mirrors/gstreamer
mirrors/gstreamer
1
1
Fork 0
mirror of https://gitlab.freedesktop.org/gstreamer/gstreamer.git synced 2024-11-23 18:21:04 +00:00
Code Issues 1 Projects Releases Wiki Activity

docs: improve decodebin2 docs a little

Browse source 
Mention that new pads may be created even after no-more-pads.

https://bugzilla.gnome.org/show_bug.cgi?id=634584
This commit is contained in:
Tim-Philipp Müller 2010-11-22 00:54:35 +00:00
parent 8fb2c27ed0
commit 1bfefc50af
1 changed files with 15 additions and 11 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

26
gst/playback/gstdecodebin2.c
View file

@ -33,8 +33,9 @@
*
* The following section describes how decodebin2 works internally.
*
* The first part of decodebin2 is it's typefind element, which tries
* to get the type of the input stream. If the type is found autoplugging starts.
* The first part of decodebin2 is its typefind element, which tries
* to determine the media type of the input stream. If the type is found
* autoplugging starts.
*
* decodebin2 internally organizes the elements it autoplugged into GstDecodeChains
* and GstDecodeGroups. A decode chain is a single chain of decoding, this
@ -47,23 +48,27 @@
* A decode group combines a number of chains that are created by a
* demuxer element. All those chains are connected through a multiqueue to
* the demuxer. A new group for the same demuxer is only created if the
* demuxer has signaled no-more pads, in which case all following pads
* demuxer has signaled no-more-pads, in which case all following pads
* create a new chain in the new group.
*
* This continues until the top-level decode chain is complete. A decode
* chain is complete if it either ends with a blocked endpad, if autoplugging
* stopped because no suitable plugins could be found or if the active group
* is complete. A decode group OTOH is complete if all child chains are complete.
* is complete. A decode group on the other hand is complete if all child
* chains are complete.
*
* If this happens at some point, all endpads of all active groups are exposed.
* For this decodebin2 adds the endpads, signals no-more-pads and then unblocks
* them. Now playback starts.
*
* If one of the chains that end on a endpad receives EOS decodebin2 checks upwards
* via the parent pointers if all chains and groups are drained. In that case
* everything goes into EOS.
* If there is a chain where the active group is drained but there exist next groups
* the active group is hidden (endpads are removed) and the next group is exposed.
* If one of the chains that end on a endpad receives EOS decodebin2 checks
* if all chains and groups are drained. In that case everything goes into EOS.
* If there is a chain where the active group is drained but there exist next
* groups, the active group is hidden (endpads are removed) and the next group
* is exposed. This means that in some cases more pads may be created even
* after the initial no-more-pads signal. This happens for example with
* so-called "chained oggs", most commonly found among ogg/vorbis internet
* radio streams.
*
* Note 1: If we're talking about blocked endpads this really means that the
* *target* pads of the endpads are blocked. Pads that are exposed to the outside
@ -3416,8 +3421,7 @@ gst_decode_pad_new (GstDecodeBin * dbin, GstPad * pad, GstDecodeChain * chain)
pad_tmpl = gst_static_pad_template_get (&decoder_bin_src_template);
dpad =
g_object_new (GST_TYPE_DECODE_PAD, "direction", GST_PAD_DIRECTION (pad),
"template", pad_tmpl,
NULL);
"template", pad_tmpl, NULL);
gst_ghost_pad_construct (GST_GHOST_PAD_CAST (dpad));
gst_ghost_pad_set_target (GST_GHOST_PAD_CAST (dpad), pad);
dpad->chain = chain;