Original commit message from CVS:
* ext/esd/esdsink.c: (gst_esdsink_finalize), (gst_esdsink_getcaps),
(gst_esdsink_open), (gst_esdsink_close):
* ext/esd/esdsink.h:
Move the esd_get_server_info() into gst_esdsink_open() and fail
with a decent error message on errors.
Original commit message from CVS:
Patch by: Mark Nauwelaerts <manauw at skynet bet>
* gst/avi/gstavimux.c: (gst_avi_mux_do_audio_buffer),
(gst_avi_mux_do_video_buffer):
Work around gst_buffer_make_metadata_writable() bug that
results in avimux marking all frames in the index as
keyframes (#340859).
Original commit message from CVS:
* ext/flac/gstflacdec.c: (gst_flac_dec_loop):
* ext/flac/gstflacdec.h:
Handle segment seeks that include the end of the file as stop point
properly: when the decoder hits EOS we want to send a SEGMENT_DONE
message instead of an EOS event in case we're in segment seek
mode (fixes#340699).
Original commit message from CVS:
Patch by: Michal Benes <michal dot benes at xeris dot cz>
* gst/matroska/matroska-demux.c: (gst_matroska_demux_reset):
Don't leak caps when freeing the stream context (#340623).
Original commit message from CVS:
* gst/matroska/matroska-mux.c:
(gst_matroska_mux_stream_is_vorbis_header),
(gst_matroska_mux_write_data):
Don't strcmp() NULL strings.
Only start new clusters on video keyframes, not on any
random audio buffer that doesn't have the DELTA_UNIT
flag set (fixes 'make check' again).
Original commit message from CVS:
Patch by: Mark Nauwelaerts <manauw at skynet be>
* gst/matroska/matroska-mux.c: (gst_matroska_mux_best_pad),
(gst_matroska_mux_stream_is_vorbis_header),
(gst_matroska_mux_write_data):
Don't misinterpret GST_CLOCK_TIME_NONE as very high timestamp
value and then dead-lock when muxing vorbis audio streams
(the three vorbis header buffers carry no timestamp, and it
would try to mux these after all video buffers). Fixes#340346.
Improve clustering: start a new cluster also whenever we get
a keyframe.
Original commit message from CVS:
* ext/cairo/gsttimeoverlay.c: (gst_cairo_time_overlay_transform):
Fix timeoverlay for non-multiple-of-4 widths. This fourcc crap
SUCKS.
Original commit message from CVS:
* gst/id3demux/gstid3demux.c: (gst_id3demux_chain),
(gst_id3demux_sink_activate):
Let core insert default error message for TYPE_NOT_FOUND
errors, it's just as good as our own and has the added
bonus of being translated.
Original commit message from CVS:
* gst/apetag/gsttagdemux.c: (gst_tag_demux_init),
(gst_tag_demux_sink_event):
* gst/id3demux/gstid3demux.c: (gst_id3demux_init),
(gst_id3demux_sink_event):
Post an error message when we get an EOS event and were not
able to find out the type of stream.
* tests/check/elements/id3v2mux.c: (fill_mp3_buffer), (got_buffer),
(test_taglib_id3mux_with_tags):
Decrease num-buffers to 16 per iteration again, otherwise the
many memcpy()s and reallocations in the test will hammer slow
CPUs completely and make the test timeout.
Original commit message from CVS:
* configure.ac:
figure out where plugins-base plugins are
* tests/check/Makefile.am:
use plugins-base plugins, so we have typefind functions
* tests/check/elements/id3v2mux.c: (test_taglib_id3mux_with_tags):
increase num-buffers, this makes sure the test errors out instead
of timing out when no typefind functions are present
Original commit message from CVS:
2006-04-28 Zaheer Abbas Merali <zaheerabbas at merali dot org>
* sys/osxaudio/gstosxaudiosink.c:
(plugin_init):
Register osxaudiosrc to the plugin.
* sys/osxaudio/gstosxaudiosrc.c:
(gst_osx_audio_src_osxelement_do_init),
(gst_osx_audio_src_base_init), (gst_osx_audio_src_class_init),
(gst_osx_audio_src_init), (gst_osx_audio_src_set_property),
(gst_osx_audio_src_get_property),
(gst_osx_audio_src_create_ringbuffer), (gst_osx_audio_src_io_proc),
(gst_osx_audio_src_osxelement_init):
* sys/osxaudio/gstosxaudiosrc.h:
Port of osxaudiosrc to 0.10.
