Currently there is no other way to unlock a buffer pool other then
stopping it. This may have the effect of freeing all the buffers,
which is too heavy for a seek. This patch add a method to enter and
leave flushing state. As a convenience, flush_start/flush_stop
virtual are added so pool implementation can also unblock their own
internal poll atomically with the rest of the pool. This is fully
backward compatible with doing stop/start to actually flush the pool
(as being done in GstBaseSrc).
https://bugzilla.gnome.org/show_bug.cgi?id=727611
When we call gst_buffer_pool_set_config() the pool may return FALSE and
slightly change the parameters. This helper is useful to do the minial required
validation before accepting the modified configuration.
https://bugzilla.gnome.org/show_bug.cgi?id=727916
If a pool config is being configured again, check if the configuration have changed.
If not, skip that step. Finally, if the pool is active, try deactivating it.
https://bugzilla.gnome.org/show_bug.cgi?id=728268
Keep it simple. Likely also makes things easier for bindings,
and efficiency clearly has not been a consideration given how
the existing code handled these lists.
In order to be deterministic, multiple waiting GstClockIDs needs to be
released at the same time, or else one can get into the situation that
the one being released first can add itself back again before the next
one waiting is released.
Test added for new API and old tests rewritten to comply.
From the test case:
/* This test creates a multiqueue with 2 streams. One receives
* a constant flow of buffers, the other only gets one buffer, and then
* new-segment events, and returns not-linked. The multiqueue should not fill.
*/
If one of the queues goes EOS and the other returns NOT_LINKED the stream
can be considerered EOS as a NOT_LINKED means that one of the branches has no
sink downstream that will block the EOS message posting.
https://bugzilla.gnome.org/show_bug.cgi?id=725917
Tag allocated buffers with TAG_MEMORY. When they are released later,
only add them back to the pool if the tag is still there and the memory
has not been changed, otherwise throw the buffer away.
Add unit test to check various scenarios.
Fixes https://bugzilla.gnome.org/show_bug.cgi?id=724481
Store the eos event seqnum and use it when creating the
new eos event to be pushed downstream. To know if the eos
was caused by the eos events received on send_event, a
'forced_eos' flag is used to use the correct seqnum on
the event pushed downstream.
Useful if the application wants to check if the EOS message
was generated from its own pushed EOS or from another source
(stream really finished).
Also adds a test for this
https://bugzilla.gnome.org/show_bug.cgi?id=722791
Baseparse stores buffers for reverse playback to push on the next
DISCONT, the issue was that it wouldn't ever check for a discont
on passthrough mode as it skips all real parsing. This test
was create to verify this issue and prevent it from happening again
https://bugzilla.gnome.org/show_bug.cgi?id=721941
Checking twice the lower bound is great (you never know, it might change
between the two calls by someone using emacs butterfly-mode), but it's a bit
more useful to check the higher bound are also identical.
Detected by Coverity
gst_parse_launchv, gst_parse_launchv_full and gst_parse_launch_full
all return floating refs, the same as gst_parse_launch, which just
calls gst_parse_launch_full internally anyway.
Add a unit test assertion to check it's true.
Spotted by nemequ on IRC.
The check itself is racy.
(CK_FORK=no GST_CHECK=test_output_order make elements/multiqueue.forever).
The problem is indeed the test and not the actual element behaviour.
The objects to push are being pulled out of the single internal queues in the
right order and at the right time...
But between:
* the moment the global multiqueue lock is released (which was used to detect
if we should pop and push downstream the next buffer)
* and the moment it is received by the source pad (which does the check)
=> another single queue (like the unlinked pad) might pop and push a buffer
downstream
What should we do ? Putting a bigger margin of error (say 5 buffers) doesn't
help, it'll eventually fail.
I can't see how we can detect this reliably.
https://bugzilla.gnome.org/show_bug.cgi?id=708661
Wrap caps strings so that it can handle serialization and deserialization
of caps inside caps. Otherwise the values from the internal caps are parsed
as if they were from the upper one
https://bugzilla.gnome.org/show_bug.cgi?id=708772
Bash 3's completion doesn't split words by characters in
COMP_WORDBREAKS. In particular it doesn't split at "=" signs. Now
_gst_launch_parse handles both bash 3 and 4 format of COMP_WORDS.
