platform/upstream/gstreamer.git
9 years agodocs: Make sure gst_debug_bin_to_dot_data() is documented
Arun Raghavan [Thu, 23 Jul 2015 06:48:51 +0000 (12:18 +0530)]
docs: Make sure gst_debug_bin_to_dot_data() is documented

Thanks to Nirbheek Chauhan <nirbheek@centricular.com> for pointing this
out.

9 years agoharness: Fix race for gst_harness_element_ref
Stian Selnes [Wed, 5 Aug 2015 08:07:50 +0000 (10:07 +0200)]
harness: Fix race for gst_harness_element_ref

In order for gst_harness_new_full to be MT-safe the increase and
decrease of HARNESS_REF must be MT-safe. This allows for creating
multiple harnesses from different threads wrapping the same element.

https://bugzilla.gnome.org/show_bug.cgi?id=754661

9 years agoharness: Allow-none for custom stress init func
Stian Selnes [Wed, 5 Aug 2015 07:59:39 +0000 (09:59 +0200)]
harness: Allow-none for custom stress init func

It should be allowed to not have a function to initialize the user data
since it's often not necessary; it may already be initialized.

https://bugzilla.gnome.org/show_bug.cgi?id=754661

9 years agodocs: remove signal that no longer exists from docs
Tim-Philipp Müller [Sun, 6 Sep 2015 08:58:09 +0000 (09:58 +0100)]
docs: remove signal that no longer exists from docs

9 years agopo: update for translated string changes
Tim-Philipp Müller [Sat, 5 Sep 2015 10:20:49 +0000 (11:20 +0100)]
po: update for translated string changes

9 years agotools: gst-launch: fix --exclude command line option
Tim-Philipp Müller [Sat, 5 Sep 2015 10:18:27 +0000 (11:18 +0100)]
tools: gst-launch: fix --exclude command line option

This has not worked (as in: crashed) since 2005, so
perhaps it should just be removed instead.

9 years agoRevert "queue2: Process SEEKING query"
Tim-Philipp Müller [Mon, 31 Aug 2015 11:07:10 +0000 (12:07 +0100)]
Revert "queue2: Process SEEKING query"

This caused problems with oggdemux when queue2 was
operating in queue mode and the souphttpsrc upstream
is not seekable because the server doesn't support
range requests. It would then still claim seekability
and then things go wrong from there.

This reverts commit 7b0b93dafe4ac547552cdb66ade5d8aa0405e7b4.

https://bugzilla.gnome.org/show_bug.cgi?id=753887

9 years agoharness: misc bugfixes
Havard Graff [Sat, 29 Aug 2015 18:14:44 +0000 (20:14 +0200)]
harness: misc bugfixes

1. Get a list of pad templates from the element class, not the
factory. This allows us to interact with test-elements that does
not have a factory.

2. Use the pad_template_caps in caps-queries when caps is not set
explicitly on the pad. Not doing so is simply wrong, and prohibits
interactions with special templates used for testing.

https://bugzilla.gnome.org/show_bug.cgi?id=754193

9 years agotests: event: fix build failure
Thiago Santos [Wed, 26 Aug 2015 12:29:05 +0000 (09:29 -0300)]
tests: event: fix build failure

gst/gstevent.c:250:5: error: ‘for’ loop initial declarations are only
allowed in C99 or C11 mode

9 years agobin: Make sure we don't add/remove a bin to/from itself
Vivia Nikolaidou [Mon, 24 Aug 2015 18:04:37 +0000 (21:04 +0300)]
bin: Make sure we don't add/remove a bin to/from itself

Doing so would deadlock from trying to acquire the object lock twice

https://bugzilla.gnome.org/show_bug.cgi?id=754036

9 years agobasetransform: Reconfigure before propose_allocation
Nicolas Dufresne [Fri, 21 Aug 2015 21:28:48 +0000 (14:28 -0700)]
basetransform: Reconfigure before propose_allocation

There exist cases where a reconfigure event was propagated from
downstream, but caps didn't change. In this case, we would
reconfigure only when the next buffer arrives. The problem is that
due to the allocation query being cached, the return query parameters
endup outdated.

