 * GST_SECOND);
_EXPECT_SEEK_SEGMENT (data[SOURCE_SRC], 7 * GST_SECOND,
(GstClockTime) (rate * 4 * GST_SECOND) + 7 * GST_SECOND);
/* final part only involves post source */
_EXPECT_SEEK_SEGMENT (data[SINK_SINK], 7 * GST_SECOND, 12 * GST_SECOND);
_EXPECT_SEEK_SEGMENT (data[NLE_IDENTITY_SRC], 7 * GST_SECOND,
12 * GST_SECOND);
_EXPECT_SEEK_SEGMENT (data[IDENTITY_SRC], 12 * GST_SECOND, 17 * GST_SECOND);
_EXPECT_SEEK_SEGMENT (data[NLE_POST_SRC], 7 * GST_SECOND, 12 * GST_SECOND);
/* nleobject will convert the seek by removing start and adding
* inpoint */
_EXPECT_SEEK_SEGMENT (data[POST_SRC], 20 * GST_SECOND, 25 * GST_SECOND);
/* expect 1 EOS from each, apart from identity, which will get 3 since
* part of 3 stacks */
for (j = 0; j < NUM_DATA; j++)
data[j]->expect_num_eos = 1;
data[IDENTITY_SRC]->expect_num_eos = 3;
data[NLE_IDENTITY_SRC]->expect_num_eos = 3;
bus = gst_element_get_bus (GST_ELEMENT (pipeline));
GST_DEBUG ("Setting pipeline to PLAYING");
fail_if (gst_element_set_state (GST_ELEMENT (pipeline),
GST_STATE_PLAYING) == GST_STATE_CHANGE_FAILURE);
GST_DEBUG ("Let's poll the bus");
carry_on = TRUE;
while (carry_on) {
message = gst_bus_poll (bus, GST_MESSAGE_ANY, GST_SECOND / 10);
if (message) {
switch (GST_MESSAGE_TYPE (message)) {
case GST_MESSAGE_EOS:
if (message->src == GST_OBJECT (pipeline)) {
GST_DEBUG ("Setting pipeline to NULL");
fail_unless (gst_element_set_state (GST_ELEMENT (pipeline),
GST_STATE_NULL) == GST_STATE_CHANGE_SUCCESS);
carry_on = FALSE;
}
break;
case GST_MESSAGE_ERROR:
fail_error_message (message);
break;
default:
break;
Handle changing playback rate Before this patch, NLE and GES did not support NleOperations (respectively GESEffects) that changed the speed/tempo/rate at which the source plays. For example, the 'pitch' element can make audio play faster or slower. In GES 1.5.90 and before, an NleOperation containing the pitch element to change the rate (or tempo) would cause a pipeline state change to PAUSED after that stack; that has been fixed in 1.5.91 (see #755012 [0]). But even then, in 1.5.91 and later, NleComposition would send segment events to its NleSources assuming that one source second is equal to one pipeline second. The resulting early EOS event (in the case of a source rate higher than 1.0) would cause it to switch stacks too early, causing confusion in the timeline and spectacularly messed up output. This patch fixes that by searching for rate-changing elements in GESTrackElements such as GESEffects. If such rate-changing elements are found, their final effect on the playing rate is stored in the corresponding NleObject as the 'media duration factor', named like this because the 'media duration', or source duration, of an NleObject can be computed by multiplying the duration with the media duration factor of that object and its parents (this is called the 'recursive media duration factor'). For example, a 4-second NleSource with an NleOperation with a media duration factor of 2.0 will have an 8-second media duration, which means that for playing 4 seconds in the pipeline, the seek event sent to it must span 8 seconds of media. (So, the 'duration' of an NleObject or GES object always refers to its duration in the timeline, not the media duration.) To summarize: * Rate-changing elements are registered in the GESEffectClass (pitch::tempo and pitch::rate are registered by default); * GESTimelineElement is responsible for detecting rate-changing elements and computing the media_duration_factor; * GESTrackElement is responsible for storing the media_duration_factor in NleObject; * NleComposition is responsible for the recursive_media_duration_factor; * The latter property finally fixes media time computations in NleObject. NLE and GES tests are included. [0] https://bugzilla.gnome.org/show_bug.cgi?id=755012 Differential Revision: https://phabricator.freedesktop.org/D276
2015-12-20 13:03:57 +00:00
}
nleobject: stop using media-duration-factor The property had been deprecated and is unused. This property is not needed. Any internal time effect that an nleoperation wraps is itself responsible for converting seek/segment timestamps. Previously, the ghostpads were performing a rate conversion after the rate element had already done so, essentially doubling their effect on seeks and segment times. This was always unnecessary, but went unnoticed by the tempochange test because it was using an identity element rather than an actual rate-changing element. Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-editing-services/-/merge_requests/160>
2020-04-08 16:11:14 +00:00
gst_message_unref (message);
Handle changing playback rate Before this patch, NLE and GES did not support NleOperations (respectively GESEffects) that changed the speed/tempo/rate at which the source plays. For example, the 'pitch' element can make audio play faster or slower. In GES 1.5.90 and before, an NleOperation containing the pitch element to change the rate (or tempo) would cause a pipeline state change to PAUSED after that stack; that has been fixed in 1.5.91 (see #755012 [0]). But even then, in 1.5.91 and later, NleComposition would send segment events to its NleSources assuming that one source second is equal to one pipeline second. The resulting early EOS event (in the case of a source rate higher than 1.0) would cause it to switch stacks too early, causing confusion in the timeline and spectacularly messed up output. This patch fixes that by searching for rate-changing elements in GESTrackElements such as GESEffects. If such rate-changing elements are found, their final effect on the playing rate is stored in the corresponding NleObject as the 'media duration factor', named like this because the 'media duration', or source duration, of an NleObject can be computed by multiplying the duration with the media