Don't unref the event if it hasn't been handled, because the caller
assumes it is still valid and might reuse it.
I ran into this problem when transcoding an AVI (with mp3 inside)
to gpp.
https://bugzilla.gnome.org/show_bug.cgi?id=639555
That is, as such formats allow subclass to extract position from frame,
it is possible to extract duration (if not otherwise provided)
from (near) last frame, and a seek can fairly accurately target the required
position.
Fixes#631389.
Arrange for upstream as well as downstream flushing when seeking.
Also determine upstream size as well as seekability. Adjust some comments
to reality and employ debug statement in proper order.
This reverts commit b5a3d60363.
Reverting this for now, since no one really seems to remember why this
property exists or what it could possibly be good for. It seems to have
been in the original mp3parse since the beginning of time and was back-
ported from there.
Seekability, like duration, etc is unlikely to change (frequently), and
the default assumption covers most cases, so let subclass set when needed.
At the same time, allow subclass to indicate if it has seek-metadata (table)
available, and possibly have it provide an average bitrate.
This allows the child class to chain its event handler with
GstBaseParse, so that subclasses don't have to duplicate all the default
event handling logic.
https://bugzilla.gnome.org/show_bug.cgi?id=622276
We wait to parse a minimum number of frames (10, arbitrarily) before
emiting bitrate tags so that our early estimates are not wildly
inaccurate for streams that start with a silence. If the stream ends
before that, we just emit the tags anyway.
While it _would_ be nicer to be specify the threshold to start pushing
the tags in terms of duration, this would introduce more complexity than
this merits.
https://bugzilla.gnome.org/show_bug.cgi?id=614991
This makes baseparse keep a running average of the stream bitrate, as
well as the minimum and maximum bitrates. Subclasses can override a
vfunc to make sure that per-frame overhead from the container is not
accounted for in the bitrate calculation.
We take care not to override the bitrate, minimum-bitrate, and
maximum-bitrate tags if they have been posted upstream. We also
rate-limit the emission of bitrate so that it is only triggered by a
change of >10 kbps.
Perform sanity check on type of seek, and only perform one that is
appropriately supported. Adjust downstream newsegment event
to first buffer timestamp that is sent downstream.
In particular, consider DISCONT == !sync, and allow subclass to query
sync state, as it may want to perform additional checks depending
on whether sync was achieved earlier on.
Also arrange for subclass to query whether leftover data is being drained.
In particular, (optionally) provide baseparse with a notion of frames per second
(and therefore also frame duration) and have it track frame and byte counts.
This way, subclass can provide baseparse with fps and have it provide default
buffer time metadata and conversions, though subclass can still install
callbacks to handle such itself.
After all, stream is as-is, and there is little molding to downstream's
taste that can be done. If subclass can and wants to do so, it can
still override as such.
Also handle the case gracefully where the subclass decides to drop
the first buffers and has no caps set yet. It's still required to
have valid caps set when the first buffer should be passed downstream.
Sending the flush-start event forward before taking the stream lock actually
works, in contrast to deadlocking in downstream preroll_wait (hunk 1).
After that we get the chain function being stuck in a busy loop. This is fixed
by updating the minimum frame size inside the synchronization loop because the
subclass asks for more data in this way (hunk 2).
Finally, this leads to a very probable crash because the subclass can find a
valid frame with a size greater than the currently available data in the
adapter. This makes the subsequent gst_adapter_take_buffer call return NULL,
which is not expected (hunk 3).
Baseparse internaly breaks the semantics of a _chain function by calling it with
buffer==NULL. The reson I belived it was okay to remove it was that there is
also an unchecked access to buffer later in _chain. Actually that code is wrong,
as it most probably wants to set discont on the outgoing buffer.
This allows to only create the socketpair when it is really required instead
of always creating it and immediately destroying it again for child buses.
https://bugzilla.gnome.org/show_bug.cgi?id=647005
This is used by GstBin to create a child bus without
a socketpair because child buses will always work
synchronous. Otherwise too many sockets could be
created and the limit of file descriptors for the
process could be reached.