In this patch we refactor the reconfigurating code into a function, and
along with reconfiguring when a new buffer comes in, we also reconfigure
when a query allocation arrives.

https://bugzilla.gnome.org/show_bug.cgi?id=753850

9 years agobasesrc-test: Fix race testing segment update
Nicolas Dufresne [Fri, 7 Aug 2015 19:39:59 +0000 (15:39 -0400)]
basesrc-test: Fix race testing segment update

As this test is using a short sleep (GST_USECOND, which is 10ms
in microsecond), sometimes that EOS event is received before the
loop in basesrc have run _do_seek() and pushed the update segment.
To solve this issue, we wait for the initial segment (and flush it)
then we wait for the second segment before sending EOS.

https://bugzilla.gnome.org/show_bug.cgi?id=753365

9 years agobin: Add NLE to GST_PLUGIN_PATH
Thibault Saunier [Wed, 19 Aug 2015 09:46:07 +0000 (11:46 +0200)]
bin: Add NLE to GST_PLUGIN_PATH

9 years agoRelease 1.5.90
Sebastian Dröge [Wed, 19 Aug 2015 09:50:56 +0000 (12:50 +0300)]
Release 1.5.90

9 years agoUpdate .po files
Sebastian Dröge [Wed, 19 Aug 2015 09:33:41 +0000 (12:33 +0300)]
Update .po files

9 years agopo: Update translations
Sebastian Dröge [Wed, 19 Aug 2015 08:17:29 +0000 (11:17 +0300)]
po: Update translations

9 years agobaseparse: avoid tag list spam if upstream provides bitrate tags already
Tim-Philipp Müller [Tue, 18 Aug 2015 14:44:02 +0000 (15:44 +0100)]
baseparse: avoid tag list spam if upstream provides bitrate tags already

Explicitly keep track again whether upstream tags or parser tags
already contain bitrate information, and only force a tag update
for a bitrate if we are actually going to add the bitrate to the
taglist later. This fixes constant re-sending of the same taglist,
because upstream provided a bitrate already and we didn't add it,
so we didn't save the 'posted' bitrate, which would then in turn
again trigger the 'bitrate has changed too much, update tags'
code path. Fixes tag spam with m4a files for example.

https://bugzilla.gnome.org/show_bug.cgi?id=679768

9 years agodebugutils: bring the dot style a bit closer to what we use in the docs
Stefan Sauer [Mon, 17 Aug 2015 20:06:11 +0000 (22:06 +0200)]
debugutils: bring the dot style a bit closer to what we use in the docs

Use round corners for bins and elements. Put sink pads on the left and src pads
on the right of elements.

9 years agobaseparse: fix tag handling
Tim-Philipp Müller [Sat, 15 Aug 2015 17:30:15 +0000 (18:30 +0100)]
baseparse: fix tag handling

In 0.10 there were no sticky events, and all tag events
sent would just be merged with the previously-received
tags. In 1.x we have sticky events, and the tags in the
tag event(s) should at all times carry the complete tags,
so we can't just push some tags and then just push tags
with just bitrates to update the bitrates, etc.

Instead we need to keep track of the upstream stream tags
received, of the tags set by the video decoder subclass,
and send an updated tag event with the combined tags
including our own bitrate tags (if applicable) whenever
the upstream tags, the subclass tags or any of our bitrates
change.

https://bugzilla.gnome.org/show_bug.cgi?id=679768

9 years agobaseparse: add API for subclass to set tags
Tim-Philipp Müller [Sun, 16 Aug 2015 09:15:56 +0000 (10:15 +0100)]
baseparse: add API for subclass to set tags

This is needed so that we can do proper tag handling
all around, and combine the upstream tags with the
tags set by the subclass and any extra tags the
base class may want to add.

API: gst_base_parse_merge_tags()

https://bugzilla.gnome.org/show_bug.cgi?id=679768

9 years agobaseparse: save upstream stream tags
Tim-Philipp Müller [Sat, 15 Aug 2015 15:01:28 +0000 (16:01 +0100)]
baseparse: save upstream stream tags

We'll need those later.

https://bugzilla.gnome.org/show_bug.cgi?id=679768

9 years agobaseparse: minor code simplification
Tim-Philipp Müller [Sat, 15 Aug 2015 15:39:40 +0000 (16:39 +0100)]
baseparse: minor code simplification

Use gst_pad_peer_query_duration() and remove a few
unnecessary levels of indentation. Rest of code might
looks a bit questionable, but leave it as is for now.

9 years agopad: Break sticky event array iterations if the type is bigger than the one we look for
Sebastian Dröge [Sat, 15 Aug 2015 15:59:21 +0000 (17:59 +0200)]
pad: Break sticky event array iterations if the type is bigger than the one we look for

Microoptimization we can do because the array is sorted by type.