duration factor of that object and its parents (this is called the 'recursive media duration factor'). For example, a 4-second NleSource with an NleOperation with a media duration factor of 2.0 will have an 8-second media duration, which means that for playing 4 seconds in the pipeline, the seek event sent to it must span 8 seconds of media. (So, the 'duration' of an NleObject or GES object always refers to its duration in the timeline, not the media duration.) To summarize: * Rate-changing elements are registered in the GESEffectClass (pitch::tempo and pitch::rate are registered by default); * GESTimelineElement is responsible for detecting rate-changing elements and computing the media_duration_factor; * GESTrackElement is responsible for storing the media_duration_factor in NleObject; * NleComposition is responsible for the recursive_media_duration_factor; * The latter property finally fixes media time computations in NleObject. NLE and GES tests are included. [0] https://bugzilla.gnome.org/show_bug.cgi?id=755012 Differential Revision: https://phabricator.freedesktop.org/D276
2015-12-20 13:03:57 +00:00
}
}
nleobject: stop using media-duration-factor The property had been deprecated and is unused. This property is not needed. Any internal time effect that an nleoperation wraps is itself responsible for converting seek/segment timestamps. Previously, the ghostpads were performing a rate conversion after the rate element had already done so, essentially doubling their effect on seeks and segment times. This was always unnecessary, but went unnoticed by the tempochange test because it was using an identity element rather than an actual rate-changing element. Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-editing-services/-/merge_requests/160>
2020-04-08 16:11:14 +00:00
for (j = 0; j < NUM_DATA; j++)
_pad_event_data_check_received (data[j]);
Handle changing playback rate Before this patch, NLE and GES did not support NleOperations (respectively GESEffects) that changed the speed/tempo/rate at which the source plays. For example, the 'pitch' element can make audio play faster or slower. In GES 1.5.90 and before, an NleOperation containing the pitch element to change the rate (or tempo) would cause a pipeline state change to PAUSED after that stack; that has been fixed in 1.5.91 (see #755012 [0]). But even then, in 1.5.91 and later, NleComposition would send segment events to its NleSources assuming that one source second is equal to one pipeline second. The resulting early EOS event (in the case of a source rate higher than 1.0) would cause it to switch stacks too early, causing confusion in the timeline and spectacularly messed up output. This patch fixes that by searching for rate-changing elements in GESTrackElements such as GESEffects. If such rate-changing elements are found, their final effect on the playing rate is stored in the corresponding NleObject as the 'media duration factor', named like this because the 'media duration', or source duration, of an NleObject can be computed by multiplying the duration with the media duration factor of that object and its parents (this is called the 'recursive media duration factor'). For example, a 4-second NleSource with an NleOperation with a media duration factor of 2.0 will have an 8-second media duration, which means that for playing 4 seconds in the pipeline, the seek event sent to it must span 8 seconds of media. (So, the 'duration' of an NleObject or GES object always refers to its duration in the timeline, not the media duration.) To summarize: * Rate-changing elements are registered in the GESEffectClass (pitch::tempo and pitch::rate are registered by default); * GESTimelineElement is responsible for detecting rate-changing elements and computing the media_duration_factor; * GESTrackElement is responsible for storing the media_duration_factor in NleObject; * NleComposition is responsible for the recursive_media_duration_factor; * The latter property finally fixes media time computations in NleObject. NLE and GES tests are included. [0] https://bugzilla.gnome.org/show_bug.cgi?id=755012 Differential Revision: https://phabricator.freedesktop.org/D276
2015-12-20 13:03:57 +00:00
nleobject: stop using media-duration-factor The property had been deprecated and is unused. This property is not needed. Any internal time effect that an nleoperation wraps is itself responsible for converting seek/segment timestamps. Previously, the ghostpads were performing a rate conversion after the rate element had already done so, essentially doubling their effect on seeks and segment times. This was always unnecessary, but went unnoticed by the tempochange test because it was using an identity element rather than an actual rate-changing element. Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-editing-services/-/merge_requests/160>
2020-04-08 16:11:14 +00:00
ASSERT_OBJECT_REFCOUNT_BETWEEN (pipeline, "main pipeline", 1, 2);
gst_object_unref (pipeline);
ASSERT_OBJECT_REFCOUNT_BETWEEN (bus, "main bus", 1, 2);
gst_object_unref (bus);
g_free (data);
}
}