Fixes bug #646624.
if set->control_pending is set to 0 but we didn't not succed reading
the control socket, future calls to gst_poll_wait() will be awaiken
by the control socket which will not be released properly because
set->control_pending is already 0, causing an infinite loop.
This caused "re-declaration" problems.
./clutter-gst-video-sink.c: In function ‘clutter_gst_video_sink_init_interfaces’:
./clutter-gst-video-sink.c:231:1: warning: declaration of ‘ClutterGstVideoSink’ shadows a global declaration [-Wshadow]
./clutter-gst-video-sink.h:64:44: warning: shadowed declaration is here [-Wshadow]
https://bugzilla.gnome.org/show_bug.cgi?id=646531
Some applications are requesting the same pad name multiple times
and the behaviour is undefined and different from element to element
but we don't want to break applications that work just fine.
In 0.11 this check should be an assertion again, although elements
have to do manual checking if the pad already exists again because
it can't be done in a threadsafe way here.
Allow for automatic merging of memory block in the _map function and automatic
freeing of the temporary memory.
Remove some unneeded functions.
Add possibility to force writable spanned memory.
basesrc's default event handler returns TRUE regardless of whether the
event is handled or not. This fixes the handler to conform with the
expected behaviour (which is to only return TRUE when the event has
actually benn handled). gst_bin_do_latency_func() depended on this
(incorrect) behaviour, and is now modified as well.
(Remaining 1-liner change in gstbasesrc.c is to keep gst-indent happy)
Just like gst_caps_intersect, but adds a new parameter 'mode'
that allows selecting the intersection algorithm to use.
Currently we have GST_CAPS_INTERSECT_MODE_ZIG_ZAG (default) and
GST_CAPS_INTERSECT_MODE_FIRST.
API: gst_caps_intersect_full
API: GstCapsIntersectMode
API: GST_CAPS_INTERSECT_MODE_ZIG_ZAG
API: GST_CAPS_INTERSECT_MODE_FIRST
https://bugzilla.gnome.org/show_bug.cgi?id=617045
Adding a buffer to the end of a GstBufferList is supposed to be a fast
operation, but it was not since the iterator does not advance its
nextpointer when adding buffers and GList does not have a tail pointer.
Using a GQueue to store the buffers makes it easier to add buffers to
the end of the list and this operation will now be much more efficient.
Adding an entire GList of buffers using
gst_buffer_list_iterator_add_list() will however have to iterate over
the list being added to be able to update the tail pointer in the
GQueue.
GST_DISABLE_DEPRECATED should only affect visibility of declarations in headers,
not actually remove symbols. See GitDeveloperGuidelines and DeprecatingAPI
pages in wiki.
https://bugzilla.gnome.org/show_bug.cgi?id=402141
Add new functions to clarify how the caps are compared to the template caps of
the element factory. Improve the docs to point out the difference.
Deprecate: gst_element_factory_can_{src|sink}_caps
API: add gst_element_factory_can_{src|sink}_{any|all}_capps
https://bugzilla.gnome.org/show_bug.cgi?id=402141
Replace subbuffer and copy vmethods by a more generic transform function that
can then be parametrised by transform specific data. This should allow us to
implement make-writable and more future transform functions.
Check if structure has been created before appending it to the caps. Free the
caps in the case of an error to not conceal it be returning empty caps.
Fixes#642271
Keep a pointer to the bufferpool. Release the buffer to the pool when
finalizing. Make sure the pool sets itself as the pool member of buffers that it
sends out.
Move some methods around.
Make sure we check for config parsing errors.
Increment the outstanding buffers before calling acquire so that we can be sure
that set_active() doesn't free the pool from under us.
Add start/stop methods to allow for bulk allocation of buffers.
Free buffers only when all outstanding buffers returned.
Make things more threadsafe wrt flushing and starting/stopping by
keeping track of start and stop method calls.
Use a lock to protect concurrect execution of set_config and set_active.
Start freeing the buffers when flushing and all buffers are returned to the
pool.
Make a copy of the config to avoid crashing with concurrent access.
Keep track if the buffer is configured and block activation when not configured
yet.
Keep track of outstanding buffers and disallow configuration when not all
buffers are returned to the pool. We need to do this or else we might end up
with wrong buffers in the pool.