9 years agogstpad: Add a new GST_PROBE_HANDLED return value for probes
Edward Hervey [Wed, 29 Apr 2015 13:49:17 +0000 (15:49 +0200)]
gstpad: Add a new GST_PROBE_HANDLED return value for probes

In some cases, probes might want to handle the buffer/event/query
themselves and stop the data from travelling further downstream.

While this was somewhat possible with buffer/events and using
GST_PROBE_DROP, it was not applicable to queries, and would result
in the query failing.

With this new GST_PROBE_HANDLED value, the buffer/event/query will
be considered as successfully handled, will not be pushed further
and the appropriate return value (TRUE or GST_FLOW_OK) will be returned

This also allows probes to return a non-default GstFlowReturn when dealing
with buffer push. This can be done by setting the
GST_PAD_PROBE_INFO_FLOW_RETURN() field accordingly

https://bugzilla.gnome.org/show_bug.cgi?id=748643

9 years agogstversion: Add missing include in .in file.
Mathieu Duponchelle [Sat, 15 Aug 2015 11:25:35 +0000 (13:25 +0200)]
gstversion: Add missing include in .in file.

9 years agoHeaders: add missing includes.
Mathieu Duponchelle [Mon, 10 Aug 2015 22:35:21 +0000 (00:35 +0200)]
Headers: add missing includes.

9 years agodocs: add the new pad accept-template flag to the docs
Thiago Santos [Sat, 15 Aug 2015 09:41:14 +0000 (06:41 -0300)]
docs: add the new pad accept-template flag to the docs

9 years agodocs: section entry missing for gst_direct_control_binding_new_absolute
Nirbheek Chauhan [Fri, 14 Aug 2015 17:14:50 +0000 (22:44 +0530)]
docs: section entry missing for gst_direct_control_binding_new_absolute

9 years agotests: pad: tests for accept-caps default handling
Thiago Santos [Fri, 14 Aug 2015 11:14:52 +0000 (08:14 -0300)]
tests: pad: tests for accept-caps default handling

Check if all the default 4 accept-caps possibilities are working:
subset or intersect check and query-caps or template caps comparisons.

https://bugzilla.gnome.org/show_bug.cgi?id=753623

9 years agopad: add GST_PAD_FLAG_ACCEPT_TEMPLATE
Thiago Santos [Fri, 14 Aug 2015 10:51:07 +0000 (07:51 -0300)]
pad: add GST_PAD_FLAG_ACCEPT_TEMPLATE

It will make the default accept-caps handler use the pad template
caps instead of the query-caps result to check if the caps is
acceptable. This is aligned with what the design docs says the
accept-caps should do (be non-recursive) and should be faster. It
is *not* enabled by default, though.

API: GST_PAD_FLAG_ACCEPT_TEMPLATE
API: GST_PAD_IS_ACCEPT_TEMPLATE
API: GST_PAD_SET_ACCEPT_TEMPLATE
API: GST_PAD_UNSET_ACCEPT_TEMPLATE

https://bugzilla.gnome.org/show_bug.cgi?id=753623

9 years agocheck: Rename states unit test
Edward Hervey [Fri, 14 Aug 2015 09:10:03 +0000 (11:10 +0200)]
check: Rename states unit test

Makes it easier to differentiate from other modules states unit test

9 years agobasetransform: rework accept-caps
Thiago Santos [Thu, 13 Aug 2015 16:08:03 +0000 (13:08 -0300)]
basetransform: rework accept-caps

According to the design docs:
The ACCEPT_CAPS query is not required to work recursively, it can simply
return TRUE if a subsequent CAPS event with those caps would return
success.

So make it a shallow check instead of recursivelly check downstream.

https://bugzilla.gnome.org/show_bug.cgi?id=748635

9 years agobasetransform: remove some dead code
Thiago Santos [Thu, 13 Aug 2015 15:44:29 +0000 (12:44 -0300)]
basetransform: remove some dead code

Doesn't seem like it is going to get back to life anytime soon

Also removes a {} block that was likely used to keep the dead
code around.

