Original commit message from CVS:
Added lots of neat stuff to fakesrc:
- ability to g_malloc buffer data or to use subbuffers.
- fixed/0/random sized buffers with min/max size
- fill types: none/0/random/patterns
- set the parent buffer size
- an arg for dumping the buffer contents
Original commit message from CVS:
Commit my pending changes.
- revert WHERE stuff
- added fast type checking
- add preliminary stack trace functionality
- minor fixes/cleanups.
Original commit message from CVS:
- if you call gst_element_request_pad_by_name() with, say "sink4" as the pad name
and there is a "sink%d" template, the request_new_pad function will be called.
this is to allow for, for example, requesting a certain channel on a multichannel
device.
- to this end, request_new_pad now has the following prototype:
GstPad* (*request_new_pad) (GstElement *element, GstPadTemplate *templ, const gchar *name)
if request_pad_by_name was called with a name like "sink%d" (i.e. no parsing
was necessary) then NULL is passed as 'name'.
- all plugins that use request pads have been changed to the new api.
also, some incremental (untested) alsa updates, switching to the bytestream api...
Original commit message from CVS:
Sync with my copy before wtay starts hacking...
Decided to keep the GstBuffer name through most code because it's really
messy otherwise, and it doesn't really mean much either way. Saves lots
of stupid casting. fakesrc num_buffers=N ! fakesink ends with a segfault
on EOS right now, not tracked down yet.
Original commit message from CVS:
gstinfo.[ch], cothreads.c: added initial support for -finstrument_functions
gstbin.c: removed a reference to config.h
gstbuffer.[ch]: added gst_buffer_is_span_fast(), used it in gst_buffer_span
elements/gstfilesrc.c: initial work fleshing out the event handling code
everywhere else: wrapped XML stuff in #ifndef's
Original commit message from CVS:
s/Dparam/DParam/ <taaz> i'll whine about an irrelevant thing: Dparam vs DParam. though Dparam is more consistantly screwed up with everything else ;)
Original commit message from CVS:
Updated the event system to use event,timestamp,data instead of a pointer.
Updated elements to use the event system correctly.
Changed "size" to "filesize" in disksrc.
Original commit message from CVS:
Lots of modifications to the plugin system.
- Added a GstPluginfeature object that serves as a base class for all
plugin contents.
- changed the plugin API, everyhting is now added with
gst_plugin_add_feature
- typefactories are named now so that they can be located easily and filled
in at plugin load.
- mime types like "video/raw image/raw" are gone for now.
- lots of plugin updates (style and API changes)
- tested with an without registry.
- updates to various tools.
- added a little testsuite to test/show how you can load plugins (4 modes)
Test this one, Almost everything has changed :-)
Original commit message from CVS:
Added prototypes for gst_pad_peek, gst_pad_select and gst_pad_selectv.
Added different scheduling methods in aggregator.
Fixed a bug in fakesrc when setting up the scheduling functions.
Original commit message from CVS:
Some code cleanups.
Added printouts of the buffer timestamps in identity, fakesrc, fakesink.
Added an aggregator test element.
Original commit message from CVS:
Convert XMMS and LIBGHTTP checks to use GST_CHECK_FEATURE()
XMMS check is turned off by default (though not in a very neat way)
Original commit message from CVS:
Changed a stupid assert in request_pad.
Some fixes for pullregion and EOS conditions.
Remove an unneeded check in the scheduler (check for NULL buffer)
some EOS fixes for pullregion in disksrc.
Removed the macro in the gstparse.h header 'cause it's internal to gstparse.c
Added a check in gstparse for NULL element.
Original commit message from CVS:
Added some caps != NULL checks.
Patched up a problem with the thread handling when the iteration fails.
Cleaned up the output of -inspect a tiny bit.
