Original commit message from CVS:
2004-07-21 Julien MOUTTE <julien@moutte.net>
* sys/ximage/ximagesink.c: (gst_ximagesink_ximage_new),
(gst_ximagesink_renegotiate_size), (gst_ximagesink_sink_link),
(gst_ximagesink_chain), (gst_ximagesink_set_xwindow_id): Optimize
images creation for both elements. We don't create the image on caps
nego or renego, we just destroy the internal one if present if it does
not match the needs. The chain function takes care of creating a new
image when needed.
* sys/xvimage/xvimagesink.c: (gst_xvimagesink_xvimage_new),
(gst_xvimagesink_xwindow_decorate), (gst_xvimagesink_sink_link),
(gst_xvimagesink_chain), (gst_xvimagesink_buffer_alloc),
(gst_xvimagesink_set_xwindow_id): Additionally xvimage now contains
the image format information. The buffer pool checks for the context
image format and discard images with different formats.
* sys/xvimage/xvimagesink.h: Adding im_format in the xvimage structure.
Original commit message from CVS:
* gst-libs/gst/riff/riff-media.c:
(gst_riff_create_audio_caps_with_data):
Fix double end-to-native symbol conversion (#148021).
Original commit message from CVS:
* gst/tcp/gstmultifdsink.c: (gst_multifdsink_class_init),
(gst_multifdsink_add), (gst_multifdsink_get_stats),
(gst_multifdsink_client_remove),
(gst_multifdsink_handle_client_write),
(gst_multifdsink_queue_buffer):
* gst/tcp/gstmultifdsink.h:
More multifdsink stats. Avoid deadlock by releasing locks
before sending out a signal.
Original commit message from CVS:
* gst/tcp/gstmultifdsink.c: (gst_multifdsink_class_init),
(gst_multifdsink_init), (gst_multifdsink_add),
(gst_multifdsink_client_remove),
(gst_multifdsink_handle_client_write),
(gst_multifdsink_queue_buffer), (gst_multifdsink_chain),
(gst_multifdsink_set_property), (gst_multifdsink_get_property),
(gst_multifdsink_init_send):
* gst/tcp/gstmultifdsink.h:
Added more stats, added timeout for a client, fixed some typos
and added some comments.
Original commit message from CVS:
2004-07-16 Julien MOUTTE <julien@moutte.net>
* sys/ximage/ximagesink.c: (gst_ximagesink_handle_xerror),
(gst_ximagesink_check_xshm_calls), (gst_ximagesink_ximage_new),
(gst_ximagesink_ximage_destroy), (gst_ximagesink_sink_link),
(gst_ximagesink_chain), (gst_ximagesink_buffer_free),
(gst_ximagesink_buffer_alloc):
* sys/ximage/ximagesink.h:
* sys/xvimage/xvimagesink.c: (gst_xvimagesink_check_xshm_calls),
(gst_xvimagesink_xvimage_new), (gst_xvimagesink_xvimage_destroy),
(gst_xvimagesink_chain), (gst_xvimagesink_buffer_free),
(gst_xvimagesink_buffer_alloc):
* sys/xvimage/xvimagesink.h: Getting the 2 video sinks synchronized
again. Using internal data pointer of the x(v)image to store image's
data to be coherent with the buffer alloc mechanism. Investigated the
image destruction code to be sure that everything gets freed correctly.
Original commit message from CVS:
* gst-libs/gst/video/video.h:
Added 32 bits RGBA. Not sure if we should use another mime-type
for alpha rgb. Currently the presence of the alpha_mask property
signals an alpha channel. Ronald?
Original commit message from CVS:
* sys/ximage/ximagesink.c: (gst_ximagesink_xcontext_get):
* sys/xvimage/xvimagesink.c: (gst_xvimagesink_get_xv_support):
FPS seems to be 0.0 to MAX everywhere else.
Original commit message from CVS:
* gst-libs/gst/riff/riff-media.c:
(gst_riff_create_video_caps_with_data):
mp42/mp43 (no caps) exist too.
* gst/matroska/matroska-demux.c: (gst_matroska_demux_video_caps):
Set pixel_width/height; we've got them in-caps.
* gst/typefind/gsttypefindfunctions.c: (plugin_init):
* gst/wavparse/gstwavparse.c: (plugin_init):
Both are valid primary.