Handle changing playback rate Before this patch, NLE and GES did not support NleOperations (respectively GESEffects) that changed the speed/tempo/rate at which the source plays. For example, the 'pitch' element can make audio play faster or slower. In GES 1.5.90 and before, an NleOperation containing the pitch element to change the rate (or tempo) would cause a pipeline state change to PAUSED after that stack; that has been fixed in 1.5.91 (see #755012 [0]). But even then, in 1.5.91 and later, NleComposition would send segment events to its NleSources assuming that one source second is equal to one pipeline second. The resulting early EOS event (in the case of a source rate higher than 1.0) would cause it to switch stacks too early, causing confusion in the timeline and spectacularly messed up output. This patch fixes that by searching for rate-changing elements in GESTrackElements such as GESEffects. If such rate-changing elements are found, their final effect on the playing rate is stored in the corresponding NleObject as the 'media duration factor', named like this because the 'media duration', or source duration, of an NleObject can be computed by multiplying the duration with the media duration factor of that object and its parents (this is called the 'recursive media duration factor'). For example, a 4-second NleSource with an NleOperation with a media duration factor of 2.0 will have an 8-second media duration, which means that for playing 4 seconds in the pipeline, the seek event sent to it must span 8 seconds of media. (So, the 'duration' of an NleObject or GES object always refers to its duration in the timeline, not the media duration.) To summarize: * Rate-changing elements are registered in the GESEffectClass (pitch::tempo and pitch::rate are registered by default); * GESTimelineElement is responsible for detecting rate-changing elements and computing the media_duration_factor; * GESTrackElement is responsible for storing the media_duration_factor in NleObject; * NleComposition is responsible for the recursive_media_duration_factor; * The latter property finally fixes media time computations in NleObject. NLE and GES tests are included. [0] https://bugzilla.gnome.org/show_bug.cgi?id=755012 Differential Revision: https://phabricator.freedesktop.org/D276
2015-12-20 13:03:57 +00:00
nleobject: stop using media-duration-factor The property had been deprecated and is unused. This property is not needed. Any internal time effect that an nleoperation wraps is itself responsible for converting seek/segment timestamps. Previously, the ghostpads were performing a rate conversion after the rate element had already done so, essentially doubling their effect on seeks and segment times. This was always unnecessary, but went unnoticed by the tempochange test because it was using an identity element rather than an actual rate-changing element. Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-editing-services/-/merge_requests/160>
2020-04-08 16:11:14 +00:00
GST_END_TEST;
Handle changing playback rate Before this patch, NLE and GES did not support NleOperations (respectively GESEffects) that changed the speed/tempo/rate at which the source plays. For example, the 'pitch' element can make audio play faster or slower. In GES 1.5.90 and before, an NleOperation containing the pitch element to change the rate (or tempo) would cause a pipeline state change to PAUSED after that stack; that has been fixed in 1.5.91 (see #755012 [0]). But even then, in 1.5.91 and later, NleComposition would send segment events to its NleSources assuming that one source second is equal to one pipeline second. The resulting early EOS event (in the case of a source rate higher than 1.0) would cause it to switch stacks too early, causing confusion in the timeline and spectacularly messed up output. This patch fixes that by searching for rate-changing elements in GESTrackElements such as GESEffects. If such rate-changing elements are found, their final effect on the playing rate is stored in the corresponding NleObject as the 'media duration factor', named like this because the 'media duration', or source duration, of an NleObject can be computed by multiplying the duration with the media duration factor of that object and its parents (this is called the 'recursive media duration factor'). For example, a 4-second NleSource with an NleOperation with a media duration factor of 2.0 will have an 8-second media duration, which means that for playing 4 seconds in the pipeline, the seek event sent to it must span 8 seconds of media. (So, the 'duration' of an NleObject or GES object always refers to its duration in the timeline, not the media duration.) To summarize: * Rate-changing elements are registered in the GESEffectClass (pitch::tempo and pitch::rate are registered by default); * GESTimelineElement is responsible for detecting rate-changing elements and computing the media_duration_factor; * GESTrackElement is responsible for storing the media_duration_factor in NleObject; * NleComposition is responsible for the recursive_media_duration_factor; * The latter property finally fixes media time computations in NleObject. NLE and GES tests are included. [0] https://bugzilla.gnome.org/show_bug.cgi?id=755012 Differential Revision: https://phabricator.freedesktop.org/D276
2015-12-20 13:03:57 +00:00
nleobject: stop using media-duration-factor The property had been deprecated and is unused. This property is not needed. Any internal time effect that an nleoperation wraps is itself responsible for converting seek/segment timestamps. Previously, the ghostpads were performing a rate conversion after the rate element had already done so, essentially doubling their effect on seeks and segment times. This was always unnecessary, but went unnoticed by the tempochange test because it was using an identity element rather than an actual rate-changing element. Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-editing-services/-/merge_requests/160>
2020-04-08 16:11:14 +00:00
#define _WAIT_UNTIL_ASYNC_DONE \
{ \
GST_DEBUG ("Let's poll the bus"); \
carry_on = TRUE; \
while (carry_on) { \
message = gst_bus_poll (bus, GST_MESSAGE_ANY, GST_SECOND / 10); \
if (message) { \
switch (GST_MESSAGE_TYPE (message)) { \
case GST_MESSAGE_EOS: \
fail_if (TRUE, "Received EOS"); \
break; \
case GST_MESSAGE_ERROR: \
fail_error_message (message); \
break; \
case GST_MESSAGE_ASYNC_DONE: \
carry_on = FALSE; \
break; \
default: \
break; \
} \
gst_message_unref (message); \
} \
} \
}