* sys/osxaudio/Makefile.am:
Add osxaudiosrc
Original commit message from CVS:
* gst/id3demux/id3v2frames.c: (has_utf16_bom),
(parse_split_strings):
Recognise and skip any byte order marker (BOM) in
UTF-16 strings.
Original commit message from CVS:
* gst/auparse/gstauparse.c: (gst_au_parse_base_init),
(gst_au_parse_class_init), (gst_au_parse_init),
(gst_au_parse_reset), (gst_au_parse_add_srcpad),
(gst_au_parse_remove_srcpad), (gst_au_parse_parse_header),
(gst_au_parse_chain), (gst_au_parse_src_convert),
(gst_au_parse_src_query), (gst_au_parse_handle_seek),
(gst_au_parse_sink_event), (gst_au_parse_src_event),
(gst_au_parse_change_state):
* gst/auparse/gstauparse.h:
Rewrite auparse to suck a little bit less: make source pad
dynamic, so decodebin/playbin work with non-raw formats
like alaw/mulaw; add query function for duration/position
queries; check whether we have enough data before attempting
to parse the header (instead of crashing when that is not the
case); work around audioconvert sucking by swapping endianness
to the native endianness ourselves for float formats; send
initial newsegment event. Fixes#161712.
Original commit message from CVS:
* ext/esd/esdsink.c: (gst_esdsink_prepare), (gst_esdsink_delay):
Always write ESD_BUF_SIZE bytes and use ESD_MAX_WRITE_SIZE as
the size of the ringbuffer. This should fix hangs with older
esd sound servers.
Original commit message from CVS:
* ext/jpeg/gstjpegdec.c: (gst_jpeg_dec_chain):
Source pad has fixed caps. If we don't set this, bad
things happen when the window is resized.
Original commit message from CVS:
* gst/matroska/Makefile.am:
* gst/matroska/matroska-demux.c: (gst_matroska_demux_add_stream),
(gst_matroska_demux_handle_src_event):
* gst/matroska/matroska-ids.c:
(gst_matroska_track_init_video_context),
(gst_matroska_track_init_audio_context),
(gst_matroska_track_init_subtitle_context),
(gst_matroska_track_init_complex_context):
* gst/matroska/matroska-ids.h:
Handle case where the TrackType ebml chunk does not come before the
TrackInfoAudio or TrackInfoVideo ebml chunk (#339446). Ignore QoS
events.
Original commit message from CVS:
* gst/rtp/gstrtpmp4gpay.c: (gst_rtp_mp4g_pay_setcaps):
* gst/rtp/gstrtpmp4vpay.c: (gst_rtp_mp4v_pay_setcaps):
It's codec_data, not codec_info.
Original commit message from CVS:
Patch by: Mark Nauwelaerts <manauw at skynet dot be>
* gst/matroska/matroska-demux.c: (gst_matroska_demux_video_caps):
Handle codec_data for VfW compatibility codec IDs (#339451)
* gst/matroska/matroska-mux.c:
(gst_matroska_mux_video_pad_setcaps):
Same here, handle codec_data and add additional caps we can handle
now to the pad template (huffyuv, dv and h263 video) (#339451)
Original commit message from CVS:
Patch by: Josef Zlomek <josef dot zlomek at itonis dot tv>
* gst/matroska/matroska-mux.c:
(gst_matroska_mux_create_buffer_header),
(gst_matroska_mux_write_data):
Fix timestamping of B-frames, use signed integers, do
some rounding (#339678).
Original commit message from CVS:
* gst/matroska/matroska-demux.c: (gst_matroskademux_do_index_seek):
Fix a bad conversion using gst_guint64_to_gdouble.
fabs ((gdouble) demux->index[entry].time - (gdouble) seek_pos) can not be
replaced by fabs (gst_guint64_to_gdouble (demux->index[entry].time - seek_pos)) as the
difference could be negative. fabs (gst_guint64_to_gdouble (demux->index[entry].time) -
gst_guint64_to_gdouble (seek_pos)) is the good solution. Thanks to Tim who has seen my
mistake.