* sys/oss/gstossmixer.c:
Remove i18n hack and enable translations.
Original commit message from CVS:
* sys/xvimage/xvimagesink.c: (gst_xvimagesink_check_xshm_calls),
(gst_xvimagesink_xvimage_new), (gst_xvimagesink_xvimage_destroy):
fix for non-shm xv. Original patch by Tim Ringenbach (fixes#147248)
Original commit message from CVS:
* ext/alsa/gstalsa.c: (gst_alsa_open_audio),
(gst_alsa_sw_params_dump), (gst_alsa_hw_params_dump),
(gst_alsa_close_audio):
disable some of the debugging code for now. Writing debugging to a
buffer is broken in current alsalib releases.
Original commit message from CVS:
* ext/ogg/gstoggdemux.c: (gst_ogg_demux_seek_before):
When trying to find the stream length, seek back N pages
instead of just one, where N is the number of streams in
the current chain.
Original commit message from CVS:
* ext/gdk_pixbuf/gstgdkpixbuf.c: (gst_gdk_pixbuf_get_capslist), (gst_gdk_pixbuf_chain):
Add svg and pcx to template caps, and ensure that getcaps returns a subset
of the template caps.
Copy each row manually for output, as gdkpixbuf may pad the
rowstride to a 32-bit word boundary.
Original commit message from CVS:
* gst-libs/gst/riff/riff-media.c: (gst_riff_create_audio_caps),
(gst_riff_create_video_template_caps):
Fix the template caps to include some more media types.
Original commit message from CVS:
2004-07-11 Andy Wingo <wingo@pobox.com>
* gst/audioconvert/gstaudioconvert.c (gst_audio_convert_link): For
float, "any" caps -> buffer_frames=[0,MAX].
* gst/interleave/interleave.c (interleave_getcaps): Seems the core
doesn't intersect our caps with the template any more. Do it
ourselves.
(interleave_buffered_loop): Use g_newa instead of malloc/free.
Original commit message from CVS:
* gst/mpegstream/gstmpegdemux.c: (gst_mpeg_demux_new_output_pad),
(gst_mpeg_demux_parse_packet), (gst_mpeg_demux_process_private):
Add pad to element *after* setting the pad functions so that
the scheduler can use the correct ones.
Original commit message from CVS:
* ext/theora/theoradec.c: (theora_dec_from_granulepos),
(theora_dec_src_query), (theora_dec_src_event), (theora_dec_chain):
Sync to keyframe after seek
Original commit message from CVS:
* gst-libs/gst/play/play.c: (gst_play_pipeline_setup),
(gst_play_get_length_callback), (gst_play_set_location),
(gst_play_seek_to_time), (gst_play_set_data_src),
(gst_play_set_video_sink), (gst_play_set_audio_sink),
(gst_play_set_visualization), (gst_play_connect_visualization),
(gst_play_get_sink_element):
- add debugging info
- fix looking up sink elements by iterating over complete caps
- put everything except for source and autoplugger in a complete bin
Original commit message from CVS:
* gst/videobox/gstvideobox.c: (gst_video_box_fill_get_type),
(gst_video_box_class_init), (gst_video_box_set_property),
(gst_video_box_i420), (gst_video_box_ayuv), (gst_video_box_chain):
Use pad_alloc where possible.
Original commit message from CVS:
* ext/theora/theora.c: (plugin_init):
* ext/theora/theoradec.c: (theora_dec_from_granulepos),
(theora_dec_src_query), (theora_dec_chain):
* ext/theora/theoraenc.c: (gst_theora_enc_class_init),
(theora_enc_sink_link), (theora_buffer_from_packet),
(theora_push_packet), (theora_enc_chain):
Fix theora granulepos calculation.
Fix overflow in duration/position calculation.
Bump rank to PRIMARY for theoradec.
Use granulepos of last packet to calculate position.
Set keyframe flag on buffers when needed.
Original commit message from CVS:
* ext/alsa/gstalsa.c: (gst_alsa_xrun_recovery):
only restart audio when we indeed have an xrun to fix repeated
xruns. Fix suggested by Giuliano Pochini.