Handle changing playback rate Before this patch, NLE and GES did not support NleOperations (respectively GESEffects) that changed the speed/tempo/rate at which the source plays. For example, the 'pitch' element can make audio play faster or slower. In GES 1.5.90 and before, an NleOperation containing the pitch element to change the rate (or tempo) would cause a pipeline state change to PAUSED after that stack; that has been fixed in 1.5.91 (see #755012 [0]). But even then, in 1.5.91 and later, NleComposition would send segment events to its NleSources assuming that one source second is equal to one pipeline second. The resulting early EOS event (in the case of a source rate higher than 1.0) would cause it to switch stacks too early, causing confusion in the timeline and spectacularly messed up output. This patch fixes that by searching for rate-changing elements in GESTrackElements such as GESEffects. If such rate-changing elements are found, their final effect on the playing rate is stored in the corresponding NleObject as the 'media duration factor', named like this because the 'media duration', or source duration, of an NleObject can be computed by multiplying the duration with the media duration factor of that object and its parents (this is called the 'recursive media duration factor'). For example, a 4-second NleSource with an NleOperation with a media duration factor of 2.0 will have an 8-second media duration, which means that for playing 4 seconds in the pipeline, the seek event sent to it must span 8 seconds of media. (So, the 'duration' of an NleObject or GES object always refers to its duration in the timeline, not the media duration.) To summarize: * Rate-changing elements are registered in the GESEffectClass (pitch::tempo and pitch::rate are registered by default); * GESTimelineElement is responsible for detecting rate-changing elements and computing the media_duration_factor; * GESTrackElement is responsible for storing the media_duration_factor in NleObject; * NleComposition is responsible for the recursive_media_duration_factor; * The latter property finally fixes media time computations in NleObject. NLE and GES tests are included. [0] https://bugzilla.gnome.org/show_bug.cgi?id=755012 Differential Revision: https://phabricator.freedesktop.org/D276
2015-12-20 13:03:57 +00:00
nleobject: stop using media-duration-factor The property had been deprecated and is unused. This property is not needed. Any internal time effect that an nleoperation wraps is itself responsible for converting seek/segment timestamps. Previously, the ghostpads were performing a rate conversion after the rate element had already done so, essentially doubling their effect on seeks and segment times. This was always unnecessary, but went unnoticed by the tempochange test because it was using an identity element rather than an actual rate-changing element. Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-editing-services/-/merge_requests/160>
2020-04-08 16:11:14 +00:00
GST_START_TEST (test_tempochange_seek)
{
GstElement *pipeline;
GstBus *bus;
GstMessage *message;
gboolean carry_on;
PadEventData **data;
gdouble rates[3] = { 2.0, 0.25, 1.0 };
guint i, j;
GstClockTime offset = 0.1 * GST_SECOND;
tests: nle: Call ges_init() only inside test cases Some NLE tests were calling ges_init() from the test suite initialization function. This was causing them to deadlock when running with glib 2.76.3. ges_init() causes a GThreadPool to be initialized. Even if GES at this point doesn't request any thread to be created, since glib 2.63.4+ (see https://gitlab.gnome.org/GNOME/glib/-/commit/8aeca4fa647bfd0f35c4a86b1e6ca6e955519ca5) the first time a GThreadPool is initialized a "pool-spawner" thread is created, which is later used by g_thread_pool_push(). The default behavior of the GStreamer check tests is to fork for every test case. This is not safe if any thread has been created at this point. In this particular case, GThreadPool preserves the state that says a "pool-spawner" thread has been created, and will have access to its mutex and condition variable, but their queues will have different contents as the memory has been forked. In consequence, calls to g_thread_pool_push() will deadlock. The deadlock will not occur if running the tests with CK_FORK=no. This patch modifies the affected tests to only call ges_init() from inside the test cases, fixing the deadlock. Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/4915>
2023-06-21 14:30:52 +00:00
late_ges_init ();
nleobject: stop using media-duration-factor The property had been deprecated and is unused. This property is not needed. Any internal time effect that an nleoperation wraps is itself responsible for converting seek/segment timestamps. Previously, the ghostpads were performing a rate conversion after the rate element had already done so, essentially doubling their effect on seeks and segment times. This was always unnecessary, but went unnoticed by the tempochange test because it was using an identity element rather than an actual rate-changing element. Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-editing-services/-/merge_requests/160>
2020-04-08 16:11:14 +00:00
for (i = 0; i < G_N_ELEMENTS (rates); i++) {
gdouble rate = rates[i];
GST_DEBUG ("rate = %g", rate);
pipeline = gst_pipeline_new ("test_pipeline");
data = _setup_test (pipeline, rate);
/* initial seek from the pause */
_EXPECT_SEEK_SEGMENT (data[SINK_SINK], 0, 3 * GST_SECOND);
_EXPECT_SEEK_SEGMENT (data[NLE_IDENTITY_SRC], 0, 3 * GST_SECOND);
_EXPECT_SEEK_SEGMENT (data[IDENTITY_SRC], 5 * GST_SECOND, 8 * GST_SECOND);
_EXPECT_SEEK_SEGMENT (data[NLE_PREV_SRC], 0, 3 * GST_SECOND);
_EXPECT_SEEK_SEGMENT (data[PREV_SRC], 99 * GST_SECOND, 102 * GST_SECOND);
GST_DEBUG ("Setting pipeline to PAUSED");
fail_unless (gst_element_set_state (GST_ELEMENT (pipeline),