Original commit message from CVS:
* gst/matroska/matroska-demux.c: (gst_matroskademux_do_index_seek):
Use gst_guint64_to_gdouble for conversions
* win32/vs6/gst_plugins_good.dsw:
* win32/vs6/libgsticydemux.dsp:
Add a project file for icydemux
Original commit message from CVS:
Patch by: Fabrizio Gennari <fabrizio dot ge at tiscali dot it>
* gst/avi/gstavidemux.c: (gst_avi_demux_parse_stream),
(gst_avi_demux_parse_index), (gst_avi_demux_massage_index):
When splitting audio chunks, the block alignment is not taken in
consideration, so the smaller chunks could be of size which is
not a multiple of the block alignment. Fixes#336904
Original commit message from CVS:
* gst/debug/progressreport.c: (gst_progress_report_finalize),
(gst_progress_report_class_init), (gst_progress_report_init),
(gst_progress_report_do_query), (gst_progress_report_report),
(gst_progress_report_set_property),
(gst_progress_report_get_property):
Add 'format' property to force querying to a particular format.
Original commit message from CVS:
2006-04-21 Andy Wingo <wingo@pobox.com>
* ext/dv/gstdv.c (plugin_init): libdv is a marginal decoder, at
best, on big endian systems. Drop its rank in that case. OTOH on
x86 it's quite fine. See changes from today in gst-ffmpeg as well.
Original commit message from CVS:
Patch by: Fabrizio Gennari <fabrizio dot ge at tiscali dot it>
* gst/avi/gstavidemux.c: (gst_avi_demux_peek_tag),
(gst_avi_demux_next_data_buffer), (gst_avi_demux_stream_scan):
Fix index creation when we have to scan the file to create
an index. There may be other types of RIFF 'LIST' chunks than
'movi' and we need to skip them properly as well or we'll end up
reading garbage (#336889). Some other cosmetic changes.
Original commit message from CVS:
* ext/flac/gstflacdec.c: (gst_flac_dec_loop),
(gst_flac_dec_handle_seek_event):
Add support for segment seeks (fixes#338290). Also demote
some recurring debug message from DEBUG to LOG level.
Original commit message from CVS:
* gst/matroska/matroska-demux.c: (gst_matroska_demux_add_stream),
(gst_matroskademux_do_index_seek),
(gst_matroska_demux_handle_seek_event),
(gst_matroska_demux_parse_blockgroup_or_simpleblock):
* gst/matroska/matroska-ids.h:
Set DISCONT flag on first buffer after a discontinuity.
Fix newsegment events sent when seeking and honour KEY_UNIT
seek flag. Create pad with bogus caps if we don't recognise
the stream codec id.
* gst/matroska/matroska-demux.h:
Fix GObject macros.
Original commit message from CVS:
Patch by: Mark Nauwelaerts <manauw at skynet dot be>
* gst/matroska/matroska-demux.c:
(gst_matroska_demux_handle_seek_event), (gst_matroska_demux_loop):
Handle end of segment properly when set; don't dead-lock when
posting start of segment message when doing a segment seek.
Fixes#338810.
Original commit message from CVS:
* gst/matroska/matroska-demux.c: (gst_matroska_demux_audio_caps),
(gst_matroska_demux_plugin_init):
Make mpeg2 aac audio work: create artificial private codec data
chunk which faad2 seems to require, just as we do for mpeg4 aac.
Also call gst_riff_init(). Partially fixes#338767.
Original commit message from CVS:
* gst/wavenc/gstwavenc.c: (gst_wavenc_base_init),
(gst_wavenc_class_init), (gst_wavenc_init),
(gst_wavenc_create_header_buf), (gst_wavenc_push_header),
(gst_wavenc_sink_setcaps), (get_id_from_name), (gst_wavenc_event),
(gst_wavenc_chain), (gst_wavenc_change_state):
* gst/wavenc/gstwavenc.h:
Set caps on first outgoing buffer, so that it doesn't error out
immediately with a non-negotiated error (#338716). Rewrite and
clean up a bit; fix setcaps function to parse things properly;
fix sink caps (8bit audio is unsigned and doesn't have depth);
use boilerplate macros; remove unused properties stuff.
Original commit message from CVS:
* ext/gdk_pixbuf/gstgdkpixbuf.c:
Leave JPEG decoding to our jpegdec plugin. gdkpixbufdec cannot
handle MJPEG streams and might be autoplugged for those if the
user doesn't have jpegdec installed (resulting in a cryptic error
message about huffman tables). Better to disable JPEG decoding here
and let the user figure out that she needs to install jpegdec.
