Original commit message from CVS:
* gst/rtsp/gstrtspsrc.c: (gst_rtspsrc_base_init),
(gst_rtspsrc_stream_configure_transport), (gst_rtspsrc_push_event),
(gst_rtspsrc_loop), (gst_rtspsrc_uri_set_uri):
* gst/rtsp/gstrtspsrc.h:
Export sometimes source pad with correct caps on the template, create
the ghostpad from the template.
Remove RTCP template as we never expose RTCP.
Protect against invalid body size.
Avoid memcpy when creating the output buffer.
Properly post an error and send EOS when the loop function is shut down.
Original commit message from CVS:
Based on patch by: Lutz Mueller <lutz at topfrose dot de>
* gst/rtsp/gstrtspsrc.c: (gst_rtspsrc_class_init),
(gst_rtspsrc_init), (gst_rtspsrc_set_property), (gst_rtspsrc_open),
(gst_rtspsrc_uri_get_uri), (gst_rtspsrc_uri_set_uri):
* gst/rtsp/gstrtspsrc.h:
Make sure we can never set an invalid location.
* gst/rtsp/rtspmessage.c: (rtsp_message_steal_body):
* gst/rtsp/rtspmessage.h:
Added _steal_body method for future use.
* gst/rtsp/rtspurl.c: (rtsp_url_parse), (rtsp_url_free):
Make freeing of NULL url return immediatly.
Original commit message from CVS:
Based on patch by: Lutz Mueller <lutz at topfrose dot de>
* gst/rtsp/gstrtspsrc.c: (_do_init), (gst_rtspsrc_class_init),
(gst_rtspsrc_init), (gst_rtspsrc_stream_setup_rtp),
(gst_rtspsrc_stream_configure_transport), (gst_rtspsrc_play),
(gst_rtspsrc_change_state):
* gst/rtsp/gstrtspsrc.h:
Use boilerplate.
Make rtspsrc subclass GstBin to make state changes easier.
Add Range header field on the PLAY request.
Original commit message from CVS:
Based on patch by: Thijs Vermeir <thijs dot vermeir at barco dot com>
* gst/rtsp/gstrtspsrc.c: (gst_rtspsrc_set_state),
(gst_rtspsrc_media_to_caps), (gst_rtspsrc_stream_setup_rtp),
(gst_rtspsrc_stream_configure_transport), (gst_rtspsrc_open),
(gst_rtspsrc_close), (gst_rtspsrc_play), (gst_rtspsrc_pause):
* gst/rtsp/rtspconnection.c: (inet_aton):
Small cleanups.
when multicast is selected as the transport, create UDP sources and
connect to the multicast group.
Move parsing and setting of caps to a common place.
Fixes#349894.
Original commit message from CVS:
Patch by: Yves Lefebvre <ivanohe at abacom dot com>
* gst/avi/gstavimux.c: (gst_avi_mux_stop_file):
Correctly set the dwLength in strh.
With this patch, the file duration is now displayed correctly in window
media player and the AVI plays completely. Fixes#356147
Original commit message from CVS:
Patch by: Darren Kenny <darren dot kenny at sun dot com>
* sys/sunaudio/gstsunaudiomixerctrl.c:
(gst_sunaudiomixer_ctrl_build_list):
Set the output track as the MASTER so that the gnome-settings-daemon
keybindings for changing the volume using the keyboard works.
Fixes#356142.
Original commit message from CVS:
* gst/multipart/multipartdemux.c: (gst_multipart_demux_chain):
Fix documentation, it is not possible to control the framerate of jpegdec
using filtered caps yet. Fixes#355210.
Return the downstream GstFlowReturn instead of GST_FLOW_OK so that we
stop when there is an error.
Original commit message from CVS:
* gst/apetag/gsttagdemux.c: (gst_tag_demux_chain_parse_tag):
* gst/id3demux/gstid3demux.c: (gst_id3demux_chain):
Don't interpret a first buffer with an offset of NONE as
'from the middle of the stream', but only a first buffer
that has a valid buffer offset that's non-zero (see #345449).
