This ensures smoother playback. It looks weird if we first do a big
jump, then play a couple of consecutive frames, just to again skip ahead
quite a bit because we ran late again.
When dealing with key-unit trick mode downloads, the goal is to
provide the best "Quality of Experience". This is achieved by:
1) maximizing the number of frames displayed per second
2) avoiding "stalling" as much as possible (i.e. not downloading and
decoding frames fast enough)
This implementation achives this by:
1) Knowing very precisely the current keyframe being download (i.e
more accurate than at the fragment level which might contain more
than one keyfram). This is the new "actual_position" variable
introduced by this commit
2) Knowing the position of downstream (provided by QoS and stored
in the adaptivedemuxstream qos_earliest_time variable)
3) Knowing how long it takes to request and fully download a keyframe
(the average_download_time variable)
Taking those 3 variables into account, whenever a keyframe has been
pushed downstream we calculate a "target time" (target_time variable)
which is the ideal next keyframe time to request so that:
1) It will be requested/downloaded/demuxed/decoded in time to be
displayed without being too late
2) It will not be too far ahead that it would cause too few frames
per second to be displayed.
How far ahead we will request is inversily proportional to how close
the actual position (actual_position) is from the downstream
position (qos_earliest_time). The more is buffered between the source
and the sink, the "closer" the target time will be, and therefore
the more frames per seconds will be displayed (up to the limit
of keyframes_per_second * absolute_rate).
The previous code was handling both as separate steps and then tried to
combine the results, but this resulted in all kinds of bugs which showed
themselves as failures during seeking and offset tracking getting wrong.
This also showed itself with gst-validate on the sample stream.
The rewritten code now parses everything in one go and tracks the
current offset only once, and as a side effect simplifies the code a
lot.
Also added is detection of SIDX that point to other SIDX instead of
actual media segments, e.g. with this stream:
http://dash.akamaized.net/dash264/TestCases/1a/sony/SNE_DASH_SD_CASE1A_REVISED.mpd
Support for this will have to be added at some point but that should
also be easier with the rewritten code.
https://bugzilla.gnome.org/show_bug.cgi?id=781233
This reverts commit c9fbf3459a.
The representation ID comparision here was wrong and triggering always
if the ID did *not* change, causing needless redownloading of the
header. The sample stream provided in the bug does not exist anymore.
There is no guarantee that the index positions are the same between
representations, and assuming this easily causes us to get into invalid
index positions.
We don't have to do yet another additional request but can just download the
data directly.
Also unify the key-unit only mode buffer pushing and extract it into its own
function now that it became more complicated.
https://bugzilla.gnome.org/show_bug.cgi?id=741104
This simplifies the code but also removes a bug with tracking of the remaining
size for the initial subfragment: we were not considering the size between the
index and the start of the first moof here.
https://bugzilla.gnome.org/show_bug.cgi?id=764684
If MPD@suggestedPresentationDelay is not present in the manifest,
dashdemux selects the fragment closest to the most recently generated
fragment. This causes a playback issue because this choice does not allow
the DASH client to build up any buffer of downloaded fragments without
pausing playback. This is because by definition new fragments appear on
the server in real-time (e.g. if segment duration is 4 seconds, a new
fragment will appear on the server every 4 seconds). If the starting
playback position was n*segmentDuration seconds behind "now", the DASH
client could download up to 'n' fragments faster than realtime before it
reached the point where it needed to wait for fragments to appear on the
server.
The MPD@suggestedPresentationDelay attribute allows a content publisher
to provide a suggested starting position that is behind the current
"live" position.
If the MPD@suggestedPresentationDelay attribute is not present, provide
a suitable default value as a property of the dashdemux element. To
allow the default presentation delay to be specified either using
fragments or seconds, the property is a string that contains a number
and a unit (e.g. "10 seconds", "4 fragments", "2500ms").