9 years agobasetransform: respect accept-caps intersect flag
Thiago Santos [Tue, 11 Aug 2015 11:07:53 +0000 (08:07 -0300)]
basetransform: respect accept-caps intersect flag

GstPad has a flag for suggesting if the accept-caps
query should use intersect instead of the default
subset caps operation to verify if the caps would be
acceptable.

basetransform currently always uses the subset check and
this patch makes it honor the flag for using intersect
if it is set.

https://bugzilla.gnome.org/show_bug.cgi?id=748635

9 years agobasetransform: remove unreachable return statement
Vineeth TM [Wed, 12 Aug 2015 04:12:38 +0000 (13:12 +0900)]
basetransform: remove unreachable return statement

https://bugzilla.gnome.org/show_bug.cgi?id=753538

9 years agotests: ignore new harness test binary
Tim-Philipp Müller [Tue, 11 Aug 2015 10:09:24 +0000 (11:09 +0100)]
tests: ignore new harness test binary

9 years agodatetime: accept just a time as ISO 8601 string and use today's date then
Vivia Nikolaidou [Mon, 10 Aug 2015 12:31:37 +0000 (15:31 +0300)]
datetime: accept just a time as ISO 8601 string and use today's date then

If no date and only a time is given in gst_date_time_new_from_iso8601_string(),
assume that it is "today" and try to parse the time-only string. "Today" is
assumed to be in the timezone provided by the user (if any), otherwise Z -
just like the behavior of the existing code.

https://bugzilla.gnome.org/show_bug.cgi?id=753455

9 years agoharness: add _set_forwarding function
Havard Graff [Thu, 23 Jul 2015 22:41:57 +0000 (00:41 +0200)]
harness: add _set_forwarding function

To be able to disable the slightly "magic" forwarding of the
necessary events between the harnesses.

Also introduce a new test-suite for GstHarness, that documents the
feature, and should hopefully expand into documenting most of the
features the harness possesses.

https://bugzilla.gnome.org/show_bug.cgi?id=752746

9 years agodevicemonitor: fix provider leak
Wim Taymans [Sat, 8 Aug 2015 15:59:51 +0000 (17:59 +0200)]
devicemonitor: fix provider leak

9 years agopad: Fix previous commit
Edward Hervey [Sat, 8 Aug 2015 13:28:19 +0000 (15:28 +0200)]
pad: Fix previous commit

We want to get the caps query *result*

9 years agodeviceprovider: Add method to hide devices from a provider
Wim Taymans [Thu, 16 Jul 2015 16:56:00 +0000 (18:56 +0200)]
deviceprovider: Add method to hide devices from a provider

Add methods to add/remove the providers that should be hidden by this
provider. Also make a method to get a list of hidden providers.

This makes it possible to have multiple systems monitor the same devices
and remove duplicates.

Add a property to see all devices, even duplicate ones from hidden
providers.

9 years agopad: get_allowed_caps() should go through both pads
Edward Hervey [Sat, 8 Aug 2015 12:42:52 +0000 (14:42 +0200)]
pad: get_allowed_caps() should go through both pads

The previous implementation was doing a direct call to the peer pad,
which resulted in query probes never being called on the original pad.

Instead of that, get the peer pad caps by using gst_pad_peer_query()
which will call probes in the expected fashion.

9 years agovalue: free caps during failure
Vineeth TM [Fri, 7 Aug 2015 01:08:21 +0000 (10:08 +0900)]
value: free caps during failure

While calling gst_value_deserialize_sample, if there is a failure
after caps is ref'ed, then caps is getting leaked. Hence checking for
caps in fail: goto condition and unref'ing it

https://bugzilla.gnome.org/show_bug.cgi?id=753338

9 years agoregistry: Add plugins to the registry we are loading and not default one
Thibault Saunier [Tue, 21 Jul 2015 11:35:33 +0000 (13:35 +0200)]
registry: Add plugins to the registry we are loading and not default one

When running gst_registry_scan_plugin_file we were losing the
information about the registry being loaded and ended up adding the
plugin to the default registry which was not correct.

https://bugzilla.gnome.org/show_bug.cgi?id=752662

9 years agobasesink: Only drop buffer if their PTS is out of segment
Nicolas Dufresne [Wed, 5 Aug 2015 19:51:27 +0000 (15:51 -0400)]
basesink: Only drop buffer if their PTS is out of segment

As of now, even for stream completly inside segment, there is no
guarantied that the DTS will be inside the segment. Specifically
for H.264 with B-Frames, the first few frames often have DTS that
are before the segment.

Instead of using the sync timestamp to clip out of segment buffer,
take the duration from the start/stop provided by the sub-class, and
check if the pts and pts_end is out of segment.

https://bugzilla.gnome.org/show_bug.cgi?id=752791

9 years agoharness: don't run code inside g_assert
Luis de Bethencourt [Wed, 5 Aug 2015 13:05:25 +0000 (14:05 +0100)]
harness: don't run code inside g_assert

Even though asserts can't be disabled in GstHarness, Coverity still
complains about running code inside them. Moving the code to outside the
g_asserts().