Original commit message from CVS:
* gst/icydemux/gsticydemux.c: (gst_icydemux_reset),
(gst_icydemux_typefind_or_forward):
* gst/icydemux/gsticydemux.h:
When we merge/collect multiple incoming buffers for typefinding
purposes, keep an initial 0 offset on the first outgoing buffer
as well (otherwise id3demux won't work right). Fixes#345449.
Also Make buffer metadata writable before setting buffer caps.
* tests/check/elements/icydemux.c: (typefind_succeed),
(cleanup_icydemux), (push_data), (GST_START_TEST),
(icydemux_suite):
Small test case for the above.
Original commit message from CVS:
* gst/avi/gstavidemux.c: (gst_avi_demux_peek_chunk),
(gst_avi_demux_stream_index), (gst_avi_demux_sync),
(gst_avi_demux_stream_header_push),
(gst_avi_demux_process_next_entry), (gst_avi_demux_stream_data),
(gst_avi_demux_loop):
More code reuse and better logging in _peek_chunk(). Reintroduce check
for chunk sizes before reading them (avoid oom). Better handling for
invalid chunksizes when streaming.
Original commit message from CVS:
* gst/spectrum/gstspectrum.c: (gst_spectrum_class_init),
(gst_spectrum_start), (gst_spectrum_stop), (gst_spectrum_event):
Implements stop() to clear the adapter and event() to clear the
adapter on FLUSH_STOP and EOS.
Original commit message from CVS:
* gst/level/gstlevel.c: (gst_level_set_property):
* gst/level/gstlevel.h:
Fix type mixup in level->interval (gdouble<->guint64). Spotted by
René Stadler
Original commit message from CVS:
* gst/spectrum/gstspectrum.c: (gst_spectrum_init),
(gst_spectrum_set_property):
* gst/spectrum/gstspectrum.h:
Fix type mixup in spectrum->interval (gdouble<->guint64). Spotted by
René Stadler
Original commit message from CVS:
* gst/spectrum/demo-osssrc.c: (draw_spectrum), (main):
Use more defines
* gst/spectrum/gstspectrum.c: (gst_spectrum_init),
(gst_spectrum_dispose), (gst_spectrum_set_caps),
(gst_spectrum_transform_ip):
* gst/spectrum/gstspectrum.h:
Apply some of the spectrum cleanup changes suggested in #348085.
Original commit message from CVS:
* configure.ac:
Bump requirements of -base (videocrop test case needs this).
* gst/videocrop/gstvideocrop.c:
Document sloppy handling of subsampled chroma planes if
left/top cropping is an odd number.
* tests/check/elements/videocrop.c: (handoff_cb),
(videocrop_test_cropping_init_context),
(videocrop_test_cropping_deinit_context),
(videocrop_test_cropping), (check_1x1_buffer), (GST_START_TEST),
(videocrop_suite), (main):
Add another unit test that crops the input to 1x1 (and checks
that that pixel has the expected values in a number of formats).
Original commit message from CVS:
* gst/videocrop/Makefile.am:
* gst/videocrop/gstvideocrop.c: (gst_video_crop_class_init),
(gst_video_crop_transform_packed),
(gst_video_crop_transform_planar):
Some quick tests indicate that it doesn't make a great deal
of sense to use liboil here, at least not for the memcpy()s
we do, so remove liboil usage until there is clear evidence
it actually makes a positive difference somewhere.
Original commit message from CVS:
* gst/avi/gstavidemux.c: (gst_avi_demux_read_subindexes_pull),
(gst_avi_demux_sync), (gst_avi_demux_stream_header_push),
(gst_avi_demux_stream_data):
Revert one change to fix streaming avi (adapter size != data size).