Original commit message from CVS:
* ext/gdk_pixbuf/gstgdkpixbuf.c: (gst_gdk_pixbuf_sink_setcaps),
(gst_gdk_pixbuf_class_init), (gst_gdk_pixbuf_init),
(gst_gdk_pixbuf_flush), (gst_gdk_pixbuf_chain):
* ext/gdk_pixbuf/gstgdkpixbuf.h:
Make work with packetised/framed input (e.g. png-in-quicktime). Use
GST_ELEMENT_ERROR when we return GST_FLOW_ERROR. Add some
GST_DEBUG_FUNCPTR here and there. Use GST_LOG for recurring
debug messages. Fix boilerplate macros.
Original commit message from CVS:
* ext/gdk_pixbuf/gstgdkpixbuf.c: (gst_gdk_pixbuf_get_capslist),
(gst_gdk_pixbuf_set_property), (gst_gdk_pixbuf_get_property):
No need to special-case for Gdk-2.0 any longer, we require
Gdk 2.2 or newer; minor clean-ups.
Original commit message from CVS:
* ext/shout2/gstshout2.c: (gst_shout2send_base_init),
(gst_shout2send_class_init), (gst_shout2send_init),
(set_shout_metadata), (gst_shout2send_set_metadata),
(gst_shout2send_event), (gst_shout2send_start),
(gst_shout2send_connect), (gst_shout2send_stop),
(gst_shout2send_render), (gst_shout2send_set_property),
(gst_shout2send_get_property), (gst_shout2send_setcaps),
(plugin_init):
* ext/shout2/gstshout2.h:
* po/POTFILES.in:
Rewrite a bit: use GstBaseSink::start and stop instead of a state
change function; use GST_ELEMENT_ERROR for error reporting, not
g_error() or GST_ERROR(); don't unref caps in setcaps function,
will cause crashes or assertion failures; remove (unused) "sync"
property, basesink already has such a property; misc. other
minor fixes and cleanups.
Original commit message from CVS:
* ext/esd/esdsink.c: (gst_esdsink_open), (gst_esdsink_prepare):
* ext/esd/gstesd.c: (plugin_init):
* po/POTFILES.in:
Add translatable error message for when we cannot
connect to the sound server, as "Cannot open resource
for writing" isn't really an acceptable message to show
to the user in this case.
Original commit message from CVS:
Patch by: Philippe Valembois
* ext/shout2/gstshout2.c: (gst_shout2send_init),
(gst_shout2send_set_metadata), (gst_shout2send_event),
(gst_shout2send_render), (gst_shout2send_change_state):
* ext/shout2/gstshout2.h:
Handle tags being received before the connection to
the server is established properly (see #338636).
Original commit message from CVS:
* ext/shout2/gstshout2.c: (gst_shout2send_render):
Don't crash in case the connection to the server fails:
don't set pointer to NULL by assigning FALSE; error out
properly by using GST_ELEMENT_ERROR and returning
GST_FLOW_ERROR (fixes#338636). Lastly, free connection
before resetting the pointer.
Original commit message from CVS:
2006-04-12 Philippe Kalaf <philippe.kalaf@collabora.co.uk>
* gst/rtp/gstrtppcmapay.c:
* gst/rtp/gstrtppcmapay.h:
* gst/rtp/gstrtppcmupay.c:
* gst/rtp/gstrtppcmupay.h:
Ported mulaw and alaw payloaders to use new base class
* gst/rtp/Makefile.am:
* gst/rtp/gstrtp.c:
* gst/rtp/gstrtpilbcpay.c:
* gst/rtp/gstrtpilbcpay.h:
* gst/rtp/gstrtpilbcdepay.c:
* gst/rtp/gstrtpilbcdepay.h:
Added new iLBC payloader/depayloader. Payloader uses new audio payload base
class.
Original commit message from CVS:
* ext/gdk_pixbuf/gstgdkpixbuf.c: (gst_gdk_pixbuf_sink_setcaps),
(gst_gdk_pixbuf_get_capslist), (gst_gdk_pixbuf_sink_getcaps),
(gst_gdk_pixbuf_class_init), (gst_gdk_pixbuf_init),
(gst_gdk_pixbuf_flush), (gst_gdk_pixbuf_sink_event),
(gst_gdk_pixbuf_chain):
Some cleanups.
Added RGBA as a possible output format.