Add return value to set_active.
Small cleanups. Fix finalize.
Use a GstStructure to provide the pool with the right configuration. Also
provide some helper methods to configure such a structure.
don't pass the config in alloc_buffer, pool implementation will already have
parsed it during set_config.
Update defs
Make separate api for getting and adding metadata. This allows us to pass extra
parameters to the init functions when creating metadata, which is needed for
specific API implementations.
Add beginnings of memory metadata.
Add first implementation of arbitrary buffer metadata. We use a simple linked
linked of slice allocated metadata chunks. Future implementations could use
something more performant.
Add get, remove, iterate methods to handle the metadata.
Add function that (unlike the GLib equivalent) also accepts paths that
aren't absolute and will clean up relative markers such as ./ and ../
before forming a URI.
Fixes warnings with e.g. filesrc location=foo ! typefind caused by the
recent switch to g_filename_to_uri(), but also actually creates valid
URIs for the first time.
Windows code paths could need some more work, e.g. we don't clean up
the relative markers there for now (because path could have \ and /
as separators).
API: gst_filename_to_uri()
Add an owner private field where the owner of a buffer can store some extra
information. We can use this to implement most of the subclassing that happens
now. Later this will be removed and replaced by arbitrary buffer metadata.
This was required to add a new MEDIA4 buffer flag for indicating
progressive/mixed telecine video buffers. There is no space for
additional flags in GstBuffer, so steal one from GstMiniObject.
https://bugzilla.gnome.org/show_bug.cgi?id=642671
This commit changes the request pad behaviour for plugins and applications.
Reopens Bug #402562
The proper fix for that bug is to keep track of created request pads.
This reverts commit a5e44ffffa.
Use new GstPoll functionality to wakeup the mainloop.
Use an atomic queue on the writer side to post the messages.
The reader side it protected with the lock still because we don't want multiple
concurrent readers.
Add an atomic queue. The queue can be used from multiple threads simultaneously
and without taking any locks or doing any blocking operations. This makes it
highly scalable for things like the bus, bufferpools and object recycling.
We need to ensure we call gst_pad_check_link() with the two pads in the correct
order. The order depends on wheter we iterate src or sink pads.
Signed-off-by: Chen Rui <rui.chen@tieto.com>
Check the sinkpad for the flushing state before calling the chainfunction on the
pad. We do this by checking the cache (which is also cleared on the srcpad when
the sink is set to flushing).
Fixes#641928
Failing to do so in the Windows case (implicitly triggered otherwise)
would have a subsequent _wait return immediately leading to high CPU
usage timeout loops.
Fixes#640675.
Make adding/removing gst_debug_log_default() work reliably in all
circumstances. The problem was that depending on platform and linker
flags the function argument might resolve to different addresses,
which made it impossible to remove the default log function added
in gst_init() from application code (because the pointer values
didn't match). The new approach should keep things simple by passing
NULL for the default function, which the code in libgstreamer can
then handle.
https://bugzilla.gnome.org/show_bug.cgi?id=625396https://bugzilla.gnome.org/show_bug.cgi?id=640771
Makes gst_bus_add_watch(), gst_bus_add_watch_full(), gst_bus_add_signal_watch(),
and gst_bus_add_signal_watch_full() convenience functions automatically pick up
any non-default main contexts set for the current thread via
g_main_thread_push_thread_default().
gstelement.c: In function ‘gst_element_get_request_pad’:
gstelement.c:1052:18: error: variable ‘tmp’ set but not used [-Werror=unused-but-set-variable]
https://bugzilla.gnome.org/show_bug.cgi?id=640850
The description string was changed to an inlined string a while back.
(But: no need to intern the const strings here, we just use the interning
to avoid allocating duplicates and make memory management easier,
since the strings will be around for the life-time of the app anyway).
https://bugzilla.gnome.org/show_bug.cgi?id=640071
Only replace existing plugins by blacklisted ones if they correspond
to the exact same plugin. If they're not the same, keep the existing
valid one.
Fixes#638941
Ignore plugins which have been moved into coreelements, so it's
still possible to just upgrade GStreamer core without having to
upgrade the whole stack.