GST_STATE_PAUSED) == GST_STATE_CHANGE_ASYNC);
bus = gst_element_get_bus (GST_ELEMENT (pipeline));
_WAIT_UNTIL_ASYNC_DONE;
for (j = 0; j < NUM_DATA; j++)
_pad_event_data_check_received (data[j]);
/* first seek for just after the start of the rate effect */
/* NOTE: neither prev nor post should receive anything */
/* sink will receive two seeks: one that initiates the pre-roll, and
* then the seek with the stop set */
/* expect no segment for the first seek */
_pad_event_data_add_expect_seek (data[SINK_SINK], 3 * GST_SECOND + offset,
GST_CLOCK_TIME_NONE);
_EXPECT_SEEK_SEGMENT (data[SINK_SINK], 3 * GST_SECOND + offset,
7 * GST_SECOND);
_EXPECT_SEEK_SEGMENT (data[NLE_IDENTITY_SRC], 3 * GST_SECOND + offset,
7 * GST_SECOND);
_EXPECT_SEEK_SEGMENT (data[IDENTITY_SRC], 8 * GST_SECOND + offset,
12 * GST_SECOND);
_EXPECT_SEEK_SEGMENT (data[NLE_OPER_SRC], 3 * GST_SECOND + offset,
7 * GST_SECOND);
_EXPECT_SEEK_SEGMENT (data[PITCH_SRC], offset, 4 * GST_SECOND);
_EXPECT_SEEK_SEGMENT (data[PITCH_SINK], rate * offset,
rate * 4 * GST_SECOND);
_EXPECT_SEEK_SEGMENT (data[NLE_OPER_SINK],
3 * GST_SECOND + (GstClockTime) (rate * offset),
3 * GST_SECOND + (GstClockTime) (rate * 4 * GST_SECOND));
_EXPECT_SEEK_SEGMENT (data[NLE_SOURCE_SRC],
3 * GST_SECOND + (GstClockTime) (rate * offset),
3 * GST_SECOND + (GstClockTime) (rate * 4 * GST_SECOND));
_EXPECT_SEEK_SEGMENT (data[SOURCE_SRC],
7 * GST_SECOND + (GstClockTime) (rate * offset),
7 * GST_SECOND + (GstClockTime) (rate * 4 * GST_SECOND));
/* perform seek */
fail_unless (gst_element_seek_simple (pipeline, GST_FORMAT_TIME,
GST_SEEK_FLAG_FLUSH, 3 * GST_SECOND + offset));
_WAIT_UNTIL_ASYNC_DONE;
for (j = 0; j < NUM_DATA; j++)
_pad_event_data_check_received (data[j]);
/* now seek to just before the end */
_pad_event_data_add_expect_seek (data[SINK_SINK], 7 * GST_SECOND - offset,
GST_CLOCK_TIME_NONE);
_EXPECT_SEEK_SEGMENT (data[SINK_SINK], 7 * GST_SECOND - offset,
7 * GST_SECOND);
_EXPECT_SEEK_SEGMENT (data[NLE_IDENTITY_SRC], 7 * GST_SECOND - offset,
7 * GST_SECOND);
_EXPECT_SEEK_SEGMENT (data[IDENTITY_SRC], 12 * GST_SECOND - offset,
12 * GST_SECOND);
_EXPECT_SEEK_SEGMENT (data[NLE_OPER_SRC], 7 * GST_SECOND - offset,
7 * GST_SECOND);
_EXPECT_SEEK_SEGMENT (data[PITCH_SRC], 4 * GST_SECOND - offset,
4 * GST_SECOND);
_EXPECT_SEEK_SEGMENT (data[PITCH_SINK],
rate * (4 * GST_SECOND) - rate * offset, rate * 4 * GST_SECOND);
_EXPECT_SEEK_SEGMENT (data[NLE_OPER_SINK],
3 * GST_SECOND + (GstClockTime) (rate * (4 * GST_SECOND - offset)),
3 * GST_SECOND + (GstClockTime) (rate * 4 * GST_SECOND));
_EXPECT_SEEK_SEGMENT (data[NLE_SOURCE_SRC],
3 * GST_SECOND + (GstClockTime) (rate * (4 * GST_SECOND - offset)),
3 * GST_SECOND + (GstClockTime) (rate * 4 * GST_SECOND));
_EXPECT_SEEK_SEGMENT (data[SOURCE_SRC],
7 * GST_SECOND + (GstClockTime) (rate * (4 * GST_SECOND - offset)),
7 * GST_SECOND + (GstClockTime) (rate * 4 * GST_SECOND));
/* perform seek */
fail_unless (gst_element_seek_simple (pipeline, GST_FORMAT_TIME,
GST_SEEK_FLAG_FLUSH, 7 * GST_SECOND - offset));
_WAIT_UNTIL_ASYNC_DONE;
for (j = 0; j < NUM_DATA; j++)
_pad_event_data_check_received (data[j]);
GST_DEBUG ("Setting pipeline to NULL");
fail_unless (gst_element_set_state (GST_ELEMENT (pipeline),
GST_STATE_NULL) == GST_STATE_CHANGE_SUCCESS);
ASSERT_OBJECT_REFCOUNT_BETWEEN (pipeline, "main pipeline", 1, 2);
gst_object_unref (pipeline);
ASSERT_OBJECT_REFCOUNT_BETWEEN (bus, "main bus", 1, 2);
gst_object_unref (bus);
g_free (data);
}
Handle changing playback rate Before this patch, NLE and GES did not support NleOperations (respectively GESEffects) that changed the speed/tempo/rate at which the source plays. For example, the 'pitch' element can make audio play faster or slower. In GES 1.5.90 and before, an NleOperation containing the pitch element to change the rate (or tempo) would cause a pipeline state change to PAUSED after that stack; that has been fixed in 1.5.91 (see #755012 [0]). But even then, in 1.5.91 and later, NleComposition would send segment events to its NleSources assuming that one source second is equal to one pipeline second. The resulting early EOS event (in the case of a source rate higher than 1.0) would cause it to switch stacks too early, causing confusion in the timeline and spectacularly messed up output. This patch fixes that by searching for rate-changing elements in GESTrackElements such as GESEffects. If such rate-changing elements are found, their final effect on the playing rate is stored in the corresponding NleObject as the 'media duration factor', named like this because the 'media duration', or source duration, of an NleObject can be computed by multiplying the duration with the media duration factor of that object and its parents (this is called the 'recursive media duration factor'). For example, a 4-second NleSource with an NleOperation with a media duration factor of 2.0 will have an 8-second media duration, which means that for playing 4 seconds in the pipeline, the seek event sent to it must span 8 seconds of media. (So, the 'duration' of an NleObject or GES object always refers to its duration in the timeline, not the media duration.) To summarize: * Rate-changing elements are registered in the GESEffectClass (pitch::tempo and pitch::rate are registered by default); * GESTimelineElement is responsible for detecting rate-changing elements and computing the media_duration_factor; * GESTrackElement is responsible for storing the media_duration_factor in NleObject; * NleComposition is responsible for the recursive_media_duration_factor; * The latter property finally fixes media time computations in NleObject. NLE and GES tests are included. [0] https://bugzilla.gnome.org/show_bug.cgi?id=755012 Differential Revision: https://phabricator.freedesktop.org/D276
2015-12-20 13:03:57 +00:00
}
GST_END_TEST;