Correctly free the supported mimetypes.
deprecate silent arg, it's not used.
Return result from _alloc_buffer to peer.
Original commit message from CVS:
* gst/rtp/gstrtpmp4vdepay.c: (gst_rtp_mp4v_depay_process):
Don't leak memory allocated by gst_buffer_new_and_alloc() by
overwriting GST_BUFFER_MALLOCDATA.
Original commit message from CVS:
* ext/libpng/gstpngdec.c: (gst_pngdec_init),
(user_endrow_callback), (user_end_callback),
(gst_pngdec_caps_create_and_set), (gst_pngdec_chain),
(gst_pngdec_sink_setcaps), (gst_pngdec_sink_event),
(gst_pngdec_libpng_clear), (gst_pngdec_change_state):
* ext/libpng/gstpngdec.h:
Handle more than one frame if the content is framed,
like with png-in-quicktime (#331917).
Original commit message from CVS:
* sys/oss/Makefile.am:
* sys/oss/common.h:
* sys/oss/gstosssink.c: (gst_oss_sink_init), (gst_oss_sink_open),
(gst_oss_sink_prepare), (gst_oss_sink_unprepare):
* sys/oss/gstosssrc.c: (gst_oss_src_prepare),
(gst_oss_src_unprepare):
- the user-visible error strings were in the wrong category
- and the messages were not marked for translation
- which is actually a good thing, because they were exactly
the kind of message you would never want anyone to see
- the macros were using variables that didn't exist in the macro
arguments
- and they were obviously copied from each other and then modified
- so a common header makes sense
Original commit message from CVS:
* ext/annodex/gstcmmlenc.c: (gst_cmml_enc_set_header_on_caps):
Use copies of header buffers for caps to avoid circular refcounting
problems (as in theoradec, vorbisdec).
* tests/check/elements/cmmldec.c: (GST_START_TEST):
Fix a typo in test that meant it was testing the wrong thing.
* tests/check/elements/cmmlenc.c: (check_headers):
Fix refcount checks now that we use buffer-copies for caps.
Original commit message from CVS:
* gst/matroska/matroska-demux.c: (gst_matroska_demux_base_init),
(gst_matroska_demux_handle_seek_event),
(gst_matroska_demux_video_caps), (gst_matroska_demux_audio_caps),
(gst_matroska_demux_subtitle_caps),
(gst_matroska_demux_plugin_init):
Use static pad templates with ANY caps for audio and video
source pads and get rid of a lot of unnecessary (and partially
broken) code for the template caps. Clean up caps finding
functions. Fixes playback of audio files/streams that do not
contain the sample rate and/or number of channels in the audio
context (happens a lot with vorbis/mp3 .mka files it seems).
Fixes#337183.
Also add myself to copyright holders.
Original commit message from CVS:
* ext/annodex/gstcmmlutils.c: (gst_cmml_track_list_del_clip):
Use g_list_delete_link () instead of g_list_remove_link () so that
we free the link as well as the contained data.
Original commit message from CVS:
Patch by: Ryan Lortie (desrt) <desrt at destr dot ca>
* gst/avi/gstavidemux.c: (gst_avi_demux_parse_superindex),
(gst_avi_demux_parse_stream), (gst_avi_demux_parse_index),
(gst_avi_demux_stream_header):
Fix some crashers with empty chunks. (Fixes#337749)
Original commit message from CVS:
* gst/level/gstlevel.c: (gst_level_set_caps),(gst_level_transform_ip):
use G_GINT64_CONSTANT for INT64 constants
* gst/videofilter/gstvideobalance.c:
define rint for WIN32 #define rint(x) (floor((x)+0.5))
* win32/vs6/libgstavi.dsp:
add missing libraries for the link and remove avimux.c from
the project as it isn't ported to 0.10 yet
Original commit message from CVS:
* gst/matroska/ebml-read.c: (gst_ebml_read_sint):
Even better would be if we actually did the right thing
here (also, G_GUINT64_CONSTANT only exists since GLib-2.10).
Original commit message from CVS:
* gst/matroska/ebml-read.c: (gst_ebml_read_sint):
Can't just replace 1LL with 1L here just because MSVC doesn't
support it, as it might lead to incorrect results when doing the
bitshifting here. Using GLib's G_GUINT64_CONSTANT() macro to
force a 64-bit constant in a way that all compilers are happy with.