This reverts commit f9039c2204.
We use -DG_DISABLE_ASSERTS for releases and pre-releases, but
don't want to disable pad name checking for releases in general,
I think. Need a better solution here. Fixes pad unit test in
pre-release/release mode.
Avoid unnecessary malloc/free to get the file basename on MSVC to avoid
unnecessary overhead when doing GST_DEBUG=foo:5 or so (since it would be
done before the category log level filtering).
The new request_new_pad_full vmethod provides an additional caps field,
which allows elements to take better decision process.
Also, add a gst_element_request_pad() function to allow developers to be
able to specify which pad template they want a pad of.
Convert gstutils to use that new method instead of the old one when more
efficient.
This is useful for being able to request pads in a more flexible way,
especially when the element can provide pads whose caps depend on
runtime configuration and therefore can't provide pre-registered
pad templates.
API: GstElement::request_new_pad_full
API: gst_element_request_pad
https://bugzilla.gnome.org/show_bug.cgi?id=637300
We need to reset the revents field of each pollfd when reading the results
from select else we'll end up with stray info from previous calls to
select.
Fix freeing of partially-inited list value when both values
passed are equal and we want to return a single non-list
value as result. Fixes unit test. Also fix up docs a bit.
https://bugzilla.gnome.org/show_bug.cgi?id=637776
The -Bsymbolic change causes us to get a different address when internaly
looking up the function than what application would get when the use the symbol
that they see. This made removing the default loghandler to fail, as it is set
internally and removed externaly.
Make the _get_caps functions behave like the _get_caps_reffed variants and
remove the _reffed variants. This means that _get_caps doesn't return a writable
caps anymore and an explicit _make_writable() is needed before modifying the
caps.
Make GstObject extend from GInitiallyUnowned, remove the FLOATING flag and use
GObject methods for managing the floating ref.
Remove class lock, it was a workaround for a glib < 2.8 bug.
Remove the parent-set and parent-unset signals, attempt to implement with notify
but disabled because deadlocks in deep-notify.
All functions in this file can access the structure field of a query directly.
This avoids having to call gst_query_get_structure() to get it, along with being
able to remove some function variables that were used to store the result of that
function.
All functions in this file can access the structure field of an event directly.
This avoids having to call gst_query_get_structure() to get it, along with being
able to remove some function variables that were used to store the result of that
function.
There is no need to cache the peer chainfunction as we can just as efficiently
get to it from the peer object. Also not caching the chain function works better
because then we automatically get the new chainfunctions when they change.
Build the cache while we push data. When we don't have a cache, we run the
slowpath and collect cacheable properties. When all conditions are met, keep the
cached data around so that we can more efficiently push data around.
Fix returning of timezones on systems with gdatetime
to use floats on the math expression to avoid
truncating the fractional part.
Also adds a test for covering this case.
Otherwise the source will stay at NULL, the event is passed to the
peerpad via gst_pad_send_event() and then the peerpad is set as
source of the event instead of the originating pad.
This change always defines the restrict keyword if a
non-C99 C compiler is used. In the case of GCC >= 4
it will be defined to __restrict__, in all other
cases to nothing. This allows to use the restrict
keyword unconditionally.
Adds 2 variants for the gst_date_time_from_unix_epoch function,
one for UTC and another for local time.
API: gst_date_time_new_from_unix_epoch_utc
API: gst_date_time_new_from_unix_epoch_local_time
Fixes#653031https://bugzilla.gnome.org/show_bug.cgi?id=635031
Basically we're not meant to put anything more complex than simple numbers,
due to the definition of G_GUINT64_CONSTANT:
G_GUINT64_CONSTANT(val) (val##UL)
Which previously resulted in .... 1 << 49UL
gst_mini_object_unref() has guards that check the type already, so
we don't really need to re-check it here again while getting the
class (there's not really much point to that anyway, since we don't
check the return value of the get_class, so we'd crash anyway if
we're not dealing with a mini object, the only question would
be if there'd be a warning before the crash or not).
Track elements tagged with the IS_SOURCE flag in a similar way we track the sink
elements. This allows us to efficiently dispatch downstream events to the right
elements.