Original commit message from CVS:
Patch by: Frédéric Riss <frederic.riss at gmail dot com>
* gst/matroska/matroska-demux.c: (gst_matroska_track_free),
(gst_matroska_demux_reset),
(gst_matroska_demux_read_track_encodings),
(gst_matroska_demux_add_stream), (gst_matroska_decode_buffer),
(gst_matroska_demux_parse_blockgroup_or_simpleblock),
(gst_matroska_demux_subtitle_caps):
* gst/matroska/matroska-ids.h:
Add support for VOBSUB subtitle tracks and zlib-compressed
tracks. Make sure we start on a keyframe after a seek. (#343348)
Original commit message from CVS:
* gst/matroska/matroska-demux.c: (gst_matroska_demux_push_hdr_buf),
(gst_matroska_demux_push_flac_codec_priv_data),
(gst_matroska_demux_push_xiph_codec_priv_data),
(gst_matroska_demux_parse_blockgroup_or_simpleblock),
(gst_matroska_demux_video_caps), (gst_matroska_demux_audio_caps):
* gst/matroska/matroska-ids.h:
Add basic FLAC support (#311586), not perfect yet though, needs some
tweaking in flacdec; also, seeking could be better.
Do better bounds checking when deserialising vorbis stream headers
to make sure we don't read beyond the end of the buffer on bad input.
Original commit message from CVS:
Patch by: Alessandro Decina <alessandro at nnva dot org>
* ext/annodex/gstcmmldec.c: (gst_cmml_dec_chain):
Seeking back in a file containing a CMML stream errors out if the seek
goes back up to the CMML headers. This is because after the seek the xml
processing instruction <?xml ...?> is submitted to the xml parser again,
which results in an error. The attached patch fixes the problem.
Fixes#353908.
* ext/annodex/gstcmmlenc.h:
Fix authors name.
Original commit message from CVS:
* tests/check/elements/videocrop.c: (handoff_cb),
(buffer_probe_cb), (test_caps_transform), (test_passthrough),
(notgst_value_list_get_nth_int), (videocrop_suite):
More tests: check passthrough mode and caps transform in
both directions with fixed values, ranges and lists.
Original commit message from CVS:
* configure.ac:
* gst/videocrop/Makefile.am:
* gst/videocrop/gstvideocrop.c: (gst_video_crop_base_init),
(gst_video_crop_class_init), (gst_video_crop_init),
(gst_video_crop_get_image_details_from_caps),
(gst_video_crop_get_unit_size), (gst_video_crop_transform_packed),
(gst_video_crop_transform_planar), (gst_video_crop_transform),
(gst_video_crop_transform_dimension),
(gst_video_crop_transform_dimension_value),
(gst_video_crop_transform_caps), (gst_video_crop_set_caps),
(gst_video_crop_set_property), (gst_video_crop_get_property),
(plugin_init):
Port/rewrite videocrop from scratch for GStreamer-0.10, and make
it support all formats videoscale supports (#345653).
Original commit message from CVS:
* gst/qtdemux/qtdemux.c:
(gst_qtdemux_do_seek):
Reset each streams last_flow to GST_FLOW_OK.
(gst_qtdemux_activate_segment):
Removing mystic modifications for good.
Original commit message from CVS:
* gst/qtdemux/qtdemux.c: (gst_qtdemux_activate_segment),
(qtdemux_parse_tree):
put back 'segment start<=stop' change that was mystically reverted by
the last commit
Original commit message from CVS:
* gst/qtdemux/qtdemux.c: (gst_qtdemux_activate_segment),
(gst_qtdemux_add_stream), (qtdemux_parse_trak),
(qtdemux_video_caps):
Make sure segment start<=stop in weird quicktime files.
Original commit message from CVS:
2006-08-28 Andy Wingo <wingo@pobox.com>
* ext/raw1394/gstdv1394src.c (gst_dv1394src_from_raw1394handle):
New helper function to lessen the ifdefs.
(GST_INFO_OBJECT):
(gst_dv1394src_iso_receive): Use it.
(gst_dv1394src_create): Also use the control sockets in iec61883
mode.
(gst_dv1394src_start, gst_dv1394src_stop): Always use a separate
handle for AVC operations; fixes#348233.