CID #1311326, #1311327, #1311328

9 years agodevicemonitor: get a list of currently monitored providers
Wim Taymans [Fri, 17 Jul 2015 08:18:02 +0000 (10:18 +0200)]
devicemonitor: get a list of currently monitored providers

Get a list of the currently monitored providers.

9 years agopad: fix invalid unref after IDLE probe on non-OK flow return
George Kiagiadakis [Sun, 2 Aug 2015 15:38:14 +0000 (17:38 +0200)]
pad: fix invalid unref after IDLE probe on non-OK flow return

In case there is an IDLE probe fired from gst_pad_push_data and it
doesn't return GST_FLOW_OK, the code jumps to the probe_stopped
label which tries to unref the data object. However, at this point
the data object belongs downstream and must not be touched.
By setting data = NULL, the code skips this unref.

https://bugzilla.gnome.org//show_bug.cgi?id=753151

9 years agobuffer: Fix the name of the parentbuffermeta debug category.
Jan Schmidt [Tue, 4 Aug 2015 10:08:04 +0000 (20:08 +1000)]
buffer: Fix the name of the parentbuffermeta debug category.

Don't use 'glbufferrefmeta' as the debug category for the
parent buffer meta.

9 years agoqueue2: not update upstream size with negative value
Eunhae Choi [Tue, 4 Aug 2015 04:45:09 +0000 (13:45 +0900)]
queue2: not update upstream size with negative value

upstream_size can be negative but queue->upstream_size is unsigned type.
to get a chance to update queue->upstream_size in gst_queue2_get_range()
it should keep the default value.

https://bugzilla.gnome.org/show_bug.cgi?id=753011

9 years agobuffer: Remove extra debug symbol from exports
Jan Schmidt [Tue, 4 Aug 2015 09:59:28 +0000 (19:59 +1000)]
buffer: Remove extra debug symbol from exports

Don't export the debug variable for the parent_buffer_meta.

This was accidentally exported and shouldn't be public

9 years agofilesink: use GST_INFO_OBJECT for more detail
Stefan Sauer [Mon, 3 Aug 2015 22:11:24 +0000 (00:11 +0200)]
filesink: use GST_INFO_OBJECT for more detail

Helps to distiguish multiple filesinks.

9 years agodocs: info: remove 0.8 terminology from log level description
Tim-Philipp Müller [Thu, 30 Jul 2015 16:29:25 +0000 (17:29 +0100)]
docs: info: remove 0.8 terminology from log level description

We don't "iterate" bins or pipelines any more.

9 years agotests: baseparse: fix buffer leak in unit test
Tim-Philipp Müller [Thu, 30 Jul 2015 11:17:16 +0000 (12:17 +0100)]
tests: baseparse: fix buffer leak in unit test

Fixes make check-valgrind

9 years agodoc/seekflags: Fix cross references
Nicolas Dufresne [Wed, 29 Jul 2015 01:14:22 +0000 (21:14 -0400)]
doc/seekflags: Fix cross references

This fixes miss-use of @ instead of % to refer to enumeration
values.

9 years agodocs: add a few more new symbols and defines
Tim-Philipp Müller [Tue, 28 Jul 2015 21:30:54 +0000 (22:30 +0100)]
docs: add a few more new symbols and defines

9 years agodoc/capsfilter: Document filtering modes
Nicolas Dufresne [Tue, 28 Jul 2015 20:57:43 +0000 (16:57 -0400)]
doc/capsfilter: Document filtering modes

This is documentation for the HTML documentation.

9 years agodoc/filesink: Add BufferMode enumeration
Nicolas Dufresne [Tue, 28 Jul 2015 20:50:55 +0000 (16:50 -0400)]
doc/filesink: Add BufferMode enumeration

This is purely for documentation purpose. This way the values will
show up in the HTML documentation.

9 years agodoc/gsthardness: Fix typo in GstAllocationParams
Nicolas Dufresne [Tue, 28 Jul 2015 19:50:40 +0000 (15:50 -0400)]
doc/gsthardness: Fix typo in GstAllocationParams

It's not GstAllocatorParams but GstAllocationParams.

9 years agodoc/gstharness: Remove unknown parameter
Nicolas Dufresne [Tue, 28 Jul 2015 19:46:09 +0000 (15:46 -0400)]
doc/gstharness: Remove unknown parameter

sink_elenment_name is not a parameter of gst_harness_add_sink_harness()
function, but still it show up in documentation.