tests: nle: Call ges_init() only inside test cases Some NLE tests were calling ges_init() from the test suite initialization function. This was causing them to deadlock when running with glib 2.76.3. ges_init() causes a GThreadPool to be initialized. Even if GES at this point doesn't request any thread to be created, since glib 2.63.4+ (see https://gitlab.gnome.org/GNOME/glib/-/commit/8aeca4fa647bfd0f35c4a86b1e6ca6e955519ca5) the first time a GThreadPool is initialized a "pool-spawner" thread is created, which is later used by g_thread_pool_push(). The default behavior of the GStreamer check tests is to fork for every test case. This is not safe if any thread has been created at this point. In this particular case, GThreadPool preserves the state that says a "pool-spawner" thread has been created, and will have access to its mutex and condition variable, but their queues will have different contents as the memory has been forked. In consequence, calls to g_thread_pool_push() will deadlock. The deadlock will not occur if running the tests with CK_FORK=no. This patch modifies the affected tests to only call ges_init() from inside the test cases, fixing the deadlock. Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/4915>
2023-06-21 14:30:52 +00:00
static void
late_ges_init ()
Handle changing playback rate Before this patch, NLE and GES did not support NleOperations (respectively GESEffects) that changed the speed/tempo/rate at which the source plays. For example, the 'pitch' element can make audio play faster or slower. In GES 1.5.90 and before, an NleOperation containing the pitch element to change the rate (or tempo) would cause a pipeline state change to PAUSED after that stack; that has been fixed in 1.5.91 (see #755012 [0]). But even then, in 1.5.91 and later, NleComposition would send segment events to its NleSources assuming that one source second is equal to one pipeline second. The resulting early EOS event (in the case of a source rate higher than 1.0) would cause it to switch stacks too early, causing confusion in the timeline and spectacularly messed up output. This patch fixes that by searching for rate-changing elements in GESTrackElements such as GESEffects. If such rate-changing elements are found, their final effect on the playing rate is stored in the corresponding NleObject as the 'media duration factor', named like this because the 'media duration', or source duration, of an NleObject can be computed by multiplying the duration with the media duration factor of that object and its parents (this is called the 'recursive media duration factor'). For example, a 4-second NleSource with an NleOperation with a media duration factor of 2.0 will have an 8-second media duration, which means that for playing 4 seconds in the pipeline, the seek event sent to it must span 8 seconds of media. (So, the 'duration' of an NleObject or GES object always refers to its duration in the timeline, not the media duration.) To summarize: * Rate-changing elements are registered in the GESEffectClass (pitch::tempo and pitch::rate are registered by default); * GESTimelineElement is responsible for detecting rate-changing elements and computing the media_duration_factor; * GESTrackElement is responsible for storing the media_duration_factor in NleObject; * NleComposition is responsible for the recursive_media_duration_factor; * The latter property finally fixes media time computations in NleObject. NLE and GES tests are included. [0] https://bugzilla.gnome.org/show_bug.cgi?id=755012 Differential Revision: https://phabricator.freedesktop.org/D276
2015-12-20 13:03:57 +00:00
{
tests: nle: Call ges_init() only inside test cases Some NLE tests were calling ges_init() from the test suite initialization function. This was causing them to deadlock when running with glib 2.76.3. ges_init() causes a GThreadPool to be initialized. Even if GES at this point doesn't request any thread to be created, since glib 2.63.4+ (see https://gitlab.gnome.org/GNOME/glib/-/commit/8aeca4fa647bfd0f35c4a86b1e6ca6e955519ca5) the first time a GThreadPool is initialized a "pool-spawner" thread is created, which is later used by g_thread_pool_push(). The default behavior of the GStreamer check tests is to fork for every test case. This is not safe if any thread has been created at this point. In this particular case, GThreadPool preserves the state that says a "pool-spawner" thread has been created, and will have access to its mutex and condition variable, but their queues will have different contents as the memory has been forked. In consequence, calls to g_thread_pool_push() will deadlock. The deadlock will not occur if running the tests with CK_FORK=no. This patch modifies the affected tests to only call ges_init() from inside the test cases, fixing the deadlock. Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/4915>
2023-06-21 14:30:52 +00:00
/* We need to do this inside the test cases, not during the initialization
* of the suite, as ges_init() will initialize thread pools, which cannot
* work properly after a fork. */