Original commit message from CVS:
* gst/avi/gstavidemux.c: (gst_avi_demux_class_init),
(gst_avi_demux_init), (gst_avi_demux_finalize),
(gst_avi_demux_reset), (gst_avi_demux_index_last),
(gst_avi_demux_index_next), (gst_avi_demux_index_entry_for_time),
(gst_avi_demux_parse_subindex), (gst_avi_demux_parse_index),
(gst_avi_demux_stream_index), (gst_avi_demux_peek_tag),
(gst_avi_demux_next_data_buffer), (gst_avi_demux_stream_scan),
(gst_avi_demux_massage_index),
(gst_avi_demux_calculate_durations_from_index),
(gst_avi_demux_stream_header_pull), (gst_avi_demux_do_seek),
(gst_avi_demux_process_next_entry), (gst_avi_demux_loop),
(gst_avi_demux_chain), (gst_avi_demux_sink_activate),
(gst_avi_demux_change_state):
* gst/avi/gstavidemux.h:
More attempts to turn this into readable code.
Don't leak adapters.
Calculate duration according to index more efficiently.
Don't try to act like we drive the pipeline in chain mode.
Original commit message from CVS:
Patch by: Alessandro Decina <alessandro at nnva dot org>
* ext/annodex/gstannodex.c: (gst_annodex_granule_to_time):
Do some extra sanity checks.
Fixes#350340.
* ext/annodex/gstcmmlenc.c: (gst_cmml_enc_change_state),
(gst_cmml_enc_parse_tag_head), (gst_cmml_enc_parse_tag_clip),
(gst_cmml_enc_push_clip), (gst_cmml_enc_push):
Check if clip->start_time is valid before adding the clip to the
track list.
Reset enc->preamble going from PAUSED to READY.
Don't use GST_FLOW_UNEXPECTED for wrong usage of the element, it is
only used for EOS.
Only post an error message if we were the one that created the fatal
GstFlowReturn value.
* ext/annodex/gstcmmlutils.c: (gst_cmml_clock_time_from_npt),
(gst_cmml_clock_time_to_granule), (gst_cmml_track_list_has_clip):
Parse the seconds field of the npt-sec time format using %llu rather than
%d and check that the value scaled by GST_SECOND doesn't overflow.
Use guint64(s) to represent the keyindex and keyoffset fields of a granulepos.
Lookup a clip's track with clip->track rather than clip->id which
makes no sense.
Identify a clip by its track and start time and not its xml id.
do some more input checking and make sure we don't do undefined shifts.
* tests/check/elements/cmmldec.c: (setup_cmmldec),
(teardown_cmmldec), (check_output_buffer_is_equal), (push_data),
(cmml_tag_message_pop), (check_headers), (push_clip_full),
(push_clip), (push_empty_clip), (check_output_clip),
(GST_START_TEST), (cmmldec_suite):
* tests/check/elements/cmmlenc.c: (setup_cmmlenc),
(teardown_cmmlenc), (check_output_buffer_is_equal), (push_data),
(check_headers), (push_clip), (check_clip_times), (check_clip),
(check_empty_clip), (GST_START_TEST), (cmmlenc_suite):
Added some more checks.
Original commit message from CVS:
* gst/audiofxgood/audiopanorama.c: (gst_audio_panorama_class_init),
(gst_audio_panorama_set_property),
(gst_audio_panorama_get_property),
(gst_audio_panorama_transform_m2s_int),
(gst_audio_panorama_transform_s2s_int),
(gst_audio_panorama_transform_m2s_float),
(gst_audio_panorama_transform_s2s_float):
* gst/audiofxgood/audiopanorama.h:
* tests/check/elements/audiopanorama.c: (GST_START_TEST):
Make also the pan-property float (saves scaling and yields better
resolution)
Original commit message from CVS:
* gst/audiofxgood/audiopanorama.c: (gst_audio_panorama_set_caps),
(gst_audio_panorama_transform_m2s_float),
(gst_audio_panorama_transform_s2s_float):
ChangeLog surgery to add cymax's real name