9 years agocapsfilter: Only remember previous filter caps if they were actually used for something
Sebastian Dröge [Tue, 28 Jul 2015 09:19:04 +0000 (12:19 +0300)]
capsfilter: Only remember previous filter caps if they were actually used for something

If nobody ever saw the previous filter caps, nothing could've negotiated with
them and we can just pretend they never existed at all.

9 years agocapsfilter: When switching caps change modes, forget all previous caps
Sebastian Dröge [Tue, 28 Jul 2015 09:16:12 +0000 (12:16 +0300)]
capsfilter: When switching caps change modes, forget all previous caps

9 years agobasetransform: Return FLOW_FLUSHING if negotiation fails during shutdown
Olivier Crête [Thu, 23 Jul 2015 22:15:05 +0000 (18:15 -0400)]
basetransform: Return FLOW_FLUSHING if negotiation fails during shutdown

https://bugzilla.gnome.org/show_bug.cgi?id=752800

9 years agoharness: Fix indendation
Olivier Crête [Wed, 22 Jul 2015 22:55:29 +0000 (18:55 -0400)]
harness: Fix indendation

9 years agobasetransform: Avoid increasing query reference
Nicolas Dufresne [Tue, 21 Jul 2015 17:14:59 +0000 (13:14 -0400)]
basetransform: Avoid increasing query reference

gst_query_find_allocation_meta() requires the query to be
writable to work. This patch ensure avoids taking a reference
on the query, so we can now check if a certain allocation meta
is present.

https://bugzilla.gnome.org/show_bug.cgi?id=752661

9 years agodocs: fix description of gst_buffer_extract_dup()
Tim-Philipp Müller [Wed, 22 Jul 2015 14:38:06 +0000 (15:38 +0100)]
docs: fix description of gst_buffer_extract_dup()

No GBytes involved.

9 years agoconcat: dot not reset pad states too early
Thiago Santos [Tue, 21 Jul 2015 03:17:28 +0000 (00:17 -0300)]
concat: dot not reset pad states too early

Resetting the flushing state of the pads at the end of the
PAUSED_TO_READY transition will make pads handle serialized
queries again which will wait for non-active pads and might
cause deadlocks when stopping the pipeline.

Move the reset to the READY_TO_PAUSED instead.

https://bugzilla.gnome.org/show_bug.cgi?id=752623

9 years agoharness: add functions for adding sub-harnesses directly
Havard Graff [Mon, 20 Jul 2015 14:18:06 +0000 (16:18 +0200)]
harness: add functions for adding sub-harnesses directly

By introducing gst_harness_add_src_harness and gst_harness_add_sink_harness
we collect all sub-harness setup in one function, making the previous
sub-harness creation functions now calls these directly, and making it
much easier (and less error-prone) to add your own src or sink-harness
using the more generic harness-creation functions.

9 years agobaseparse: Don't override gst_segment_do_seek()
Nicolas Dufresne [Fri, 17 Jul 2015 21:44:52 +0000 (17:44 -0400)]
baseparse: Don't override gst_segment_do_seek()

This line has no purpose, clearly gst_segment_do_seek() is doing
the right job, also, having the start time (a timestamp) be that
same as time (the stream time) is quite odd.

https://bugzilla.gnome.org/show_bug.cgi?id=750783

9 years agobaseparse: Fix extrapolation of seeksegment.stop
Nicolas Dufresne [Fri, 17 Jul 2015 21:43:26 +0000 (17:43 -0400)]
baseparse: Fix extrapolation of seeksegment.stop

The stop shall be relative to start if extrapolated from the
duration.

https://bugzilla.gnome.org/show_bug.cgi?id=750783

9 years agodevicemonitor: do start and stop outside of the lock
Wim Taymans [Thu, 16 Jul 2015 16:47:20 +0000 (18:47 +0200)]
devicemonitor: do start and stop outside of the lock

Release the monitor lock when calling the provider start/stop methods.
Because we release the lock now, We need to make sure we check the
cookie again and keep track of started and removed providers.

9 years agodeviceprovider: small cleanups
Wim Taymans [Thu, 16 Jul 2015 16:43:06 +0000 (18:43 +0200)]
deviceprovider: small cleanups

Protect against wrong arguments.
Clean up the header file indentation.