Handle changing playback rate Before this patch, NLE and GES did not support NleOperations (respectively GESEffects) that changed the speed/tempo/rate at which the source plays. For example, the 'pitch' element can make audio play faster or slower. In GES 1.5.90 and before, an NleOperation containing the pitch element to change the rate (or tempo) would cause a pipeline state change to PAUSED after that stack; that has been fixed in 1.5.91 (see #755012 [0]). But even then, in 1.5.91 and later, NleComposition would send segment events to its NleSources assuming that one source second is equal to one pipeline second. The resulting early EOS event (in the case of a source rate higher than 1.0) would cause it to switch stacks too early, causing confusion in the timeline and spectacularly messed up output. This patch fixes that by searching for rate-changing elements in GESTrackElements such as GESEffects. If such rate-changing elements are found, their final effect on the playing rate is stored in the corresponding NleObject as the 'media duration factor', named like this because the 'media duration', or source duration, of an NleObject can be computed by multiplying the duration with the media duration factor of that object and its parents (this is called the 'recursive media duration factor'). For example, a 4-second NleSource with an NleOperation with a media duration factor of 2.0 will have an 8-second media duration, which means that for playing 4 seconds in the pipeline, the seek event sent to it must span 8 seconds of media. (So, the 'duration' of an NleObject or GES object always refers to its duration in the timeline, not the media duration.) To summarize: * Rate-changing elements are registered in the GESEffectClass (pitch::tempo and pitch::rate are registered by default); * GESTimelineElement is responsible for detecting rate-changing elements and computing the media_duration_factor; * GESTrackElement is responsible for storing the media_duration_factor in NleObject; * NleComposition is responsible for the recursive_media_duration_factor; * The latter property finally fixes media time computations in NleObject. NLE and GES tests are included. [0] https://bugzilla.gnome.org/show_bug.cgi?id=755012 Differential Revision: https://phabricator.freedesktop.org/D276
2015-12-20 13:03:57 +00:00
introduce ges_deinit() GstDiscoverer objects were leaked by tests making the leaks detector unusable. Introduce ges_deinit(), similiar to gst_deinit(), doing some cleanup before exiting the process. https://bugzilla.gnome.org/show_bug.cgi?id=776805
2017-02-06 11:07:26 +00:00
if (atexit (ges_deinit) != 0) {
GST_ERROR ("failed to set ges_deinit as exit function");
}
Handle changing playback rate Before this patch, NLE and GES did not support NleOperations (respectively GESEffects) that changed the speed/tempo/rate at which the source plays. For example, the 'pitch' element can make audio play faster or slower. In GES 1.5.90 and before, an NleOperation containing the pitch element to change the rate (or tempo) would cause a pipeline state change to PAUSED after that stack; that has been fixed in 1.5.91 (see #755012 [0]). But even then, in 1.5.91 and later, NleComposition would send segment events to its NleSources assuming that one source second is equal to one pipeline second. The resulting early EOS event (in the case of a source rate higher than 1.0) would cause it to switch stacks too early, causing confusion in the timeline and spectacularly messed up output. This patch fixes that by searching for rate-changing elements in GESTrackElements such as GESEffects. If such rate-changing elements are found, their final effect on the playing rate is stored in the corresponding NleObject as the 'media duration factor', named like this because the 'media duration', or source duration, of an NleObject can be computed by multiplying the duration with the media duration factor of that object and its parents (this is called the 'recursive media duration factor'). For example, a 4-second NleSource with an NleOperation with a media duration factor of 2.0 will have an 8-second media duration, which means that for playing 4 seconds in the pipeline, the seek event sent to it must span 8 seconds of media. (So, the 'duration' of an NleObject or GES object always refers to its duration in the timeline, not the media duration.) To summarize: * Rate-changing elements are registered in the GESEffectClass (pitch::tempo and pitch::rate are registered by default); * GESTimelineElement is responsible for detecting rate-changing elements and computing the media_duration_factor; * GESTrackElement is responsible for storing the media_duration_factor in NleObject; * NleComposition is responsible for the recursive_media_duration_factor; * The latter property finally fixes media time computations in NleObject. NLE and GES tests are included. [0] https://bugzilla.gnome.org/show_bug.cgi?id=755012 Differential Revision: https://phabricator.freedesktop.org/D276
2015-12-20 13:03:57 +00:00
ges_init ();
tests: nle: Call ges_init() only inside test cases Some NLE tests were calling ges_init() from the test suite initialization function. This was causing them to deadlock when running with glib 2.76.3. ges_init() causes a GThreadPool to be initialized. Even if GES at this point doesn't request any thread to be created, since glib 2.63.4+ (see https://gitlab.gnome.org/GNOME/glib/-/commit/8aeca4fa647bfd0f35c4a86b1e6ca6e955519ca5) the first time a GThreadPool is initialized a "pool-spawner" thread is created, which is later used by g_thread_pool_push(). The default behavior of the GStreamer check tests is to fork for every test case. This is not safe if any thread has been created at this point. In this particular case, GThreadPool preserves the state that says a "pool-spawner" thread has been created, and will have access to its mutex and condition variable, but their queues will have different contents as the memory has been forked. In consequence, calls to g_thread_pool_push() will deadlock. The deadlock will not occur if running the tests with CK_FORK=no. This patch modifies the affected tests to only call ges_init() from inside the test cases, fixing the deadlock. Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/4915>
2023-06-21 14:30:52 +00:00
}
static Suite *
gnonlin_suite (void)
{
Suite *s = suite_create ("nle");
TCase *tc_chain = tcase_create ("tempochange");