Original commit message from CVS:
* ext/flac/gstflacdec.c: (gst_flac_dec_handle_seek_event):
Don't try to seek beyond the end of the file (would
occasionally display error dialogs in totem when seeking
to the end) (#335869). Will still throw an error though
if the file is truncated and the total_samples value in
the stream header is wrong.
Original commit message from CVS:
* ext/flac/gstflacdec.c: (gst_flac_calculate_crc8),
(gst_flac_dec_scan_got_frame), (gst_flac_dec_scan_for_last_block),
(gst_flac_dec_metadata_callback):
* ext/flac/gstflacdec.h:
If the stream header doesn't contain the total number of samples,
search for the last flac frame at the end of the file and calculate
the total duration from that frame's offset (fixes#337609).
Original commit message from CVS:
* gst/avi/gstavidemux.c: (gst_avi_demux_parse_stream):
Don't unref the GstPadTemplate returned by
gst_element_class_get_pad_template().
Original commit message from CVS:
Patch by: Brian Cameron <brian dot cameron at sun dot com>
* sys/sunaudio/gstsunaudiosink.c: (gst_sunaudiosink_init),
(gst_sunaudiosink_prepare), (gst_sunaudiosink_write):
* sys/sunaudio/gstsunaudiosink.h:
Use spec->segsize and spec->segtotal in the prepare function
to initialise the ring buffer instead of using the buffer-time
property (#337421).
Original commit message from CVS:
* Makefile.am:
* configure.ac:
* pkgconfig/.cvsignore:
* pkgconfig/Makefile.am:
* pkgconfig/gstreamer-plugins-good-uninstalled.pc.in:
add a .pc file so other modules can use good plugins in tests
Original commit message from CVS:
* ext\jpeg\smokecodec.c:
use of GST_DEBUG instead of DEBUG(a...) for WIN32
* ext\speex\gstspeexenc.c: (gst_speexenc_set_header_on_caps):
move first instruction after all variables declarations
* gst\alpha\gstalpha.c:
* gst\effectv\gstshagadelic.c:
* gst\smpte\paint.c:
* gst\videofilter\gstvideobalance.c:
define M_PI if it's not defined (it's not defined on WIN32)
* gst\cutter\gstcutter.c: (gst_cutter_chain):
* gst\id3demux\id3v2frames.c: (parse_relative_volume_adjustment_two):
* gst\level\gstlevel.c: (gst_level_set_property), (gst_level_transform_ip):
* gst\matroska\matroska-demux.c: (gst_matroska_demux_parse_info),
(gst_matroska_demux_video_caps):
* gst\matroska\matroska-mux.c: (gst_matroska_mux_start), (gst_matroska_mux_finish):
* gst\wavparse\gstwavparse.c: (gst_wavparse_stream_data):
use gst_guint64_to_gdouble for conversions
* gst\goom\filters.c: (setPixelRGB_):
fix a debug which was using undefined variable
* gst\level\gstlevel.c: (gst_level_set_caps), (gst_level_transform_ip):
* gst\matroska\ebml-read.c: (gst_ebml_read_sint):
replace LL suffix with L suffix (LL isn't supported by MSVC6.0)
* win32/vs6:
add vs6 projects files for most of plugins-good
Original commit message from CVS:
From a patch by: Michael Dominic K. <mdk at mdk dot org dot pl>
* ext/dv/gstdvdemux.c: (gst_dvdemux_class_init),
(gst_dvdemux_reset), (gst_dvdemux_src_convert),
(gst_dvdemux_send_event), (gst_dvdemux_flush), (gst_dvdemux_loop),
(gst_dvdemux_sink_activate_pull), (gst_dvdemux_change_state):
* ext/dv/gstdvdemux.h:
Seek in READY patch. Only works for pull based mode.
Fixes#323880
Original commit message from CVS:
* ext/gdk_pixbuf/gstgdkpixbuf.c: (gst_gdk_pixbuf_flush),
(gst_gdk_pixbuf_event):
Fix two crashers: don't unref the same caps twice, and
set pixbuf loader to NULL after freeing it.
Original commit message from CVS:
* ext/speex/gstspeexenc.c: (gst_speexenc_class_init),
(gst_speexenc_finalize), (gst_speexenc_sink_setcaps),
(gst_speexenc_chain):
* ext/speex/gstspeexenc.h:
Don't leak adapter.
A push *always* takes ownership of the buffer, even on
errors.
Small cleanups.