Original commit message from CVS:
* ext/alsa/gstalsaplugin.c: (gst_alsa_error_wrapper): Disable
call to gst_debug_log() if debugging is disabled (bug #145118)
Original commit message from CVS:
* ext/alsa/gstalsa.c: (gst_alsa_xrun_recovery):
use our own functions for restarting the alsa device.
* ext/alsa/gstalsasink.c: (gst_alsa_sink_check_event):
I should apply patches myself - use MIN for the third argument, not
the second, this fixes seeking
Original commit message from CVS:
* ext/flac/gstflacdec.c: (gst_flacdec_class_init),
(gst_flacdec_write): Actually, GST_PAD_CAPS() has nothing to
do with the logic.
Original commit message from CVS:
* ext/ogg/gstogg.c: (plugin_init):
we require bytestream now
* ext/ogg/gstoggdemux.c:
huge diff to implement chain setup in a fast and generic way. This
improves tag reading and startup of huge files (read: Theora videos)
quite a bit. It probably contains bugs, too, so please test.
Seeking is not improved to the fast method.
Original commit message from CVS:
* ext/ogg/gstoggdemux.c: (gst_ogg_pad_push):
* ext/ogg/gstoggmux.c:
Fix memleak in oggdemux when running unconnected pads.
doc update in mux, start working on keyframe mode.
Original commit message from CVS:
* sys/oss/gstosssink.c:
* sys/oss/gstosssrc.c:
advertise correct template caps - we indeed do non-native endianness
and 8bit audio has no endianness
* sys/ximage/ximagesink.c: (gst_ximagesink_getcaps):
* sys/xvimage/xvimagesink.c: (gst_xvimagesink_getcaps):
avoid (wrong) duplications in getcaps function and return
template caps
Original commit message from CVS:
* gst/tcp/gstmultifdsink.c: (gst_recover_policy_get_type),
(gst_multifdsink_class_init), (gst_multifdsink_add),
(gst_multifdsink_remove), (gst_multifdsink_clear),
(gst_multifdsink_client_remove),
(gst_multifdsink_handle_client_read),
(gst_multifdsink_client_queue_data),
(gst_multifdsink_client_queue_caps),
(gst_multifdsink_client_queue_buffer),
(gst_multifdsink_handle_client_write),
(gst_multifdsink_recover_client), (gst_multifdsink_queue_buffer),
(gst_multifdsink_handle_clients), (gst_multifdsink_thread),
(gst_multifdsink_init_send), (gst_multifdsink_close):
Fix wrong GList iteration that could crash the server when
more then 2 clients disconnect at the same time. Read all the
pending commands in one batch to recover from command storms under
very heavy load.
Original commit message from CVS:
2004-06-27 Julien Moutte <julien@moutte.net>
* sys/ximage/ximagesink.c: (gst_ximagesink_xwindow_decorate),
(gst_ximagesink_xwindow_new):
* sys/xvimage/xvimagesink.c: (gst_xvimagesink_xwindow_decorate),
(gst_xvimagesink_xwindow_new): I prefer locking the mutex in the
function directly. We might want to call it from somewhere else one day.
Original commit message from CVS:
2004-06-27 Julien Moutte <julien@moutte.net>
* sys/ximage/ximagesink.c: (gst_ximagesink_xwindow_decorate),
(gst_ximagesink_xwindow_new):
* sys/xvimage/xvimagesink.c: (gst_xvimagesink_xwindow_decorate),
(gst_xvimagesink_xwindow_new): Trying to fix the random behaviour of
window decorations.
Original commit message from CVS:
* ext/dv/gstdvdec.c: (gst_dvdec_class_init), (gst_dvdec_init),
(gst_dvdec_video_getcaps), (gst_dvdec_video_link),
(gst_dvdec_push), (gst_dvdec_loop), (gst_dvdec_change_state),
(gst_dvdec_set_property), (gst_dvdec_get_property):
* ext/dv/gstdvdec.h:
Implement drop_factor property to lower the framerate with
a factor.
Original commit message from CVS:
* ext/dv/gstdvdec.c: (gst_dvdec_push), (gst_dvdec_loop),
(gst_dvdec_change_state):
* ext/dv/gstdvdec.h:
Fix timestamp, duration and offset of the buffers.
