Original commit message from CVS:
Reviewed by: Ronald S. Bultje <rbultje@ronald.bitfreak.net>
* ext/libpng/gstpngdec.c: (gst_pngdec_init), (gst_pngdec_chain):
* ext/libpng/gstpngenc.c:
Fix byte-order, use proper fixed caps. Fixes#164197.
Original commit message from CVS:
Include "_stdint.h" instead of <stdint.h>. Fixes build on systems that do not have stdint.h, like Solaris 9 (fixes#166631).
Original commit message from CVS:
* sys/ximage/ximagesink.c: (gst_ximagesink_change_state):
* sys/xvimage/xvimagesink.c: (gst_xvimagesink_xwindow_clear),
(gst_xvimagesink_change_state):
Clear window on PAUSED->READY instead of READY->PAUSED. Stop
Xv video (and thereby regenerate Xv colourkey) in clear() so
that PLAY -> READY -> PLAY works (fixes#162504).
Original commit message from CVS:
* sys/v4l/gstv4lmjpegsrc.c: (gst_v4lmjpegsrc_getcaps):
Switch to list instead of range, since MJPEG-devices really just
support decimations, not any size.
Original commit message from CVS:
* ext/mpeg2dec/gstmpeg2dec.c: (gst_mpeg2dec_open_decoder),
(gst_mpeg2dec_reset), (free_all_buffers),
(gst_mpeg2dec_alloc_buffer), (handle_sequence):
* ext/mpeg2dec/gstmpeg2dec.h:
The libmpeg2 user-allocated buffer management is awkward,
to say the least. Hopefully this fixes things.
Original commit message from CVS:
2005-02-04 Andy Wingo <wingo@pobox.com>
* gst/audioconvert/bufferframesconvert.c
(buffer_frames_convert_fixate): New function, fixates to 256
frames per buffer by default. (Much better than 1.)
(buffer_frames_convert_init): Set the fixate function for both src
and sink pad.
(buffer_frames_convert_link): After success setting nonfixed caps,
get the negotiated caps so we can know how many buffer-frames it
will be. No idea how this worked at all before.
Original commit message from CVS:
* ext/mpeg2dec/gstmpeg2dec.c: (gst_mpeg2dec_init),
(gst_mpeg2dec_close_decoder), (put_buffer), (check_buffer),
(free_buffer), (free_all_buffers), (gst_mpeg2dec_alloc_buffer),
(handle_sequence), (handle_picture):
* ext/mpeg2dec/gstmpeg2dec.h:
Rearrange buffer tracking and refcounting and refactor
a little for readability.
Original commit message from CVS:
* sys/qcam/gstqcamsrc.c: (gst_qcamsrc_change_state),
(gst_qcamsrc_open):
Use GST_ELEMENT_ERROR, not g_warning, if open failed.
Original commit message from CVS:
* gst/tcp/gsttcpclientsink.c: (gst_tcpclientsink_class_init),
(gst_tcpclientsink_finalize):
* gst/tcp/gsttcpclientsrc.c: (gst_tcpclientsrc_class_init),
(gst_tcpclientsrc_finalize):
* gst/tcp/gsttcpserversink.c: (gst_tcpserversink_class_init),
(gst_tcpserversink_init), (gst_tcpserversink_finalize):
* gst/tcp/gsttcpserversrc.c: (gst_tcpserversrc_class_init),
(gst_tcpserversrc_init), (gst_tcpserversrc_finalize):
Don't leak the hostname when shutting down.
In tcpserversrc, take a copy of the default hostname.
Original commit message from CVS:
* ext/mpeg2dec/gstmpeg2dec.c:
Don't send things to NULL PAD_PEERs
* gst/deinterlace/gstdeinterlace.c: (gst_deinterlace_chain):
Copy-on-write the incoming buffer.
* gst/mpegstream/gstdvddemux.h:
* gst/mpegstream/gstmpegclock.h:
* gst/mpegstream/gstmpegdemux.c: (gst_mpeg_demux_parse_syshead),
(normal_seek), (gst_mpeg_demux_handle_src_event):
* gst/mpegstream/gstmpegdemux.h:
* gst/mpegstream/gstmpegpacketize.h:
* gst/mpegstream/gstmpegparse.c:
(gst_mpeg_parse_update_streaminfo), (gst_mpeg_parse_reset),
(gst_mpeg_parse_handle_discont), (gst_mpeg_parse_parse_packhead),
(gst_mpeg_parse_loop), (gst_mpeg_parse_get_rate),
(gst_mpeg_parse_convert_src), (gst_mpeg_parse_handle_src_query),
(gst_mpeg_parse_handle_src_event), (gst_mpeg_parse_change_state):
* gst/mpegstream/gstmpegparse.h:
* gst/mpegstream/gstrfc2250enc.h:
Various changes to the way time is computed that make seeking and
total time estimation much better here.
Use G_BEGIN/END_DECLS instead of __cplusplus
* gst/videocrop/gstvideocrop.c: (gst_video_crop_chain):
Use gst_buffer_stamp instead of only copying the TIMESTAMP
Original commit message from CVS:
* ext/polyp/polypsink.c: (gst_polypsink_base_init),
(create_context), (gst_polypsink_link): Fix silly endianness
bug. Add some debugging. Remove float from caps; it doesn't
work. Attempt to get remote audio working.