Note that "${cur%%=*}" means cur's value with the longest possible match
of "=*" deleted from the end; "${cur#*=}" means cur's value with the
shortest possible match of "*=" deleted from the beginning. See
http://www.gnu.org/software/bash/manual/html_node/Shell-Parameter-Expansion.html
Regardless of the version of bash running the unit tests, I can test for
both behaviours because the unit test populates COMP_WORDS manually. So
this tests the bash 3 behaviour:
test_gst_inspect_completion --gst-debug-level=4
and this tests the bash 4 behaviour:
test_gst_inspect_completion --gst-debug-level = 4
Compatible with bash 3.2; doesn't require the bash-completion package at
all (though the easiest way to install this script is still to install
bash-completion, and then drop this script into /etc/bash_completion.d).
Note that bash 3 doesn't break COMP_WORDS according to characters in
COMP_WORDBREAKS, so "property=val" looks like a single word, so this
won't complete property values (on bash 3). Similarly,
"--gst-debug-level=<TAB>" won't complete properly (on bash 3), but
"--gst-debug-level <TAB>" will.
For that reason, I now offer "--gst-debug-level" etc as completions
instead of "--gst-debug-level=".
Functions "_init_completion" and "_parse_help" were provided by the
bash-completion package >= 2.0; now I roll my own equivalent of
"_parse_help", and instead of "_init_completion" I use
"_get_comp_words_by_ref" which is available from bash-completion 1.2
onwards. If the bash-completion package isn't available at all I use
bash's raw facilities, at the expense of not completing properly when
the cursor is in the middle of a word.
The builtin "compopt" doesn't exist in bash 3; those users will just
have to live with the inconvenience of "property=" completing to
"property= " with a trailing space. Property values aren't completed
properly anyway on bash 3 (see above).
"[[ -v var ]]" to test whether a variable is set, also doesn't exist in
bash 3. Neither does ";;&" to fall through in a "case" statement.
In the unit tests:
* On my system (OS X), "#!/bin/bash" is bash 3.2, whereas
"#!/usr/bin/env bash" is the 4.2 version I built myself.
* I have to initialise array variables like "expected=()", or bash 3
treats "+=" as appending to an array already populated with one empty
string.
Completes options like "--gst-debug-level" and the values of some of
those options; completes gst-launch pipeline element names, property
names, and even property values (for enum or boolean properties only).
Doesn't complete all caps specifications, nor element names specified
earlier in the pipeline with "name=...".
The GStreamer version number is hard-coded into the completion script:
This patch is off the master branch and has the version hard-coded as
"1.0"; it needs to be updated if backported to the 0.10 branch. You
could always create a "gstreamer-completion.in" that has the appropriate
version inserted by "configure", but I'd rather not do that. The
hard-coded version is consistent with the previous implementation of
gstreamer-completion, which had the registry path hard-coded as
~/.gstreamer-1.0/registry.xml.
Note that GStreamer 0.10 installs "gst-inspect" and "gst-inspect-0.10".
"gst-inspect --help" only prints 4 flags (--help, --print, --gst-mm,
gst-list-mm) whereas "gst-inspect-0.10 --help-all" prints the full list
of flags. The same applies to "gst-launch" and "gst-launch-0.10".
GStreamer 1.0 only installs "gst-inspect-1.0", not "gst-inspect".
Requires bash 4; only tested with bash 4.2. Requires "bash-completion"
(which you install with your system's package manager).
Put this in /etc/bash_completion.d/ or in `pkg-config
--variable=compatdir bash-completion`, where it will be loaded at the
beginning of every new terminal session;
or in `pgk-config --variable=completionsdir bash-completion`, renamed to
match the name of the command it completes (e.g. "gst-launch-1.0", with
an additional symlink named "gst-inspect-1.0"), where it will be
autoloaded when needed.
test-gstreamer-completion.sh is (for now) in tests/misc -- it might be
worth creating "tests/check/tools", with all the necessary automake
boilerplate, and moving test-gstreamer-completion.sh there, and have it
run automatically with "make check".
IF YOU'RE NEW TO BASH COMPLETION SCRIPTS
----------------------------------------
"complete -F _gst_launch gst-launch-1.0" means that bash will run the
function "_gst_launch" to generate possible completions for the command
"gst-launch-1.0".
"_gst_launch" must return the possible completions in the array variable
COMPREPLY. (Note on bash syntax: "V=(a b c)" assigns three elements to
the array "V").
"compgen" prints a list of possible completions to standard output. Try
it:
compgen -W "abc1 abc2 def" -- "a"
compgen -f -- "/"
The last argument is the word currently being completed; compgen uses it
to filter out the non-matching completions. We put "--" first, in case
the word currently being completed starts with "-" or "--", so that it
isn't treated as a flag to compgen.