Original commit message from CVS:
* gst/tcp/gsttcpserversink.c: (gst_tcpserversink_class_init),
(gst_tcpserversink_init), (gst_tcpserversink_handle_server_read),
(gst_tcpserversink_client_remove),
(gst_tcpserversink_handle_client_read),
(gst_tcpserversink_client_queue_data),
(gst_tcpserversink_client_queue_caps),
(gst_tcpserversink_client_queue_buffer),
(gst_tcpserversink_handle_client_write),
(gst_tcpserversink_queue_buffer),
(gst_tcpserversink_handle_clients), (gst_tcpserversink_thread),
(gst_tcpserversink_chain), (gst_tcpserversink_set_property),
(gst_tcpserversink_get_property), (gst_tcpserversink_init_send),
(gst_tcpserversink_close):
* gst/tcp/gsttcpserversink.h:
Serversink rewrite. Really do non blocking writes to clients and
maintain an internal queue to handle slower clients while not
disturbing fast clients.
Original commit message from CVS:
* ext/theora/theoraenc.c (theora_enc_chain): Call
gst_pad_try_set_caps instead of gst_pad_set_explicit_caps so the
streamheader caps are set correctly.
Original commit message from CVS:
* ext/alsa/gstalsa.c: (gst_alsa_start), (gst_alsa_xrun_recovery):
* ext/alsa/gstalsa.h:
* ext/alsa/gstalsasrc.c: (gst_alsa_src_init),
(gst_alsa_src_update_avail), (gst_alsa_src_loop):
Use alsa trigger_tstamp to get the timestamp of the first
sample in the buffer for more precise sync. Some cleanups.
Original commit message from CVS:
* gst/audiorate/gstaudiorate.c: (gst_audiorate_link),
(gst_audiorate_init), (gst_audiorate_chain),
(gst_audiorate_set_property), (gst_audiorate_get_property):
* gst/videorate/gstvideorate.c: (gst_videorate_class_init),
(gst_videorate_chain):
Added some logging, fixed an overflow bug in videorate.
Original commit message from CVS:
* gst-libs/gst/colorbalance/Makefile.am:
* gst-libs/gst/mixer/Makefile.am:
* gst-libs/gst/play/Makefile.am:
* gst-libs/gst/tuner/Makefile.am:
* gst/tcp/Makefile.am:
* sys/dxr3/Makefile.am:
don't include -enumtypes.[ch] or -marshal.[ch] files in the disted
tarball.
Also add all *.list files that were missing.
* Makefile.am:
add a distcheck hook to ensure the above doesn't happen again.
Original commit message from CVS:
* ext/alsa/gstalsa.c: (gst_alsa_change_state), (gst_alsa_start),
(gst_alsa_xrun_recovery):
* ext/alsa/gstalsa.h:
* ext/alsa/gstalsasink.c: (gst_alsa_sink_check_event),
(gst_alsa_sink_loop), (gst_alsa_sink_get_time):
* ext/alsa/gstalsasrc.c: (gst_alsa_src_init),
(gst_alsa_src_get_time), (gst_alsa_src_update_avail),
(gst_alsa_src_loop):
Add clock to alsasrc. Take new capture timestamp when
restarting after an overrun. Split up some functions between
alsasrc ans alsasink.
Original commit message from CVS:
* ext/theora/theoraenc.c: (gst_theora_enc_class_init),
(theora_enc_sink_link), (theora_buffer_from_packet),
(theora_push_packet), (theora_enc_chain):
Some cleanups, make sure the timestamps are correct.
Original commit message from CVS:
* ext/alsa/gstalsa.c: (gst_alsa_get_time), (gst_alsa_clock_update),
(gst_alsa_change_state), (gst_alsa_update_avail),
(gst_alsa_xrun_recovery):
* ext/alsa/gstalsa.h:
* ext/alsa/gstalsasrc.c: (gst_alsa_src_loop):
Cleanups, take queued samples into account when reporting
the time.
Original commit message from CVS:
* gst/videorate/gstvideorate.c: (gst_videorate_class_init),
(gst_videorate_init), (gst_videorate_chain),
(gst_videorate_set_property), (gst_videorate_get_property):
Add property to make videorate silent.
Add property to prefer new frames over old ones.
Original commit message from CVS:
2004-06-22 Zaheer Abbas Merali <zaheerabbas at merali.org>
* sys/osxvideo/Makefile.am:
Workaround so that the osxvideo .so file gets linked with the