9 years agodevicemonitor: keep order of providers and devices
Wim Taymans [Thu, 16 Jul 2015 15:25:24 +0000 (17:25 +0200)]
devicemonitor: keep order of providers and devices

The deviceproviders are added to the array sorted by their rank. Make
sure we keep this ordering when removing a provider.

We use _prepend to collect the devices, use g_list_reverse to get the
devices in the right order; sorted by rank and in the same order as
returned by the provider.

9 years agoharness: fix indentation
Tim-Philipp Müller [Thu, 16 Jul 2015 16:50:49 +0000 (17:50 +0100)]
harness: fix indentation

9 years agoharness: fix pad template leak
Tim-Philipp Müller [Thu, 16 Jul 2015 16:50:06 +0000 (17:50 +0100)]
harness: fix pad template leak

9 years agodocs: drop reference to sourceforge mailing list adress
Tim-Philipp Müller [Thu, 16 Jul 2015 16:13:35 +0000 (17:13 +0100)]
docs: drop reference to sourceforge mailing list adress

9 years agoharness: don't re-establish the harness sink and src pads
Havard Graff [Thu, 16 Jul 2015 15:53:40 +0000 (17:53 +0200)]
harness: don't re-establish the harness sink and src pads

Given that the element has the possibility to have one, they should
already be there.

https://bugzilla.gnome.org/show_bug.cgi?id=752498

9 years agoharness: Improve detection of element type
Stian Selnes [Mon, 13 Jul 2015 09:03:13 +0000 (11:03 +0200)]
harness: Improve detection of element type

The element flag does not indicate wether a bin should be tested as a
source or as a sink, eg. a bin with the sink flag may still have a
source pad and a bin with the source flag may have a sink pad. In this
case it is better to determine the element type by looking at the
available pads and pad templates.

Also rename srcpad and sinkpad where it actually represents
element_srcpad_name and element_sinkpad_name.

https://bugzilla.gnome.org/show_bug.cgi?id=752493

9 years agoharness: Forward sticky events to sink harness
Stian Selnes [Mon, 13 Jul 2015 09:10:49 +0000 (11:10 +0200)]
harness: Forward sticky events to sink harness

Fixes issue where if a sink harness was added late the sticky events
would not be forwared.

https://bugzilla.gnome.org/show_bug.cgi?id=752494

9 years agoharness: make header nicer to read
Tim-Philipp Müller [Thu, 16 Jul 2015 11:36:14 +0000 (12:36 +0100)]
harness: make header nicer to read

9 years agodocs: add new function to API docs
Tim-Philipp Müller [Thu, 16 Jul 2015 09:36:36 +0000 (10:36 +0100)]
docs: add new function to API docs

9 years agodevice: add generic struct with properties
Wim Taymans [Wed, 15 Jul 2015 16:21:13 +0000 (18:21 +0200)]
device: add generic struct with properties

Add a generic structure to hold any additional properties about the
device.

9 years agotee: fix typo in allow-not-linked property description
Tim-Philipp Müller [Tue, 14 Jul 2015 11:44:12 +0000 (12:44 +0100)]
tee: fix typo in allow-not-linked property description

9 years agodocs: bus: mention main loop requirement in gst_bus_add_watch() docs
Tim-Philipp Müller [Mon, 13 Jul 2015 13:24:34 +0000 (14:24 +0100)]
docs: bus: mention main loop requirement in gst_bus_add_watch() docs

9 years agotask: add function guard for _set_lock() and fix guard for _join()
Prashant Gotarne [Wed, 18 Mar 2015 10:35:34 +0000 (16:05 +0530)]
task: add function guard for _set_lock() and fix guard for _join()

Should only access the object structure after checking
it's valid in gst_task_join().

https://bugzilla.gnome.org/show_bug.cgi?id=746385
https://bugzilla.gnome.org/show_bug.cgi?id=746431

9 years agoprotection: implement meta transform function
Philippe Normand [Tue, 19 May 2015 16:58:11 +0000 (18:58 +0200)]
protection: implement meta transform function

Copy the GstMeta contents over to the new buffer.

https://bugzilla.gnome.org/show_bug.cgi?id=749590

9 years agobaseparse: estimate duration on EOS
Vineeth TM [Fri, 10 Jul 2015 00:12:15 +0000 (09:12 +0900)]
baseparse: estimate duration on EOS