For the documentation of COMP_WORDS, COMP_CWORD, etc see
http://www.gnu.org/software/bash/manual/html_node/Bash-Variables.html#index-COMP_005fCWORD-180
See also:
* http://www.gnu.org/software/bash/manual/html_node/Programmable-Completion.html
* http://www.gnu.org/software/bash/manual/html_node/Programmable-Completion-Builtins.html
The bash-completion package provides the helper function
"_init_completion" which populates variables "cur", "prev", and "words".
See
http://anonscm.debian.org/gitweb/?p=bash-completion/bash-completion.git;a=blob;f=bash_completion;h=870811b4;hb=HEAD#l634
Note that by default, bash appends a space to the completed word. When
the completion is "property=" we don't want a trailing space; calling
"compopt -o nospace" modifies the currently-executing completion
accordingly. See
http://www.gnu.org/software/bash/manual/html_node/Programmable-Completion-Builtins.html#index-compopt
Fixes abort when the old specifiers are used. Fix up the conversion
specifier, it would get overwritten with 'c' below to the extension
format char, which then later is unhandled, leading to the abort.
Also fix up and enable unit test for this.
https://bugzilla.gnome.org/process_bug.cgi
These account for both possible type size mismatch AND -mms-bitfields
packing. Sizes are taken from an i686-w64-mingw32-built GStreamer,
gcc 4.8.0, mingw-w64 svn-r5685.
Fixes#697551
This is equal to any other caps features but results in unfixed caps. It
would be used by elements that only look at the buffer metadata or are
currently working in passthrough mode, and as such don't care about any
specific features.
These are meant to specify features in caps that are required
for a specific structure, for example a specific memory type
or meta.
Semantically they could be though of as an extension of the media
type name of the structures and are handled exactly like that.
pop() in collected callback.
There were three threads in the test cases that hanged: the test thread and two
threads that push buffers. Each thread push one buffer on one pad. There are
two pads in the collectpads so the second buffer will trigger the
collect-callback.
This is what happens when the hang occurs:
The first thread pushes a buffer and initializes a cookie to the value of a
counter in the collectpads object and waits on a cond for the counter to change
and for someone to consume the buffer (i.e. _pop() it).
The second thread pushes a buffer and calls the collected callback, which
signals the cond that the test thread is waiting for.
The test thread pops both buffers (without holding any lock). Each call to
_pop() increases the counter broadcasts the condition that the first thread is
now waiting for. It then joins both threads (hangs).
The first thread wakes up and returns, since its buffer has been consumed.
The second thread starts executing again. When the callback, called by the
second thread, has returned it initializes a cookie to the value of a counter,
which has already prematurely been increased by the test thread when it popped
the buffers, and wait's on a cond for the counter to change and for someone to
consume the buffer (i.e. _pop() it). Since the buffer has already been poped
and the counter has already been increased it will be stuck forever.
https://bugzilla.gnome.org/show_bug.cgi?id=685555
We previously forgot to initilize the amplitde property to the default and thus it was 0.0. Therefore a default lfo controlsource returned a series of 0.0 and the test was asserting on that.
Set operations on the bitmasks don't make much sense and result
in invalid caps when used as a channel-mask. They are now handled
exactly like integers.
This functionality was not used anywhere except for tests.
https://bugzilla.gnome.org/show_bug.cgi?id=691370
The _1_0 suffixed environment variables override the
non-suffixed ones, so if we're in an environment that
sets the _1_0 suffixed ones, such as jhbuild, we need
to set those to make sure ours actually always get
used.
Implement the same behaviour as gst_pad_push_event when pushing sticky events
fails, that is don't fail immediately but fail when data flow resumes and upstream
can aggregate properly.
This fixes segment seeks with decodebin and unlinked audio or video branches.
Fixes: https://bugzilla.gnome.org/show_bug.cgi?id=687899
Fixes negotiation taking a ridiculous amount of
time (multiple 10s of seconds on a core2) when
there are duplicate entries in lists.
Could have a negative performance impact on other
scenarios because we now have to iterate the
dest list to avoid duplicates, but we don't
have a lot of lists any more these days, and
they tend to be small anyway. The negatives
are hopefully countered by the positive effects
of reducing the list length early on in the
process. And in any case, it's the right thing
to do.
Based on patch by Andre Moreira Magalhaes.
https://bugzilla.gnome.org/show_bug.cgi?id=684981
MIME-type -> media types
Fix up the manual in various places with the 1.0 way of doing things
such as probes, static elements, scheduling, ...
Add porting from 0.10 to 1.0 chapter.
Add probe example to build.
Remove some docs for remove components such as GstMixer and
GstPropertyProbe, XML...