Original commit message from CVS:
* gst/ffmpegcolorspace/gstffmpegcolorspace.c:
(gst_ffmpegcsp_chain):
D'oh, reference the palette data, not the palette structure.
Fixes color distortion in #132341.
Original commit message from CVS:
* gst/videoscale/gstvideoscale.c: (gst_videoscale_link):
PAR can be non-fixed when not provided as argument (#162626).
Original commit message from CVS:
Reviewed by: Ronald S. Bultje <rbultje@ronald.bitfreak.net>
* gst/qtdemux/qtdemux.c: (gst_qtdemux_change_state),
(gst_qtdemux_loop_header):
Re-apply patch from #142272 that allows non-seekable sources,
re-proposed by Daniel Drake <dsd@gentoo.org>.
Original commit message from CVS:
2005-01-28 Andy Wingo <wingo@pobox.com>
* ext/dv/gstdvdec.c: Change the pixel aspect ratio of dvdec output
to reflect a different dubious internet source. Add a reference
and some commentary.
Original commit message from CVS:
* gst/playback/gststreamselector.c: (gst_stream_selector_init),
(gst_stream_selector_get_caps), (gst_stream_selector_chain):
* gst/playback/gststreamselector.h:
Be more selective when we're redoing caps negotiation from
within the chain function on a stream change.
Original commit message from CVS:
* gst/matroska/matroska-demux.c:
(gst_matroska_demux_parse_blockgroup):
Fix logic error in timing of subtitle stream synchronization.
* gst/typefind/gsttypefindfunctions.c: (qt_type_find):
Add skip-chunk, which is found in kodak-camera streams.
Original commit message from CVS:
* ext/mad/Makefile.am:
* ext/mad/gstid3demuxbin.c: (gst_id3demux_bin_get_type),
(gst_id3demux_bin_base_init), (gst_id3demux_bin_class_init),
(gst_id3demux_bin_init), (gst_id3demux_bin_remove_pad),
(found_type), (gst_id3demux_bin_change_state):
* ext/mad/gstid3tag.c: (gst_id3_tag_add_src_pad),
(gst_id3_tag_init), (gst_id3_tag_handle_event),
(gst_id3_tag_src_link), (gst_id3_tag_chain),
(gst_id3_tag_change_state), (plugin_init):
* ext/mad/gstmad.h:
Add id3demuxbin (which is a simple bin consisting of id3demux
and typefind), take over rank from id3demux, remove typefind
code from id3demux. Makes all broken mp3s that I know of work,
and thereby fixes#152688.
Original commit message from CVS:
Reviewed by: Ronald S. Bultje <rbultje@ronald.bitfreak.net>
* ext/mad/gstmad.c: (gst_mad_src_event):
* gst/avi/gstavidemux.c: (gst_avi_demux_handle_src_event):
Allow seeks on audio pad, make mad forward those (#164826).
* gst/audioscale/gstaudioscale.c: (gst_audioscale_chain):
Set duration (#165335).
Original commit message from CVS:
* gst/asfdemux/gstasfdemux.c: (gst_asf_demux_init),
(gst_asf_demux_commit_taglist), (gst_asf_demux_process_comment),
(gst_asf_demux_process_ext_content_desc),
(gst_asf_demux_change_state), (gst_asf_demux_add_audio_stream),
(gst_asf_demux_add_video_stream), (gst_asf_demux_setup_pad):
* gst/asfdemux/gstasfdemux.h:
Improve metadata display, e.g. if the metadata comes before the
streams are loaded (which is perfectly valid).
Original commit message from CVS:
2005-01-26 Amaury Jacquot <sxpert@esitcom.org>
* ext/cairo/gsttextoverlay.c: include string.h and strings.h to fix
build failure on amd64
Original commit message from CVS:
Check environment variables GST_ID3V2_TAG_ENCODING,
GST_ID3_TAG_ENCODING and GST_TAG_ENCODING for a colon-separated
list of character encodings to force interpretation of non-unicode
strings stored in an ID3v2 tag to a particular encoding. If none
is specified, try to use current locale's encoding, then fall back
to ISO-8859-1 (which will always succeed). (Resolves#149274)
Check environment variables GST_ID3V1_TAG_ENCODING,
GST_ID3_TAG_ENCODING and GST_TAG_ENCODING for a colon-separated
list of character encodings to use in case a string encountered
in an ID3v1 tag is not valid UTF-8 already. If no encoding is
specified, try to use the current locale's encoding, then fall
back to ISO-8859-1 (which will always succeed).
Original commit message from CVS:
* ext/mad/gstmad.c: (gst_mad_check_caps_reset), (gst_mad_chain):
- on half framerate, compute the rate in advance so the comparisons
don't compare wrong values
- don't use mad_synth/frame_mute anymore, this mirrors mad_decoder
behaviour
- don't use mad_header_decode anymore, mad_frame_decode does that
automatically
- when getting rid of consumed bytes, reset the stream's skiplen
(fixes#163867)