Create src pads for Representations that contain timed-text subtitles,
both when the subtitles are encapsulated in ISO BMFF (i.e., the
Representation has mimeType "application/mp4") and when they are
unencapsulated (i.e., the Representation has mimeType
"application/ttml+xml").
https://bugzilla.gnome.org/show_bug.cgi?id=747774
Unless the DASH client can compensate for the difference between its
clock and the clock used by the server, the client might request
fragments that either not yet on the server or fragments that have
already been expired from the server. This is an issue because these
requests can propagate all the way back to the origin
ISO/IEC 23009-1:2014/Amd 1 [PDAM1] defines a new UTCTiming element to allow
a DASH client to track the clock used by the server generating the
DASH stream. Multiple UTCTiming elements might be present, to indicate
support for multiple methods of UTC time gathering. Each element can
contain a white space separated list of URLs that can be contacted
to discover the UTC time from the server's perspective.
This commit provides parsing of UTCTiming elements, unit tests of this
parsing and a function to poll a time server. This function
supports the following methods:
urn:mpeg:dash:utc:ntp:2014
urn:mpeg:dash:utc:http-xsdate:2014
urn:mpeg:dash:utc:http-iso:2014
urn:mpeg:dash:utc:http-ntp:2014
The manifest update task is used to poll the clock time server,
to save having to create a new thread.
When choosing the starting fragment number and when waiting for a
fragment to become available, the difference between the server's idea
of UTC and the client's idea of UTC is taken into account. For example,
if the server's time is behind the client's idea of UTC, we wait for
longer before requesting a fragment
[PDAM1]: http://www.iso.org/iso/home/store/catalogue_tc/catalogue_detail.htm?csnumber=66068
dashdemux: support NTP time servers in UTCTiming elements
Use the gst_ntp_clock to support the use of an NTP server.
https://bugzilla.gnome.org/show_bug.cgi?id=752413
Bitrate-limit is already available in the baseclass and, even though
the bandwidth-usage name is better, hls and mss already used
bitrate-limit. This patch deprecates the bandwidth-usage and maps
it to the baseclass bitrate-limite.
Add more power to the chunk_received function (renamed to data_received)
and also to the fragment_finish function.
The data_received function must parse/decrypt the data if necessary and
also push it using the new push_buffer function that is exposed now. The
default implementation gets data from the stream adapter (all available)
and pushes it.
The fragment_finish function must also advance the fragment. The default
implementation only advances the fragment.
This allows the subsegment handling in dashdemux to continuously download
the same file from the server instead of stopping at every subsegment
boundary and starting a new request
The ISOBMFF profile allows definind subsegments in a segment. At those
subsegment boundaries the client can switch from one representation to
another as they have aligned indexes.
To handle those the 'sidx' index is parsed from the stream and the
entries point to pts/offset of the samples in the stream. Knowing that
the entries are aligned in the different representation allows the client
to switch mid fragment. In this profile a single fragment is used per
representation and the subsegments are contained in this fragment.
To notify the superclass about the subsegment boundary the chunk_received
function returns a special flow return that indicates that. In this case,
the super class will check if a more suitable bitrate is available and will
change to the same subsegment in this new representation.
It also requires special handling of the position in the stream as the
fragment advancing is now done by incrementing the index of the subsegment.
It will only advance to the next fragment once all subsegments have been
downloaded.
https://bugzilla.gnome.org/show_bug.cgi?id=741248
If EOS or ERROR happens before the download loop thread has reached its
g_cond_wait() call, then the g_cond_signal doesn't have any effect and
the download loop thread stucks later.
https://bugzilla.gnome.org/show_bug.cgi?id=735663
Set up a message handling function to be able to catch errors
from the source element and signal the cond to allow the download
loop to retry the download.
Instead, use a source element linked to a ghostpad to provide
smaller buffers and more granular control for downstream
buffering elements while also reducing startup latency
Download and push from the same task, makes code a lot simpler
to maintain. Also pushing from separate threads avoids deadlocking
when gst_pad_push blocks due to downstream queues being full.
Use a single lock for all streams instead of having separate locks.
This makes maintenance easier and at most points we would need
a single lock before iterating on all streams data. So not much
is gained from individual locks.
Handle multiple languages by using the not-linked return to stop
the download task for that stream. It can be reactivated when
a reconfigure event is received. Stopping the unused streams is
relevant to save network bandwidth
Instead of having a single download task for all streams, this
commit makes each stream have its own download loop, allowing
parallel download of fragments.