Also add test to make sure that if a pad probe is removed while it's
callback is running, the cleanup_hook isn't called again if it
returns GST_PAD_PROBE_REMOVE
define GLIB_DISABLE_DEPRECATION_WARNINGS earlier so that it is defined before
the glib headers are loaded or else we trip over the GValueArray deprecations in
gst-inspect.c.
Our check would make sure that GLib segfaults when
someone tries to instantiate an abstract type, which
is an extremely useful thing to check for.
In newer GLibs this is fixed and we get an abort with
a g_error() now it seems, so let's just remove this
check entirely.
This is because we need to be able to signal different TOCs
to downstream elements such as muxers and the application,
and because we need to send both types as events (because
the sink should post the TOC messages for the app in the
end, just like tag messages are now posted by the sinks),
and hence need to make TOC events multi-sticky.
https://bugzilla.gnome.org/show_bug.cgi?id=678742
This specifies if a given taglist applies to the complete
medium or only this specific stream. By default a taglist
has a stream scope.
Fixes bug #677619.
Add an offset field that is used to track at what position the segment was
updated. This is used to set the running time to 0 when we do a flushing
seek that doesn't update the position.
See https://bugzilla.gnome.org/show_bug.cgi?id=680306
Remove the user_data from the alloc vmethod. Subclasses that implement a new
alloc function can also implement their own vmethod to pass extra arguments. We
can then also require that custom allocators implement an alloc function so that
gst_allocator_alloc() always works.
Stop querying the duration once an element return unknown and return unknown
as a final result. This avoid eventually cutting off a stream too early.
Add a tests to docuement the behavior.
Make GstAllocator a GstObject instead of a GstMiniObject, like bufferpool.
Make a new gstallocator.c file. Make a GstAllocator subclass for the default
allocator.
Clear the initial floating ref in the init function for
busses and clocks. These objects can be set on multiple
elements, so there's no clear parent-child relationship
here. Ideally we'd just not make them derive from
GInitiallyUnowned at all, but since we want to keep
using GstObject features for debugging, we'll just do
it like this.
This should also fix some problems with bindings, which
seem to get confused when they get floating refs from
non-constructor functions (or functions annotated to
have a 'transfer full' return type). This works now:
from gi.repository import GObject, Gst
GObject.threads_init()
Gst.init(None)
pipeline=Gst.Pipeline()
bus = pipeline.get_bus()
pipeline.set_state(Gst.State.NULL)
del pipeline;
https://bugzilla.gnome.org/show_bug.cgi?id=679286https://bugzilla.gnome.org/show_bug.cgi?id=657202
This re-uses existing code and makes sure we properly serialise
and deserialise datetimes where not all fields are set (thus
fixing some warnings when serialising such datetimes).
Expose the GstAllocation structure and provide an _init function. This makes it
easier to make 'subclasses' of the allocator that contain more info.
It also allows us to expose the flags on the allocator miniobject.
Make a flag to note that the allocator uses a custom alloc function.
Add a method that memory implementations can call to initialize the standard
GstMemory structure.
Move the parent handling in the _free handler.
Rearrange some internal function parameters so that the order is consistent.
Add more memory examples
Move the locking methods from GstMemory to GstMiniObject.
Add a miniobject flag to enable LOCKABLE objects. LOCKABLE objects can
use the lock/unlock API to control the access to the object.
Add a minobject flag that allows you to lock an object in readonly mode.
Modify the _is_writable() method to check the shared counter for LOCKABLE
objects. This allows us to control writability separately from the refcount for
LOCKABLE objects.
We added a minimum length of three letters originally so we would
fail to recognise DOS/Windows-style filenames as valid URIs (as we
should). Two should be just fine as well.
Add option to avoid build binaries. When building for platforms like
android, you might want to not link any "final" binary, mostly because
it requires special link flags or other parts of code that aren't
in the C library.
https://bugzilla.gnome.org/show_bug.cgi?id=677621
Now that TOCs are refcounted and have a GType, we can just
stuff a ref of the TOC directly into the various toc
event/message/query structures and get rid of lots of
cracktastic GstStructure <-> GstToc serialisation and
deserialisation code. We lose some TOC sanity checking
in the process, but that should really be done when
it's being created anyway.
Let's keep it simple for now:
gst_toc_setter_reset_toc() -> gst_toc_setter_reset()
gst_toc_setter_get_toc_copy() -> removed
gst_toc_setter_get_toc() -> returns a ref now
gst_toc_setter_get_toc_entry_copy() -> removed,
use TOC functions instead
gst_toc_setter_get_toc_entry() -> removed,
use TOC functions instead
gst_toc_setter_add_toc_entry() -> removed,
to avoid problems with (refcount-dependent)
writability of TOC; use TOC functions instead