For files which are smaller than 1.5 seconds, the duration
estimation does not happen. So the duration will always be
displayed as 0. Updating the duration on EOS when the estimation
has not happened already

https://bugzilla.gnome.org/show_bug.cgi?id=750131

9 years agoadapter: change log message properly
Hyunjun Ko [Fri, 10 Jul 2015 02:01:21 +0000 (11:01 +0900)]
adapter: change log message properly

https://bugzilla.gnome.org/show_bug.cgi?id=752116

9 years agodocs: add StreamidDemux to documentation
Justin Joy [Wed, 8 Jul 2015 15:12:51 +0000 (00:12 +0900)]
docs: add StreamidDemux to documentation

https://bugzilla.gnome.org/show_bug.cgi?id=749873

9 years agoadapter: fix to get valid (buffer_)list
Hyunjun Ko [Wed, 8 Jul 2015 15:21:42 +0000 (00:21 +0900)]
adapter: fix to get valid (buffer_)list

get_list/get_buffer_list should be done with buffers in adapter remaining
while take_list/take_buffer_list flushes each buffer one by one.

https://bugzilla.gnome.org/show_bug.cgi?id=752116

9 years agoadapter: unit test for new get_(buffer_)list
Hyunjun Ko [Wed, 8 Jul 2015 11:06:27 +0000 (20:06 +0900)]
adapter: unit test for new get_(buffer_)list

9 years agobaseparse: put buffer in a correct state after gst_adapter_get_buffer call
Arnaud Vrac [Wed, 8 Jul 2015 10:00:56 +0000 (12:00 +0200)]
baseparse: put buffer in a correct state after gst_adapter_get_buffer call

We must make the buffer writable to write its PTS and DTS, and also
reset its duration.

The behaviour is now the same as before commit c3bcbadd, except metas
might still be attached to the buffer extracted from the adapter.

https://bugzilla.gnome.org/show_bug.cgi?id=752092

9 years agoharness: fix indentation and replace stress test function macros
Tim-Philipp Müller [Tue, 7 Jul 2015 14:02:45 +0000 (15:02 +0100)]
harness: fix indentation and replace stress test function macros

These screw with indentation and seem a bit trivial. Just copy'n'paste.

9 years agometa: transform_func: return FALSE if not supported or failed
Hyunjun Ko [Tue, 7 Jul 2015 01:46:48 +0000 (10:46 +0900)]
meta: transform_func: return FALSE if not supported or failed

https://bugzilla.gnome.org/show_bug.cgi?id=751778

9 years agoidentity: refactor and add tests using GstHarness
Havard Graff [Tue, 7 Jul 2015 09:53:07 +0000 (11:53 +0200)]
identity: refactor and add tests using GstHarness

Writing a test for unscheduling the gst_clock_id_wait inside the
identity element, found an invalid read, caused by removing the clock-id
when calling _unschedule instead of letting the code calling _wait remove
the clock-id after being unscheduled.

https://bugzilla.gnome.org/show_bug.cgi?id=752055

9 years agoharness: make sure g_assert() statements are always active
Tim-Philipp Müller [Sat, 12 Apr 2014 18:48:15 +0000 (19:48 +0100)]
harness: make sure g_assert() statements are always active

We have code with side effects inside g_assert()s, so make
sure those are always enabled here (they might otherwise
get disabled for release builds).

9 years agoharness: rename GstHarnessPrepareBuffer -> GstHarnessPrepareBufferFunc
Tim-Philipp Müller [Mon, 6 Jul 2015 23:56:41 +0000 (00:56 +0100)]
harness: rename GstHarnessPrepareBuffer -> GstHarnessPrepareBufferFunc

https://bugzilla.gnome.org/show_bug.cgi?id=751916

9 years agodocs: add GstHarness to documentation
Tim-Philipp Müller [Mon, 6 Jul 2015 23:53:48 +0000 (00:53 +0100)]
docs: add GstHarness to documentation

https://bugzilla.gnome.org/show_bug.cgi?id=751916

9 years agocheck: Add GstHarness convenience API for unit tests
Havard Graff [Mon, 16 Dec 2013 09:47:47 +0000 (10:47 +0100)]
check: Add GstHarness convenience API for unit tests

http://gstconf.ubicast.tv/videos/gstharness-again-a-follow-up/

https://bugzilla.gnome.org/show_bug.cgi?id=751916

9 years agobaseparse: reverse playback in pull mode
Vineeth TM [Mon, 6 Jul 2015 00:26:58 +0000 (09:26 +0900)]
baseparse: reverse playback in pull mode

right now reverse playback is disabled in pull mode.
enabling the code for the same and changing a bit of logic
to make reverse playback work.

https://bugzilla.gnome.org/show_bug.cgi?id=750783