Handle changing playback rate Before this patch, NLE and GES did not support NleOperations (respectively GESEffects) that changed the speed/tempo/rate at which the source plays. For example, the 'pitch' element can make audio play faster or slower. In GES 1.5.90 and before, an NleOperation containing the pitch element to change the rate (or tempo) would cause a pipeline state change to PAUSED after that stack; that has been fixed in 1.5.91 (see #755012 [0]). But even then, in 1.5.91 and later, NleComposition would send segment events to its NleSources assuming that one source second is equal to one pipeline second. The resulting early EOS event (in the case of a source rate higher than 1.0) would cause it to switch stacks too early, causing confusion in the timeline and spectacularly messed up output. This patch fixes that by searching for rate-changing elements in GESTrackElements such as GESEffects. If such rate-changing elements are found, their final effect on the playing rate is stored in the corresponding NleObject as the 'media duration factor', named like this because the 'media duration', or source duration, of an NleObject can be computed by multiplying the duration with the media duration factor of that object and its parents (this is called the 'recursive media duration factor'). For example, a 4-second NleSource with an NleOperation with a media duration factor of 2.0 will have an 8-second media duration, which means that for playing 4 seconds in the pipeline, the seek event sent to it must span 8 seconds of media. (So, the 'duration' of an NleObject or GES object always refers to its duration in the timeline, not the media duration.) To summarize: * Rate-changing elements are registered in the GESEffectClass (pitch::tempo and pitch::rate are registered by default); * GESTimelineElement is responsible for detecting rate-changing elements and computing the media_duration_factor; * GESTrackElement is responsible for storing the media_duration_factor in NleObject; * NleComposition is responsible for the recursive_media_duration_factor; * The latter property finally fixes media time computations in NleObject. NLE and GES tests are included. [0] https://bugzilla.gnome.org/show_bug.cgi?id=755012 Differential Revision: https://phabricator.freedesktop.org/D276
2015-12-20 13:03:57 +00:00
suite_add_tcase (s, tc_chain);
check: give nle_tempochange test more time These test can take longer than most under valgrind, so give them a little more time until they timeout. Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-editing-services/-/merge_requests/160>
2020-04-14 09:22:09 +00:00
/* give the tests a little more time than the default
* CK_DEFAULT_TIMEOUT=20, this is sometimes needed for running under
* valgrind */
tcase_set_timeout (tc_chain, 40.0);
nleobject: stop using media-duration-factor The property had been deprecated and is unused. This property is not needed. Any internal time effect that an nleoperation wraps is itself responsible for converting seek/segment timestamps. Previously, the ghostpads were performing a rate conversion after the rate element had already done so, essentially doubling their effect on seeks and segment times. This was always unnecessary, but went unnoticed by the tempochange test because it was using an identity element rather than an actual rate-changing element. Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-editing-services/-/merge_requests/160>
2020-04-08 16:11:14 +00:00
tcase_add_test (tc_chain, test_tempochange_play);
tcase_add_test (tc_chain, test_tempochange_seek);
Handle changing playback rate Before this patch, NLE and GES did not support NleOperations (respectively GESEffects) that changed the speed/tempo/rate at which the source plays. For example, the 'pitch' element can make audio play faster or slower. In GES 1.5.90 and before, an NleOperation containing the pitch element to change the rate (or tempo) would cause a pipeline state change to PAUSED after that stack; that has been fixed in 1.5.91 (see #755012 [0]). But even then, in 1.5.91 and later, NleComposition would send segment events to its NleSources assuming that one source second is equal to one pipeline second. The resulting early EOS event (in the case of a source rate higher than 1.0) would cause it to switch stacks too early, causing confusion in the timeline and spectacularly messed up output. This patch fixes that by searching for rate-changing elements in GESTrackElements such as GESEffects. If such rate-changing elements are found, their final effect on the playing rate is stored in the corresponding NleObject as the 'media duration factor', named like this because the 'media duration', or source duration, of an NleObject can be computed by multiplying the duration with the media duration factor of that object and its parents (this is called the 'recursive media duration factor'). For example, a 4-second NleSource with an NleOperation with a media duration factor of 2.0 will have an 8-second media duration, which means that for playing 4 seconds in the pipeline, the seek event sent to it must span 8 seconds of media. (So, the 'duration' of an NleObject or GES object always refers to its duration in the timeline, not the media duration.) To summarize: * Rate-changing elements are registered in the GESEffectClass (pitch::tempo and pitch::rate are registered by default); * GESTimelineElement is responsible for detecting rate-changing elements and computing the media_duration_factor; * GESTrackElement is responsible for storing the media_duration_factor in NleObject; * NleComposition is responsible for the recursive_media_duration_factor; * The latter property finally fixes media time computations in NleObject. NLE and GES tests are included. [0] https://bugzilla.gnome.org/show_bug.cgi?id=755012 Differential Revision: https://phabricator.freedesktop.org/D276
2015-12-20 13:03:57 +00:00
return s;
}
GST_CHECK_MAIN (gnonlin)
Reference in a new issue Copy permalink